博客列表¶

August 23, 2019
in DonkeyCar
5 min read

Donkey Car Detailed Explanation (2) – Raspberry Pi Installation

robocarstore/173807615525023113

Now the official Donkey Car is no longer provided with a lazy person's version, so this time we will share the process of installing Donkey Car on a Raspberry Pi (Raspberry Pi 3 B+).

Reference: http://docs.donkeycar.com/guide/robot_sbc/setup_raspberry_pi/

Environment

The official recommendation is to use Raspian Lite(Stretch) (352MB). However, I used the desktop version, as follows:

Raspberry Pi OS:
– Raspbian Buster with desktop
Version:July 2019
Release date:2019-07-10
Kernel version:4.19

Download Url:

https://www.raspberrypi.org/downloads/raspbian/

How to flash install Raspberry Pi, here is no more.

Assume you have already installed the Raspberry Pi system and can connect to the network, you can choose to connect to the display directly (this example's operation mode), or remotely SSH. You like it.

Open the command line, let's start!

donkeycar-rpi

Start Installation

1，Update your system before installation

To update your system, enter the following command:

Updating the system takes a long time and may need to be repeated multiple times to complete. Please be patient.

2，Configure Raspberry Pi to enable camera, I2C, expand file system

To open the Raspberry Pi configuration dialog, enter the following command in the command line:

① Enable I2C

05 Interfacing Options -> P5 I2C ->「Yes」

rpi_i2c

② Enable Camera:

05 Interfacing Options -> P1 Camera -> 「Yes」

robocarstore/173807630025267416

③ Expand File System

7 Advanced Options -> A1 Exapand Filesystem

robocarstore/173807631425048117

Set it up, then exit the configuration dialog, it will prompt whether to restart now, select 「Yes」 to restart the system.

robocarstore/173807632825076218

3，Install the dependencies required for Donkey Car

Enter the following command in the command line:

sudo apt-get install -y build-essential python3 python3-dev python3-pip python3-virtualenv python3-numpy python3-picamera python3-pandas python3-rpi.gpio i2c-tools avahi-utils joystick libopenjp2-7-dev libtiff5-dev gfortran libatlas-base-dev libopenblas-dev libhdf5-serial-dev git

After a long wait, it will be installed, if an error occurs, please run it again.

（If the problem is still not resolved, please share the problem you encountered in the comments section, and post your system version, as much information as possible, so we can help you.）

4，Install OpenCV dependencies (optional)

This step is optional, you can skip it if you don't use the OpenCV feature.

If you still want to install it, enter the following command in the command line:

sudo apt-get install libilmbase-dev libopenexr-dev libgstreamer1.0-dev libjasper-dev libwebp-dev libatlas-base-dev libavcodec-dev libavformat-dev libswscale-dev libqtgui4 libqt4-test

5，Install Virtual Environment

Enter the following command in the command line:

python3 -m virtualenv -p python3 env --system-site-packages

echo "source env/bin/activate" >> ~/.bashrc

After running the above commands, the command line will have a (env) prefix, indicating that we are operating in an environment called env.

robocarstore/173807634625230519

The purpose of the virtual environment is to avoid errors in other projects when updating a dependency or automatically clearing a dependency on the Raspberry Pi running multiple projects.

Now, every time the Raspberry Pi starts, it will automatically enter the env virtual environment. If you want to exit, enter the command line: deactivate .

6，Install Donkey Car's Python Code

Create Directory, and Enter the Directory

In the user directory, create a folder named projects (directory), and enter the folder, enter the following command:

Note: Directly enter the user directory, just enter the command line: cd , press Enter.

Clone the entire Donkey car project from GitHub

Enter the following command:

git clone https://github.com/autorope/donkeycar

Then, start a long wait, because the speed of github in China is not fast, please be patient, if it doesn't work, run it again.

Note: If you are familiar with command line instructions, you can use the scp command to download some software projects after packaging, and upload them to the current directory of the Raspberry Pi. For specific usage, please search for it yourself.

When git clone (clone) is complete, we use the following command to operate:

pip install tensorflow==1.13.1

Note:
– When running pip install -e .[pi] installation, if an error occurs, you can try to run it again.
– When running pip install tensorflow==1.13.1 , it may prompt that the downloaded file's hash code is incorrect, causing it to be unable to download. You can manually use the wget command to download it according to the error prompt, download it, and then use pip install 「downloaded file name」 to install it.

Verify Tensorflow Installation Success

Enter the following command to enter the Python interactive programming mode:

Enter the following code to check:

print(tensorflow.__version__)

robocarstore/173807637225182120

When you see the version 1.13.1, it proves that you have successfully installed Tensorflow.

Note: Python will have some prompts when loading tensorflow for the first time, please don't worry, this is not an error prompt.

7，Install Python's OpenCV (optional)

Install OpenCV

This is also not required to be installed, if you don't use the OpenCV feature.

But if you still want to install it, please enter the following command:

pip install opencv-python

Verify Whether OpenCV Installation Success

Enter the following code to enter the Python interactive programming mode:

python import cv2 print(cv2.__version__)

August 23, 2019
in DonkeyCar
3 min read

Donkey Car详解(2) – 树莓派安装

robocarstore/173807615525023113

现在官方已经没有提供懒人版的Donkey Car 了，所以这次就分享一下，小编在树莓派（Raspberry Pi 3 B+）安装Donkey Car 的过程。

本文参考： http://docs.donkeycar.com/guide/robot_sbc/setup_raspberry_pi/

环境

官方推荐使用Raspian Lite(Stretch) (352MB)。但我使用的是带桌面的版本，如下

树莓派OS：
– Raspbian Buster with desktop
Version:July 2019
Release date:2019-07-10
Kernel version:4.19

下载地址：

https://www.raspberrypi.org/downloads/raspbian/

怎么刷写安装树莓派，联网，这里就不多说了。

假设你已经安装好树莓派系统，并且能联网，可以选择连接显示器，直接操作（本示例的操作模式），也可以远程SSH操作。你喜欢就好。

打开命令行，我们开始吧！

donkeycar-rpi

开始安装

1，在安装前更新你的系统

要更新你的系统，输入以下命令：

更新系统需要花费很大时间，而且可能需要重复多次才能更新完成。请耐心操作。

2，配置树莓派开启摄像头，I2C，扩展文件系统

要打开树莓派的配置对话框，在命令行输入：

① 开启I2C

05 Interfacing Options -> P5 I2C ->「Yes」

rpi_i2c

② 开启摄像头：

05 Interfacing Options -> P1 Camera -> 「Yes」

robocarstore/173807630025267416

③ 扩展文件系统

7 Advanced Options -> A1 Exapand Filesystem

robocarstore/173807631425048117

设置好，之后退出配置对话框，会提示是否现在重启，选择「Yes」重启系统。

robocarstore/173807632825076218

3，安装Donkey Car 需要的依赖项

在命令行中输入以下命令：

sudo apt-get install -y build-essential python3 python3-dev python3-pip python3-virtualenv python3-numpy python3-picamera python3-pandas python3-rpi.gpio i2c-tools avahi-utils joystick libopenjp2-7-dev libtiff5-dev gfortran libatlas-base-dev libopenblas-dev libhdf5-serial-dev git

经过漫长的等待终将安装完成，如果爆错误，请重新运行一遍。

（如果还是解决不了问题，请在留言区分享你所遇到的问题，请贴出你的系统版本，尽可能多的信息，我们才有可能帮助到你。）

4，安装OpenCV的依赖项（可选）

这一步骤可选，如果你没有用到OpenCV的功能可以不安装。

但如果还是想安装的话，在命令行输入以下命令：

sudo apt-get install libilmbase-dev libopenexr-dev libgstreamer1.0-dev libjasper-dev libwebp-dev libatlas-base-dev libavcodec-dev libavformat-dev libswscale-dev libqtgui4 libqt4-test

5，安装虚拟环境

在命令行输入：

python3 -m virtualenv -p python3 env --system-site-packages

echo "source env/bin/activate" >> ~/.bashrc

运行完以上命令，命令行会多了个（env）前缀，表示我们在一个叫作env的虚拟环境下，进行操作。如下图：

robocarstore/173807634625230519

虚拟环境的作用是，避免树莓派运行多个项目的时候，不会因为更新了某个依赖项或者自动清除了某个依赖项，在其他项目中出现了不可预测的错误。

现在，每次开机都会自动进入env的虚拟环境，如果想退出，命令行输入：deactivate ，即可。

6，安装Donkey Car 的 Python 代码

在用户目录下，新建一个名为projects 的文件夹（目录），并进入该文件夹，输入命令如下：

注意：直接进入用户目录，只需要在命令行输入： cd ，回车即可。

从GitHub上克隆整个Donkey car项目

输入以下命令：

git clone https://github.com/autorope/donkeycar

接着，又开始漫长的等待，因为github在国内的速度不是一般的快，请耐心等待，如果一次不行，则再运行一次。

提示：如果对命令行指令比较熟悉的，可使用scp指令，通过一些软件项目打包下载后上传到树莓派的当前目录。具体用法请自行搜索。

当git clone（克隆）完成，我们使用使用以下命令操作：

pip install tensorflow==1.13.1

提示：
– 当运行 pip install -e .[pi] 安装，出现错误时，可尝试再执行一次。
– 当运行 pip install tensorflow==1.13.1 ，可能会提示下载的文件的检验哈希码不正确，导致没有办法下载，可根据错误提示的下载地址，手动使用 wget 命令下载，下载完成后，使用 pip install 「下载的文件名称」，进行安装。

验证Tensorflow安装是否成功

命令行输入python进入Python的交互编程模式：

输入以下代码检查：

print(tensorflow.__version__)

robocarstore/173807637225182120

当看到1.13.1的版本好，证明你已经安装Tensorflow成功

注意：python第一次加载tensorflow会有一些提示，请不用担心，这不是错误提示。

7，安装Python的OpenCV（可选）

安装OpenCV

这个也是非必须安装的，如果你没有用到OpenCV的功能。

但如果你还是想安装，请输入以下命令：

pip install opencv-python

验证是否安装OpenCV成功

同样进入python的交互编程模式，输入以下代码：

import cv2
print(cv2.__version__)

August 23, 2019
in DonkeyCar
3 min read

Donkey Car详解(2) – 树莓派安装

robocarstore/173807615525023113

现在官方已经没有提供懒人版的Donkey Car 了，所以这次就分享一下，小编在树莓派（Raspberry Pi 3 B+）安装Donkey Car 的过程。

本文参考： http://docs.donkeycar.com/guide/robot_sbc/setup_raspberry_pi/

环境

官方推荐使用Raspian Lite(Stretch) (352MB)。但我使用的是带桌面的版本，如下

树莓派OS：
– Raspbian Buster with desktop
Version:July 2019
Release date:2019-07-10
Kernel version:4.19

下载地址：

https://www.raspberrypi.org/downloads/raspbian/

怎么刷写安装树莓派，联网，这里就不多说了。

假设你已经安装好树莓派系统，并且能联网，可以选择连接显示器，直接操作（本示例的操作模式），也可以远程SSH操作。你喜欢就好。

打开命令行，我们开始吧！

donkeycar-rpi

开始安装

1，在安装前更新你的系统

要更新你的系统，输入以下命令：

更新系统需要花费很大时间，而且可能需要重复多次才能更新完成。请耐心操作。

2，配置树莓派开启摄像头，I2C，扩展文件系统

要打开树莓派的配置对话框，在命令行输入：

① 开启I2C

05 Interfacing Options -> P5 I2C ->「Yes」

rpi_i2c

② 开启摄像头：

05 Interfacing Options -> P1 Camera -> 「Yes」

robocarstore/173807630025267416

③ 扩展文件系统

7 Advanced Options -> A1 Exapand Filesystem

robocarstore/173807631425048117

设置好，之后退出配置对话框，会提示是否现在重启，选择「Yes」重启系统。

robocarstore/173807632825076218

3，安装Donkey Car 需要的依赖项

在命令行中输入以下命令：

sudo apt-get install -y build-essential python3 python3-dev python3-pip python3-virtualenv python3-numpy python3-picamera python3-pandas python3-rpi.gpio i2c-tools avahi-utils joystick libopenjp2-7-dev libtiff5-dev gfortran libatlas-base-dev libopenblas-dev libhdf5-serial-dev git

经过漫长的等待终将安装完成，如果爆错误，请重新运行一遍。

（如果还是解决不了问题，请在留言区分享你所遇到的问题，请贴出你的系统版本，尽可能多的信息，我们才有可能帮助到你。）

4，安装OpenCV的依赖项（可选）

这一步骤可选，如果你没有用到OpenCV的功能可以不安装。

但如果还是想安装的话，在命令行输入以下命令：

sudo apt-get install libilmbase-dev libopenexr-dev libgstreamer1.0-dev libjasper-dev libwebp-dev libatlas-base-dev libavcodec-dev libavformat-dev libswscale-dev libqtgui4 libqt4-test

5，安装虚拟环境

在命令行输入：

python3 -m virtualenv -p python3 env --system-site-packages

echo "source env/bin/activate" >> ~/.bashrc

运行完以上命令，命令行会多了个（env）前缀，表示我们在一个叫作env的虚拟环境下，进行操作。如下图：

robocarstore/173807634625230519

虚拟环境的作用是，避免树莓派运行多个项目的时候，不会因为更新了某个依赖项或者自动清除了某个依赖项，在其他项目中出现了不可预测的错误。

现在，每次开机都会自动进入env的虚拟环境，如果想退出，命令行输入：deactivate ，即可。

6，安装Donkey Car 的 Python 代码

在用户目录下，新建一个名为projects 的文件夹（目录），并进入该文件夹，输入命令如下：

注意：直接进入用户目录，只需要在命令行输入： cd ，回车即可。

从GitHub上克隆整个Donkey car项目

输入以下命令：

git clone https://github.com/autorope/donkeycar

接着，又开始漫长的等待，因为github在国内的速度不是一般的快，请耐心等待，如果一次不行，则再运行一次。

提示：如果对命令行指令比较熟悉的，可使用scp指令，通过一些软件项目打包下载后上传到树莓派的当前目录。具体用法请自行搜索。

当git clone（克隆）完成，我们使用使用以下命令操作：

pip install tensorflow==1.13.1

提示：
– 当运行 pip install -e .[pi] 安装，出现错误时，可尝试再执行一次。
– 当运行 pip install tensorflow==1.13.1 ，可能会提示下载的文件的检验哈希码不正确，导致没有办法下载，可根据错误提示的下载地址，手动使用 wget 命令下载，下载完成后，使用 pip install 「下载的文件名称」，进行安装。

验证Tensorflow安装是否成功

命令行输入python进入Python的交互编程模式：

输入以下代码检查：

print(tensorflow.__version__)

robocarstore/173807637225182120

当看到1.13.1的版本好，证明你已经安装Tensorflow成功

注意：python第一次加载tensorflow会有一些提示，请不用担心，这不是错误提示。

7，安装Python的OpenCV（可选）

安装OpenCV

这个也是非必须安装的，如果你没有用到OpenCV的功能。

但如果你还是想安装，请输入以下命令：

pip install opencv-python

验证是否安装OpenCV成功

同样进入python的交互编程模式，输入以下代码：

import cv2
print(cv2.__version__)

August 23, 2019
in DonkeyCar
5 min read

Donkey Car Detailed Explanation (2) – Raspberry Pi Installation

robocarstore/173807615525023113

Now the official Donkey Car is no longer provided with a lazy person's version, so this time we will share the process of installing Donkey Car on a Raspberry Pi (Raspberry Pi 3 B+).

Reference: http://docs.donkeycar.com/guide/robot_sbc/setup_raspberry_pi/

Environment

The official recommendation is to use Raspian Lite(Stretch) (352MB). However, I used the desktop version, as follows:

Raspberry Pi OS:
– Raspbian Buster with desktop
Version:July 2019
Release date:2019-07-10
Kernel version:4.19

Download Url:

https://www.raspberrypi.org/downloads/raspbian/

How to flash install Raspberry Pi, here is no more.

Assume you have already installed the Raspberry Pi system and can connect to the network, you can choose to connect to the display directly (this example's operation mode), or remotely SSH. You like it.

Open the command line, let's start!

donkeycar-rpi

Start Installation

1，Update your system before installation

To update your system, enter the following command:

Updating the system takes a long time and may need to be repeated multiple times to complete. Please be patient.

2，Configure Raspberry Pi to enable camera, I2C, expand file system

To open the Raspberry Pi configuration dialog, enter the following command in the command line:

① Enable I2C

05 Interfacing Options -> P5 I2C ->「Yes」

rpi_i2c

② Enable Camera:

05 Interfacing Options -> P1 Camera -> 「Yes」

robocarstore/173807630025267416

③ Expand File System

7 Advanced Options -> A1 Exapand Filesystem

robocarstore/173807631425048117

Set it up, then exit the configuration dialog, it will prompt whether to restart now, select 「Yes」 to restart the system.

robocarstore/173807632825076218

3，Install the dependencies required for Donkey Car

Enter the following command in the command line:

sudo apt-get install -y build-essential python3 python3-dev python3-pip python3-virtualenv python3-numpy python3-picamera python3-pandas python3-rpi.gpio i2c-tools avahi-utils joystick libopenjp2-7-dev libtiff5-dev gfortran libatlas-base-dev libopenblas-dev libhdf5-serial-dev git

After a long wait, it will be installed, if an error occurs, please run it again.

（If the problem is still not resolved, please share the problem you encountered in the comments section, and post your system version, as much information as possible, so we can help you.）

4，Install OpenCV dependencies (optional)

This step is optional, you can skip it if you don't use the OpenCV feature.

If you still want to install it, enter the following command in the command line:

sudo apt-get install libilmbase-dev libopenexr-dev libgstreamer1.0-dev libjasper-dev libwebp-dev libatlas-base-dev libavcodec-dev libavformat-dev libswscale-dev libqtgui4 libqt4-test

5，Install Virtual Environment

Enter the following command in the command line:

python3 -m virtualenv -p python3 env --system-site-packages

echo "source env/bin/activate" >> ~/.bashrc

After running the above commands, the command line will have a (env) prefix, indicating that we are operating in an environment called env.

robocarstore/173807634625230519

The purpose of the virtual environment is to avoid errors in other projects when updating a dependency or automatically clearing a dependency on the Raspberry Pi running multiple projects.

Now, every time the Raspberry Pi starts, it will automatically enter the env virtual environment. If you want to exit, enter the command line: deactivate .

6，Install Donkey Car's Python Code

Create Directory, and Enter the Directory

In the user directory, create a folder named projects (directory), and enter the folder, enter the following command:

Note: Directly enter the user directory, just enter the command line: cd , press Enter.

Clone the entire Donkey car project from GitHub

Enter the following command:

git clone https://github.com/autorope/donkeycar

Then, start a long wait, because the speed of github in China is not fast, please be patient, if it doesn't work, run it again.

Note: If you are familiar with command line instructions, you can use the scp command to download some software projects after packaging, and upload them to the current directory of the Raspberry Pi. For specific usage, please search for it yourself.

When git clone (clone) is complete, we use the following command to operate:

pip install tensorflow==1.13.1

Note:
– When running pip install -e .[pi] installation, if an error occurs, you can try to run it again.
– When running pip install tensorflow==1.13.1 , it may prompt that the downloaded file's hash code is incorrect, causing it to be unable to download. You can manually use the wget command to download it according to the error prompt, download it, and then use pip install 「downloaded file name」 to install it.

Verify Tensorflow Installation Success

Enter the following command to enter the Python interactive programming mode:

Enter the following code to check:

print(tensorflow.__version__)

robocarstore/173807637225182120

When you see the version 1.13.1, it proves that you have successfully installed Tensorflow.

Note: Python will have some prompts when loading tensorflow for the first time, please don't worry, this is not an error prompt.

7，Install Python's OpenCV (optional)

Install OpenCV

This is also not required to be installed, if you don't use the OpenCV feature.

But if you still want to install it, please enter the following command:

pip install opencv-python

Verify Whether OpenCV Installation Success

Enter the following code to enter the Python interactive programming mode:

python import cv2 print(cv2.__version__)

August 18, 2019
in DonkeyCar
16 min read

Use Colab to Accelerate Donkey Car Training (Tensorflow GPU)

colab

This notebook helps you quickly train your Donkey car or autonomous car model.

Reference and improved the notebook of @sachindroid8, thanks to the previous generation!

The Pros and Cons of Using Colab for Training

Efficiency

Using Colab requires a VPN, and also needs to upload data to Google, the upload time varies from person to person, I uploaded it for about 2-3 minutes, the rest is executing the code and training. The first execution of the code needs to understand what the principle is, and the subsequent execution basically does not take much time, but the training time using GPU, I have 6k images, training about 30-40 epochs early, the training time is less than 1 minute! Each Epoch only needs 1-2 seconds.

Then using my Macbook Pro training, because of the lack of GPU, each Epoch needs 30-50 seconds, training down, it takes about 40 minutes to 1 hour to complete.

Convenience

If you don't have a VPN, of course, Colab is not a choice, I know Baidu also provides some limited free servers, which can be tested later, but with a VPN, Google Colab is the best choice. Below is how to start training

Load My Notebook

Open the Using Google_Colab to Donkey_Car

Install TensorFlow 1.14.0

TensorFlow 2.x and Donkey Car 3.x still have compatibility issues, so it is not recommended to use them temporarily (2019.8.17)In [1]:

!pip install tensorflow-gpu==1.14.0

Collecting tensorflow-gpu==1.14.0
  Downloading https://files.pythonhosted.org/packages/76/04/43153bfdfcf6c9a4c38ecdb971ca9a75b9a791bb69a764d652c359aca504/tensorflow_gpu-1.14.0-cp36-cp36m-manylinux1_x86_64.whl (377.0MB)
     |████████████████████████████████| 377.0MB 86kB/s 
Requirement already satisfied: grpcio&gt;=1.8.6 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.15.0)
Requirement already satisfied: absl-py&gt;=0.7.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (0.7.1)
Requirement already satisfied: tensorflow-estimator&lt;1.15.0rc0,&gt;=1.14.0rc0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.14.0)
Requirement already satisfied: wrapt&gt;=1.11.1 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.11.2)
Requirement already satisfied: termcolor&gt;=1.1.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.1.0)
Requirement already satisfied: astor&gt;=0.6.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (0.8.0)
Requirement already satisfied: keras-applications&gt;=1.0.6 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.0.8)
Requirement already satisfied: tensorboard&lt;1.15.0,&gt;=1.14.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.14.0)
Requirement already satisfied: wheel&gt;=0.26 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (0.33.4)
Requirement already satisfied: protobuf&gt;=3.6.1 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (3.7.1)
Requirement already satisfied: six&gt;=1.10.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.12.0)
Requirement already satisfied: keras-preprocessing&gt;=1.0.5 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.1.0)
Requirement already satisfied: google-pasta&gt;=0.1.6 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (0.1.7)
Requirement already satisfied: gast&gt;=0.2.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (0.2.2)
Requirement already satisfied: numpy&lt;2.0,&gt;=1.14.5 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.16.4)
Requirement already satisfied: h5py in /usr/local/lib/python3.6/dist-packages (from keras-applications&gt;=1.0.6-&gt;tensorflow-gpu==1.14.0) (2.8.0)
Requirement already satisfied: markdown&gt;=2.6.8 in /usr/local/lib/python3.6/dist-packages (from tensorboard&lt;1.15.0,&gt;=1.14.0-&gt;tensorflow-gpu==1.14.0) (3.1.1)
Requirement already satisfied: werkzeug&gt;=0.11.15 in /usr/local/lib/python3.6/dist-packages (from tensorboard&lt;1.15.0,&gt;=1.14.0-&gt;tensorflow-gpu==1.14.0) (0.15.5)
Requirement already satisfied: setuptools&gt;=41.0.0 in /usr/local/lib/python3.6/dist-packages (from tensorboard&lt;1.15.0,&gt;=1.14.0-&gt;tensorflow-gpu==1.14.0) (41.0.1)
Installing collected packages: tensorflow-gpu
Successfully installed tensorflow-gpu-1.14.0

Check if GPU is working

If you see “Found GPU at: / device: GPU: 0”, then the GPU is working normally

If you don't see the above output, you need to check if the Runtime (Run Type) has selected the GPU hardware acceleratorIn [2]:

device_name = tf.test.gpu_device_name()

if device_name != '/device:GPU:0':

raise SystemError('GPU device not found')

print('Found GPU at: {}'.format(device_name))

Found GPU at: /device:GPU:0

克隆Donkey respository

In [3]:

!git clone https://github.com/autorope/donkeycar.git donkey

Cloning into 'donkey'...
remote: Enumerating objects: 120, done.
remote: Counting objects: 100% (120/120), done.
remote: Compressing objects: 100% (76/76), done.
remote: Total 10558 (delta 65), reused 73 (delta 31), pack-reused 10438
Receiving objects: 100% (10558/10558), 58.74 MiB | 46.70 MiB/s, done.
Resolving deltas: 100% (6528/6528), done.

Install Donkey car ¶

In [4]:

Obtaining file:///content/donkey
Requirement already satisfied: numpy in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (1.16.4)
Requirement already satisfied: pillow in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (4.3.0)
Requirement already satisfied: docopt in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (0.6.2)
Requirement already satisfied: tornado in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (4.5.3)
Requirement already satisfied: requests in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (2.21.0)
Requirement already satisfied: h5py in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (2.8.0)
Requirement already satisfied: moviepy in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (0.2.3.5)
Requirement already satisfied: pandas in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (0.24.2)
Requirement already satisfied: PrettyTable in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (0.7.2)
Collecting paho-mqtt (from donkeycar==3.1.0)
  Downloading https://files.pythonhosted.org/packages/25/63/db25e62979c2a716a74950c9ed658dce431b5cb01fde29eb6cba9489a904/paho-mqtt-1.4.0.tar.gz (88kB)
     |████████████████████████████████| 92kB 4.2MB/s 
Requirement already satisfied: olefile in /usr/local/lib/python3.6/dist-packages (from pillow-&gt;donkeycar==3.1.0) (0.46)
Requirement already satisfied: urllib3&lt;1.25,&gt;=1.21.1 in /usr/local/lib/python3.6/dist-packages (from requests-&gt;donkeycar==3.1.0) (1.24.3)
Requirement already satisfied: certifi&gt;=2017.4.17 in /usr/local/lib/python3.6/dist-packages (from requests-&gt;donkeycar==3.1.0) (2019.6.16)
Requirement already satisfied: chardet&lt;3.1.0,&gt;=3.0.2 in /usr/local/lib/python3.6/dist-packages (from requests-&gt;donkeycar==3.1.0) (3.0.4)
Requirement already satisfied: idna&lt;2.9,&gt;=2.5 in /usr/local/lib/python3.6/dist-packages (from requests-&gt;donkeycar==3.1.0) (2.8)
Requirement already satisfied: six in /usr/local/lib/python3.6/dist-packages (from h5py-&gt;donkeycar==3.1.0) (1.12.0)
Requirement already satisfied: decorator&lt;5.0,&gt;=4.0.2 in /usr/local/lib/python3.6/dist-packages (from moviepy-&gt;donkeycar==3.1.0) (4.4.0)
Requirement already satisfied: tqdm&lt;5.0,&gt;=4.11.2 in /usr/local/lib/python3.6/dist-packages (from moviepy-&gt;donkeycar==3.1.0) (4.28.1)
Requirement already satisfied: imageio&lt;3.0,&gt;=2.1.2 in /usr/local/lib/python3.6/dist-packages (from moviepy-&gt;donkeycar==3.1.0) (2.4.1)
Requirement already satisfied: pytz&gt;=2011k in /usr/local/lib/python3.6/dist-packages (from pandas-&gt;donkeycar==3.1.0) (2018.9)
Requirement already satisfied: python-dateutil&gt;=2.5.0 in /usr/local/lib/python3.6/dist-packages (from pandas-&gt;donkeycar==3.1.0) (2.5.3)
Building wheels for collected packages: paho-mqtt
  Building wheel for paho-mqtt (setup.py) ... done
  Created wheel for paho-mqtt: filename=paho_mqtt-1.4.0-cp36-none-any.whl size=48333 sha256=9a67d0c95fae2b9495c20980895d9569ef53859cbc22bb57fc2466d7a581af7b
  Stored in directory: /root/.cache/pip/wheels/82/e5/de/d90d0f397648a1b58ffeea1b5742ac8c77f71fd43b550fa5a5
Successfully built paho-mqtt
Installing collected packages: paho-mqtt, donkeycar
  Running setup.py develop for donkeycar
Successfully installed donkeycar paho-mqtt-1.4.0

Create a project

I used mycar as the project name, you can change it, but you need to modify the code after renamingIn [5]:

!donkey createcar --path /content/mycar

using donkey v3.1.0 ...
Creating car folder: /content/mycar
making dir  /content/mycar
Creating data &amp; model folders.
making dir  /content/mycar/models
making dir  /content/mycar/data
making dir  /content/mycar/logs
Copying car application template: complete
Copying car config defaults. Adjust these before starting your car.
Copying train script. Adjust these before starting your car.
Copying my car config overrides
Donkey setup complete.

Prepare data: Upload data.zip and unzip

Now you need to package the data directory that you collected from pi, save it as data.zip.

Run the following command on the pi data directory:

$ zip -r data.zip tub_3_19-08-17/

Then copy it back to your computer, ready to upload to Colab

Upload data.zip to Colab

Run the following code, you will see an upload button, click to upload the data.zip you just packagedIn [7]:

from google.colab import files

if(os.path.exists("/content/data.zip")):

os.remove("/content/data.zip")

if(os.path.exists("/content/mycar/data/data.zip")):

os.remove("/content/mycar/data/data.zip")

uploaded = files.upload()

WORK_FOLDER = "/content/mycar/data/"

if(os.path.exists(WORK_FOLDER) == False):

!mv /content/data.zip /content/mycar/data/

Clean up the uploaded files

You need to ensure that the content/mycar/data directory has a tub directory, and the directory contains images and corresponding json files

data.zip is not neededIn [0]:

!rm /content/mycar/data/data.zip

Train the model

In [21]:

!python /content/mycar/manage.py train --model /content/mycar/models/mypilot.h5

using donkey v3.1.0 ...
loading config file: /content/mycar/config.py
loading personal config over-rides

config loaded
WARNING: Logging before flag parsing goes to stderr.
W0818 04:53:15.944086 140399927523200 deprecation_wrapper.py:119] From /content/donkey/donkeycar/parts/keras.py:18: The name tf.ConfigProto is deprecated. Please use tf.compat.v1.ConfigProto instead.

W0818 04:53:15.944333 140399927523200 deprecation_wrapper.py:119] From /content/donkey/donkeycar/parts/keras.py:18: The name tf.Session is deprecated. Please use tf.compat.v1.Session instead.

2019-08-18 04:53:15.954857: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
2019-08-18 04:53:15.959505: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcuda.so.1
2019-08-18 04:53:16.083690: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1005] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-08-18 04:53:16.084227: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x2455100 executing computations on platform CUDA. Devices:
2019-08-18 04:53:16.084262: I tensorflow/compiler/xla/service/service.cc:175]   StreamExecutor device (0): Tesla T4, Compute Capability 7.5
2019-08-18 04:53:16.086194: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 2300000000 Hz
2019-08-18 04:53:16.086460: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x65e6380 executing computations on platform Host. Devices:
2019-08-18 04:53:16.086508: I tensorflow/compiler/xla/service/service.cc:175]   StreamExecutor device (0): &lt;undefined&gt;, &lt;undefined&gt;
2019-08-18 04:53:16.086686: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1005] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-08-18 04:53:16.087039: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1640] Found device 0 with properties: 
name: Tesla T4 major: 7 minor: 5 memoryClockRate(GHz): 1.59
pciBusID: 0000:00:04.0
2019-08-18 04:53:16.087321: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcudart.so.10.0
2019-08-18 04:53:16.088548: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcublas.so.10.0
2019-08-18 04:53:16.089621: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcufft.so.10.0
2019-08-18 04:53:16.089935: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcurand.so.10.0
2019-08-18 04:53:16.091399: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcusolver.so.10.0
2019-08-18 04:53:16.092394: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcusparse.so.10.0
2019-08-18 04:53:16.095470: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcudnn.so.7
2019-08-18 04:53:16.095606: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1005] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-08-18 04:53:16.096009: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1005] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-08-18 04:53:16.096350: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1763] Adding visible gpu devices: 0
2019-08-18 04:53:16.096401: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcudart.so.10.0
2019-08-18 04:53:16.097166: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1181] Device interconnect StreamExecutor with strength 1 edge matrix:
2019-08-18 04:53:16.097190: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1187]      0 
2019-08-18 04:53:16.097201: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1200] 0:   N 
2019-08-18 04:53:16.097482: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1005] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-08-18 04:53:16.097878: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1005] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-08-18 04:53:16.098226: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1326] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 14089 MB memory) -&gt; physical GPU (device: 0, name: Tesla T4, pci bus id: 0000:00:04.0, compute capability: 7.5)
"get_model_by_type" model Type is: linear
W0818 04:53:16.444941 140399927523200 deprecation.py:506] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/init_ops.py:1251: calling VarianceScaling.__init__ (from tensorflow.python.ops.init_ops) with dtype is deprecated and will be removed in a future version.
Instructions for updating:
Call initializer instance with the dtype argument instead of passing it to the constructor
training with model type &lt;class 'donkeycar.parts.keras.KerasLinear'&gt;
Model: "model"
__________________________________________________________________________________________________
Layer (type)                    Output Shape         Param #     Connected to                     
==================================================================================================
img_in (InputLayer)             [(None, 120, 160, 3) 0                                            
__________________________________________________________________________________________________
conv2d_1 (Conv2D)               (None, 58, 78, 24)   1824        img_in[0][0]                     
__________________________________________________________________________________________________
dropout (Dropout)               (None, 58, 78, 24)   0           conv2d_1[0][0]                   
__________________________________________________________________________________________________
conv2d_2 (Conv2D)               (None, 27, 37, 32)   19232       dropout[0][0]                    
__________________________________________________________________________________________________
dropout_1 (Dropout)             (None, 27, 37, 32)   0           conv2d_2[0][0]                   
__________________________________________________________________________________________________
conv2d_3 (Conv2D)               (None, 12, 17, 64)   51264       dropout_1[0][0]                  
__________________________________________________________________________________________________
dropout_2 (Dropout)             (None, 12, 17, 64)   0           conv2d_3[0][0]                   
__________________________________________________________________________________________________
conv2d_4 (Conv2D)               (None, 10, 15, 64)   36928       dropout_2[0][0]                  
__________________________________________________________________________________________________
dropout_3 (Dropout)             (None, 10, 15, 64)   0           conv2d_4[0][0]                   
__________________________________________________________________________________________________
conv2d_5 (Conv2D)               (None, 8, 13, 64)    36928       dropout_3[0][0]                  
__________________________________________________________________________________________________
dropout_4 (Dropout)             (None, 8, 13, 64)    0           conv2d_5[0][0]                   
__________________________________________________________________________________________________
flattened (Flatten)             (None, 6656)         0           dropout_4[0][0]                  
__________________________________________________________________________________________________
dense (Dense)                   (None, 100)          665700      flattened[0][0]                  
__________________________________________________________________________________________________
dropout_5 (Dropout)             (None, 100)          0           dense[0][0]                      
__________________________________________________________________________________________________
dense_1 (Dense)                 (None, 50)           5050        dropout_5[0][0]                  
__________________________________________________________________________________________________
dropout_6 (Dropout)             (None, 50)           0           dense_1[0][0]                    
__________________________________________________________________________________________________
n_outputs0 (Dense)              (None, 1)            51          dropout_6[0][0]                  
__________________________________________________________________________________________________
n_outputs1 (Dense)              (None, 1)            51          dropout_6[0][0]                  
==================================================================================================
Total params: 817,028
Trainable params: 817,028
Non-trainable params: 0
__________________________________________________________________________________________________
None
found 0 pickles writing json records and images in tub /content/mycar/data/tub_3_19-08-17
/content/mycar/data/tub_3_19-08-17
collating 5799 records ...
train: 4639, val: 1160
total records: 5799
steps_per_epoch 36
Epoch 1/100
2019-08-18 04:53:19.663907: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcublas.so.10.0
2019-08-18 04:53:19.984999: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcudnn.so.7
35/36 [============================&gt;.] - ETA: 0s - loss: 0.1681 - n_outputs0_loss: 0.1555 - n_outputs1_loss: 0.0126
Epoch 00001: val_loss improved from inf to 0.14482, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 8s 213ms/step - loss: 0.1677 - n_outputs0_loss: 0.1553 - n_outputs1_loss: 0.0124 - val_loss: 0.1448 - val_n_outputs0_loss: 0.1409 - val_n_outputs1_loss: 0.0039
Epoch 2/100
34/36 [===========================&gt;..] - ETA: 0s - loss: 0.1278 - n_outputs0_loss: 0.1234 - n_outputs1_loss: 0.0044
Epoch 00002: val_loss improved from 0.14482 to 0.08814, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 53ms/step - loss: 0.1257 - n_outputs0_loss: 0.1214 - n_outputs1_loss: 0.0043 - val_loss: 0.0881 - val_n_outputs0_loss: 0.0870 - val_n_outputs1_loss: 0.0012
Epoch 3/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0943 - n_outputs0_loss: 0.0910 - n_outputs1_loss: 0.0033
Epoch 00003: val_loss improved from 0.08814 to 0.07490, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 52ms/step - loss: 0.0938 - n_outputs0_loss: 0.0905 - n_outputs1_loss: 0.0033 - val_loss: 0.0749 - val_n_outputs0_loss: 0.0738 - val_n_outputs1_loss: 0.0011
Epoch 4/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0840 - n_outputs0_loss: 0.0813 - n_outputs1_loss: 0.0027
Epoch 00004: val_loss improved from 0.07490 to 0.07108, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 51ms/step - loss: 0.0835 - n_outputs0_loss: 0.0808 - n_outputs1_loss: 0.0027 - val_loss: 0.0711 - val_n_outputs0_loss: 0.0702 - val_n_outputs1_loss: 8.9668e-04
Epoch 5/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0767 - n_outputs0_loss: 0.0741 - n_outputs1_loss: 0.0026
Epoch 00005: val_loss did not improve from 0.07108
36/36 [==============================] - 2s 50ms/step - loss: 0.0765 - n_outputs0_loss: 0.0739 - n_outputs1_loss: 0.0026 - val_loss: 0.0722 - val_n_outputs0_loss: 0.0717 - val_n_outputs1_loss: 5.4629e-04
Epoch 6/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0750 - n_outputs0_loss: 0.0727 - n_outputs1_loss: 0.0023
Epoch 00006: val_loss improved from 0.07108 to 0.06483, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 52ms/step - loss: 0.0749 - n_outputs0_loss: 0.0725 - n_outputs1_loss: 0.0024 - val_loss: 0.0648 - val_n_outputs0_loss: 0.0641 - val_n_outputs1_loss: 7.5336e-04
Epoch 7/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0711 - n_outputs0_loss: 0.0689 - n_outputs1_loss: 0.0022
Epoch 00007: val_loss improved from 0.06483 to 0.06324, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 52ms/step - loss: 0.0711 - n_outputs0_loss: 0.0689 - n_outputs1_loss: 0.0021 - val_loss: 0.0632 - val_n_outputs0_loss: 0.0626 - val_n_outputs1_loss: 5.9058e-04
Epoch 8/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0693 - n_outputs0_loss: 0.0675 - n_outputs1_loss: 0.0018
Epoch 00008: val_loss did not improve from 0.06324
36/36 [==============================] - 2s 51ms/step - loss: 0.0690 - n_outputs0_loss: 0.0672 - n_outputs1_loss: 0.0018 - val_loss: 0.0644 - val_n_outputs0_loss: 0.0637 - val_n_outputs1_loss: 7.8261e-04
Epoch 9/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0693 - n_outputs0_loss: 0.0675 - n_outputs1_loss: 0.0018
Epoch 00009: val_loss did not improve from 0.06324
36/36 [==============================] - 2s 50ms/step - loss: 0.0690 - n_outputs0_loss: 0.0672 - n_outputs1_loss: 0.0018 - val_loss: 0.0653 - val_n_outputs0_loss: 0.0646 - val_n_outputs1_loss: 6.6657e-04
Epoch 10/100
34/36 [===========================&gt;..] - ETA: 0s - loss: 0.0651 - n_outputs0_loss: 0.0633 - n_outputs1_loss: 0.0018
Epoch 00010: val_loss did not improve from 0.06324
36/36 [==============================] - 2s 51ms/step - loss: 0.0657 - n_outputs0_loss: 0.0639 - n_outputs1_loss: 0.0018 - val_loss: 0.0699 - val_n_outputs0_loss: 0.0692 - val_n_outputs1_loss: 6.9729e-04
Epoch 11/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0630 - n_outputs0_loss: 0.0612 - n_outputs1_loss: 0.0018
Epoch 00011: val_loss did not improve from 0.06324
36/36 [==============================] - 2s 51ms/step - loss: 0.0630 - n_outputs0_loss: 0.0612 - n_outputs1_loss: 0.0018 - val_loss: 0.0633 - val_n_outputs0_loss: 0.0619 - val_n_outputs1_loss: 0.0014
Epoch 12/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0643 - n_outputs0_loss: 0.0626 - n_outputs1_loss: 0.0016
Epoch 00012: val_loss improved from 0.06324 to 0.06034, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 52ms/step - loss: 0.0644 - n_outputs0_loss: 0.0628 - n_outputs1_loss: 0.0017 - val_loss: 0.0603 - val_n_outputs0_loss: 0.0598 - val_n_outputs1_loss: 4.9525e-04
Epoch 13/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0611 - n_outputs0_loss: 0.0597 - n_outputs1_loss: 0.0014
Epoch 00013: val_loss improved from 0.06034 to 0.05636, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 52ms/step - loss: 0.0613 - n_outputs0_loss: 0.0599 - n_outputs1_loss: 0.0014 - val_loss: 0.0564 - val_n_outputs0_loss: 0.0557 - val_n_outputs1_loss: 6.3083e-04
Epoch 14/100
34/36 [===========================&gt;..] - ETA: 0s - loss: 0.0593 - n_outputs0_loss: 0.0580 - n_outputs1_loss: 0.0013
Epoch 00014: val_loss did not improve from 0.05636
36/36 [==============================] - 2s 51ms/step - loss: 0.0590 - n_outputs0_loss: 0.0577 - n_outputs1_loss: 0.0013 - val_loss: 0.0580 - val_n_outputs0_loss: 0.0575 - val_n_outputs1_loss: 5.4033e-04
Epoch 15/100
34/36 [===========================&gt;..] - ETA: 0s - loss: 0.0552 - n_outputs0_loss: 0.0540 - n_outputs1_loss: 0.0012
Epoch 00015: val_loss did not improve from 0.05636
36/36 [==============================] - 2s 51ms/step - loss: 0.0555 - n_outputs0_loss: 0.0542 - n_outputs1_loss: 0.0012 - val_loss: 0.0565 - val_n_outputs0_loss: 0.0559 - val_n_outputs1_loss: 5.6679e-04
Epoch 16/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0538 - n_outputs0_loss: 0.0527 - n_outputs1_loss: 0.0012
Epoch 00016: val_loss did not improve from 0.05636
36/36 [==============================] - 2s 50ms/step - loss: 0.0538 - n_outputs0_loss: 0.0526 - n_outputs1_loss: 0.0012 - val_loss: 0.0624 - val_n_outputs0_loss: 0.0619 - val_n_outputs1_loss: 5.0014e-04
Epoch 17/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0536 - n_outputs0_loss: 0.0525 - n_outputs1_loss: 0.0011
Epoch 00017: val_loss improved from 0.05636 to 0.05615, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 53ms/step - loss: 0.0534 - n_outputs0_loss: 0.0523 - n_outputs1_loss: 0.0011 - val_loss: 0.0561 - val_n_outputs0_loss: 0.0556 - val_n_outputs1_loss: 5.2067e-04
Epoch 18/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0502 - n_outputs0_loss: 0.0492 - n_outputs1_loss: 0.0011
Epoch 00018: val_loss improved from 0.05615 to 0.05594, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 52ms/step - loss: 0.0501 - n_outputs0_loss: 0.0490 - n_outputs1_loss: 0.0011 - val_loss: 0.0559 - val_n_outputs0_loss: 0.0555 - val_n_outputs1_loss: 4.4089e-04
Epoch 00018: early stopping
Training completed in 0:00:40.


----------- Best Eval Loss :0.055939 ---------
&lt;Figure size 640x480 with 1 Axes&gt;

Display the Loss curve of the training result

In [26]:

import matplotlib.pyplot as plt

list_of_png = glob.glob('/content/mycar/models/*png')

latest_png = max(list_of_png, key=os.path.getctime)

image = cv2.imread(latest_png)

/content/mycar/models/mypilot.h5_loss_acc_0.055939.png

Out[26]:

&lt;matplotlib.image.AxesImage at 0x7f867c896be0&gt;

robocarstore/173807730625008222

Put the trained model back into Donkey Car.

Once the model is trained, you can find the trained model in the mycar/models directory
1. Download the mypilot file to your PC or Mac

Copy the model from your PC or Mac to your pi

Finally, use the Autopilot mode to drive Donkey car

Enjoy!

August 18, 2019
in DonkeyCar
15 min read

使用Colab来加速Donkey Car训练 (Tensorflow GPU)

colab

本notebook助你快速训练你的Donkey car或者自动驾驶小车模型。

参考并改进了@sachindroid8的notebook，先向前人致敬！

先说说使用Colab进行训练的优缺点

效率

使用Colab需要梯子，也要把数据传给google，上传时间因人而异，我上传大概花来2-3分钟，剩下就是执行代码和训练。第一执行代码需要搞懂什么原理，后面执行基本不用花什么时间，然而使用GPU训练时间，我有6k图片，训练30-40个Epoch左右提前结束，训练时间少于1分钟！每个Epoch只需1-2秒。

然后使用我的Macbook Pro训练，因为没有GPU的缘故，每个Epoc需要30-50秒，训练下来，接近40分钟到1小时才能完成。

便利性

如果没有梯子，当然Colab不是一个选择，我知道百度也提供一些有限度免费的服务器，以后可以再做测试，但有梯子的话，Google Colab是最好的选择。下面便开始教你怎样开始训练

载入我的Notebook

打开使用Google_Colab来Donkey_Car

安装 TensorFlow 1.14.0

TensorFlow 2.x和Donkey Car 3.x现在还有兼容性问题，暂时不推荐使用(2019.8.17)In [1]:

!pip install tensorflow-gpu==1.14.0

Collecting tensorflow-gpu==1.14.0
  Downloading https://files.pythonhosted.org/packages/76/04/43153bfdfcf6c9a4c38ecdb971ca9a75b9a791bb69a764d652c359aca504/tensorflow_gpu-1.14.0-cp36-cp36m-manylinux1_x86_64.whl (377.0MB)
     |████████████████████████████████| 377.0MB 86kB/s 
Requirement already satisfied: grpcio&gt;=1.8.6 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.15.0)
Requirement already satisfied: absl-py&gt;=0.7.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (0.7.1)
Requirement already satisfied: tensorflow-estimator&lt;1.15.0rc0,&gt;=1.14.0rc0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.14.0)
Requirement already satisfied: wrapt&gt;=1.11.1 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.11.2)
Requirement already satisfied: termcolor&gt;=1.1.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.1.0)
Requirement already satisfied: astor&gt;=0.6.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (0.8.0)
Requirement already satisfied: keras-applications&gt;=1.0.6 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.0.8)
Requirement already satisfied: tensorboard&lt;1.15.0,&gt;=1.14.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.14.0)
Requirement already satisfied: wheel&gt;=0.26 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (0.33.4)
Requirement already satisfied: protobuf&gt;=3.6.1 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (3.7.1)
Requirement already satisfied: six&gt;=1.10.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.12.0)
Requirement already satisfied: keras-preprocessing&gt;=1.0.5 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.1.0)
Requirement already satisfied: google-pasta&gt;=0.1.6 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (0.1.7)
Requirement already satisfied: gast&gt;=0.2.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (0.2.2)
Requirement already satisfied: numpy&lt;2.0,&gt;=1.14.5 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.16.4)
Requirement already satisfied: h5py in /usr/local/lib/python3.6/dist-packages (from keras-applications&gt;=1.0.6-&gt;tensorflow-gpu==1.14.0) (2.8.0)
Requirement already satisfied: markdown&gt;=2.6.8 in /usr/local/lib/python3.6/dist-packages (from tensorboard&lt;1.15.0,&gt;=1.14.0-&gt;tensorflow-gpu==1.14.0) (3.1.1)
Requirement already satisfied: werkzeug&gt;=0.11.15 in /usr/local/lib/python3.6/dist-packages (from tensorboard&lt;1.15.0,&gt;=1.14.0-&gt;tensorflow-gpu==1.14.0) (0.15.5)
Requirement already satisfied: setuptools&gt;=41.0.0 in /usr/local/lib/python3.6/dist-packages (from tensorboard&lt;1.15.0,&gt;=1.14.0-&gt;tensorflow-gpu==1.14.0) (41.0.1)
Installing collected packages: tensorflow-gpu
Successfully installed tensorflow-gpu-1.14.0

检查GPU是否有效

如果显示”Found GPU at: / device: GPU: 0“，则GPU可以正常使用

如果没有以上输出，需要检查Runtime （运行类型）是否选择了GPU硬件加速器In [2]:

device_name = tf.test.gpu_device_name()

if device_name != '/device:GPU:0':

raise SystemError('GPU device not found')

print('Found GPU at: {}'.format(device_name))

Found GPU at: /device:GPU:0

克隆Donkey respository

In [3]:

!git clone https://github.com/autorope/donkeycar.git donkey

Cloning into 'donkey'...
remote: Enumerating objects: 120, done.
remote: Counting objects: 100% (120/120), done.
remote: Compressing objects: 100% (76/76), done.
remote: Total 10558 (delta 65), reused 73 (delta 31), pack-reused 10438
Receiving objects: 100% (10558/10558), 58.74 MiB | 46.70 MiB/s, done.
Resolving deltas: 100% (6528/6528), done.

安装 Donkey car ¶

In [4]:

Obtaining file:///content/donkey
Requirement already satisfied: numpy in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (1.16.4)
Requirement already satisfied: pillow in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (4.3.0)
Requirement already satisfied: docopt in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (0.6.2)
Requirement already satisfied: tornado in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (4.5.3)
Requirement already satisfied: requests in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (2.21.0)
Requirement already satisfied: h5py in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (2.8.0)
Requirement already satisfied: moviepy in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (0.2.3.5)
Requirement already satisfied: pandas in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (0.24.2)
Requirement already satisfied: PrettyTable in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (0.7.2)
Collecting paho-mqtt (from donkeycar==3.1.0)
  Downloading https://files.pythonhosted.org/packages/25/63/db25e62979c2a716a74950c9ed658dce431b5cb01fde29eb6cba9489a904/paho-mqtt-1.4.0.tar.gz (88kB)
     |████████████████████████████████| 92kB 4.2MB/s 
Requirement already satisfied: olefile in /usr/local/lib/python3.6/dist-packages (from pillow-&gt;donkeycar==3.1.0) (0.46)
Requirement already satisfied: urllib3&lt;1.25,&gt;=1.21.1 in /usr/local/lib/python3.6/dist-packages (from requests-&gt;donkeycar==3.1.0) (1.24.3)
Requirement already satisfied: certifi&gt;=2017.4.17 in /usr/local/lib/python3.6/dist-packages (from requests-&gt;donkeycar==3.1.0) (2019.6.16)
Requirement already satisfied: chardet&lt;3.1.0,&gt;=3.0.2 in /usr/local/lib/python3.6/dist-packages (from requests-&gt;donkeycar==3.1.0) (3.0.4)
Requirement already satisfied: idna&lt;2.9,&gt;=2.5 in /usr/local/lib/python3.6/dist-packages (from requests-&gt;donkeycar==3.1.0) (2.8)
Requirement already satisfied: six in /usr/local/lib/python3.6/dist-packages (from h5py-&gt;donkeycar==3.1.0) (1.12.0)
Requirement already satisfied: decorator&lt;5.0,&gt;=4.0.2 in /usr/local/lib/python3.6/dist-packages (from moviepy-&gt;donkeycar==3.1.0) (4.4.0)
Requirement already satisfied: tqdm&lt;5.0,&gt;=4.11.2 in /usr/local/lib/python3.6/dist-packages (from moviepy-&gt;donkeycar==3.1.0) (4.28.1)
Requirement already satisfied: imageio&lt;3.0,&gt;=2.1.2 in /usr/local/lib/python3.6/dist-packages (from moviepy-&gt;donkeycar==3.1.0) (2.4.1)
Requirement already satisfied: pytz&gt;=2011k in /usr/local/lib/python3.6/dist-packages (from pandas-&gt;donkeycar==3.1.0) (2018.9)
Requirement already satisfied: python-dateutil&gt;=2.5.0 in /usr/local/lib/python3.6/dist-packages (from pandas-&gt;donkeycar==3.1.0) (2.5.3)
Building wheels for collected packages: paho-mqtt
  Building wheel for paho-mqtt (setup.py) ... done
  Created wheel for paho-mqtt: filename=paho_mqtt-1.4.0-cp36-none-any.whl size=48333 sha256=9a67d0c95fae2b9495c20980895d9569ef53859cbc22bb57fc2466d7a581af7b
  Stored in directory: /root/.cache/pip/wheels/82/e5/de/d90d0f397648a1b58ffeea1b5742ac8c77f71fd43b550fa5a5
Successfully built paho-mqtt
Installing collected packages: paho-mqtt, donkeycar
  Running setup.py develop for donkeycar
Successfully installed donkeycar paho-mqtt-1.4.0

创建项目

我使用了mycar作为项目名称，你可以改名，但改名后需要相应修改后面但代码In [5]:

!donkey createcar --path /content/mycar

using donkey v3.1.0 ...
Creating car folder: /content/mycar
making dir  /content/mycar
Creating data &amp; model folders.
making dir  /content/mycar/models
making dir  /content/mycar/data
making dir  /content/mycar/logs
Copying car application template: complete
Copying car config defaults. Adjust these before starting your car.
Copying train script. Adjust these before starting your car.
Copying my car config overrides
Donkey setup complete.

准备数据: 上传data.zip并解压

现在你需要把pi采集回来的data目录上需要训练的目录打包，保存成data.zip.

在pi的data目录上运行:

$ zip -r data.zip tub_3_19-08-17/

然后拷贝会电脑，准备上传到Colab

上传data.zip到Colab

运行下面代码，会出现一个上传按钮，点击上传刚才打包的data.zipIn [7]:

from google.colab import files

if(os.path.exists("/content/data.zip")):

os.remove("/content/data.zip")

if(os.path.exists("/content/mycar/data/data.zip")):

os.remove("/content/mycar/data/data.zip")

uploaded = files.upload()

WORK_FOLDER = "/content/mycar/data/"

if(os.path.exists(WORK_FOLDER) == False):

!mv /content/data.zip /content/mycar/data/

清理已经上传的文件

你需要确保content/mycar/data目录下有tub目录，目录里面有图片和对应的json文件

data.zip就不用保留了In [0]:

!rm /content/mycar/data/data.zip

训练模型

In [21]:

!python /content/mycar/manage.py train --model /content/mycar/models/mypilot.h5

using donkey v3.1.0 ...
loading config file: /content/mycar/config.py
loading personal config over-rides

config loaded
WARNING: Logging before flag parsing goes to stderr.
W0818 04:53:15.944086 140399927523200 deprecation_wrapper.py:119] From /content/donkey/donkeycar/parts/keras.py:18: The name tf.ConfigProto is deprecated. Please use tf.compat.v1.ConfigProto instead.

W0818 04:53:15.944333 140399927523200 deprecation_wrapper.py:119] From /content/donkey/donkeycar/parts/keras.py:18: The name tf.Session is deprecated. Please use tf.compat.v1.Session instead.

2019-08-18 04:53:15.954857: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
2019-08-18 04:53:15.959505: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcuda.so.1
2019-08-18 04:53:16.083690: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1005] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-08-18 04:53:16.084227: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x2455100 executing computations on platform CUDA. Devices:
2019-08-18 04:53:16.084262: I tensorflow/compiler/xla/service/service.cc:175]   StreamExecutor device (0): Tesla T4, Compute Capability 7.5
2019-08-18 04:53:16.086194: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 2300000000 Hz
2019-08-18 04:53:16.086460: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x65e6380 executing computations on platform Host. Devices:
2019-08-18 04:53:16.086508: I tensorflow/compiler/xla/service/service.cc:175]   StreamExecutor device (0): &lt;undefined&gt;, &lt;undefined&gt;
2019-08-18 04:53:16.086686: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1005] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-08-18 04:53:16.087039: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1640] Found device 0 with properties: 
name: Tesla T4 major: 7 minor: 5 memoryClockRate(GHz): 1.59
pciBusID: 0000:00:04.0
2019-08-18 04:53:16.087321: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcudart.so.10.0
2019-08-18 04:53:16.088548: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcublas.so.10.0
2019-08-18 04:53:16.089621: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcufft.so.10.0
2019-08-18 04:53:16.089935: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcurand.so.10.0
2019-08-18 04:53:16.091399: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcusolver.so.10.0
2019-08-18 04:53:16.092394: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcusparse.so.10.0
2019-08-18 04:53:16.095470: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcudnn.so.7
2019-08-18 04:53:16.095606: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1005] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-08-18 04:53:16.096009: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1005] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-08-18 04:53:16.096350: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1763] Adding visible gpu devices: 0
2019-08-18 04:53:16.096401: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcudart.so.10.0
2019-08-18 04:53:16.097166: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1181] Device interconnect StreamExecutor with strength 1 edge matrix:
2019-08-18 04:53:16.097190: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1187]      0 
2019-08-18 04:53:16.097201: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1200] 0:   N 
2019-08-18 04:53:16.097482: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1005] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-08-18 04:53:16.097878: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1005] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-08-18 04:53:16.098226: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1326] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 14089 MB memory) -&gt; physical GPU (device: 0, name: Tesla T4, pci bus id: 0000:00:04.0, compute capability: 7.5)
"get_model_by_type" model Type is: linear
W0818 04:53:16.444941 140399927523200 deprecation.py:506] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/init_ops.py:1251: calling VarianceScaling.__init__ (from tensorflow.python.ops.init_ops) with dtype is deprecated and will be removed in a future version.
Instructions for updating:
Call initializer instance with the dtype argument instead of passing it to the constructor
training with model type &lt;class 'donkeycar.parts.keras.KerasLinear'&gt;
Model: "model"
__________________________________________________________________________________________________
Layer (type)                    Output Shape         Param #     Connected to                     
==================================================================================================
img_in (InputLayer)             [(None, 120, 160, 3) 0                                            
__________________________________________________________________________________________________
conv2d_1 (Conv2D)               (None, 58, 78, 24)   1824        img_in[0][0]                     
__________________________________________________________________________________________________
dropout (Dropout)               (None, 58, 78, 24)   0           conv2d_1[0][0]                   
__________________________________________________________________________________________________
conv2d_2 (Conv2D)               (None, 27, 37, 32)   19232       dropout[0][0]                    
__________________________________________________________________________________________________
dropout_1 (Dropout)             (None, 27, 37, 32)   0           conv2d_2[0][0]                   
__________________________________________________________________________________________________
conv2d_3 (Conv2D)               (None, 12, 17, 64)   51264       dropout_1[0][0]                  
__________________________________________________________________________________________________
dropout_2 (Dropout)             (None, 12, 17, 64)   0           conv2d_3[0][0]                   
__________________________________________________________________________________________________
conv2d_4 (Conv2D)               (None, 10, 15, 64)   36928       dropout_2[0][0]                  
__________________________________________________________________________________________________
dropout_3 (Dropout)             (None, 10, 15, 64)   0           conv2d_4[0][0]                   
__________________________________________________________________________________________________
conv2d_5 (Conv2D)               (None, 8, 13, 64)    36928       dropout_3[0][0]                  
__________________________________________________________________________________________________
dropout_4 (Dropout)             (None, 8, 13, 64)    0           conv2d_5[0][0]                   
__________________________________________________________________________________________________
flattened (Flatten)             (None, 6656)         0           dropout_4[0][0]                  
__________________________________________________________________________________________________
dense (Dense)                   (None, 100)          665700      flattened[0][0]                  
__________________________________________________________________________________________________
dropout_5 (Dropout)             (None, 100)          0           dense[0][0]                      
__________________________________________________________________________________________________
dense_1 (Dense)                 (None, 50)           5050        dropout_5[0][0]                  
__________________________________________________________________________________________________
dropout_6 (Dropout)             (None, 50)           0           dense_1[0][0]                    
__________________________________________________________________________________________________
n_outputs0 (Dense)              (None, 1)            51          dropout_6[0][0]                  
__________________________________________________________________________________________________
n_outputs1 (Dense)              (None, 1)            51          dropout_6[0][0]                  
==================================================================================================
Total params: 817,028
Trainable params: 817,028
Non-trainable params: 0
__________________________________________________________________________________________________
None
found 0 pickles writing json records and images in tub /content/mycar/data/tub_3_19-08-17
/content/mycar/data/tub_3_19-08-17
collating 5799 records ...
train: 4639, val: 1160
total records: 5799
steps_per_epoch 36
Epoch 1/100
2019-08-18 04:53:19.663907: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcublas.so.10.0
2019-08-18 04:53:19.984999: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcudnn.so.7
35/36 [============================&gt;.] - ETA: 0s - loss: 0.1681 - n_outputs0_loss: 0.1555 - n_outputs1_loss: 0.0126
Epoch 00001: val_loss improved from inf to 0.14482, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 8s 213ms/step - loss: 0.1677 - n_outputs0_loss: 0.1553 - n_outputs1_loss: 0.0124 - val_loss: 0.1448 - val_n_outputs0_loss: 0.1409 - val_n_outputs1_loss: 0.0039
Epoch 2/100
34/36 [===========================&gt;..] - ETA: 0s - loss: 0.1278 - n_outputs0_loss: 0.1234 - n_outputs1_loss: 0.0044
Epoch 00002: val_loss improved from 0.14482 to 0.08814, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 53ms/step - loss: 0.1257 - n_outputs0_loss: 0.1214 - n_outputs1_loss: 0.0043 - val_loss: 0.0881 - val_n_outputs0_loss: 0.0870 - val_n_outputs1_loss: 0.0012
Epoch 3/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0943 - n_outputs0_loss: 0.0910 - n_outputs1_loss: 0.0033
Epoch 00003: val_loss improved from 0.08814 to 0.07490, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 52ms/step - loss: 0.0938 - n_outputs0_loss: 0.0905 - n_outputs1_loss: 0.0033 - val_loss: 0.0749 - val_n_outputs0_loss: 0.0738 - val_n_outputs1_loss: 0.0011
Epoch 4/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0840 - n_outputs0_loss: 0.0813 - n_outputs1_loss: 0.0027
Epoch 00004: val_loss improved from 0.07490 to 0.07108, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 51ms/step - loss: 0.0835 - n_outputs0_loss: 0.0808 - n_outputs1_loss: 0.0027 - val_loss: 0.0711 - val_n_outputs0_loss: 0.0702 - val_n_outputs1_loss: 8.9668e-04
Epoch 5/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0767 - n_outputs0_loss: 0.0741 - n_outputs1_loss: 0.0026
Epoch 00005: val_loss did not improve from 0.07108
36/36 [==============================] - 2s 50ms/step - loss: 0.0765 - n_outputs0_loss: 0.0739 - n_outputs1_loss: 0.0026 - val_loss: 0.0722 - val_n_outputs0_loss: 0.0717 - val_n_outputs1_loss: 5.4629e-04
Epoch 6/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0750 - n_outputs0_loss: 0.0727 - n_outputs1_loss: 0.0023
Epoch 00006: val_loss improved from 0.07108 to 0.06483, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 52ms/step - loss: 0.0749 - n_outputs0_loss: 0.0725 - n_outputs1_loss: 0.0024 - val_loss: 0.0648 - val_n_outputs0_loss: 0.0641 - val_n_outputs1_loss: 7.5336e-04
Epoch 7/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0711 - n_outputs0_loss: 0.0689 - n_outputs1_loss: 0.0022
Epoch 00007: val_loss improved from 0.06483 to 0.06324, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 52ms/step - loss: 0.0711 - n_outputs0_loss: 0.0689 - n_outputs1_loss: 0.0021 - val_loss: 0.0632 - val_n_outputs0_loss: 0.0626 - val_n_outputs1_loss: 5.9058e-04
Epoch 8/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0693 - n_outputs0_loss: 0.0675 - n_outputs1_loss: 0.0018
Epoch 00008: val_loss did not improve from 0.06324
36/36 [==============================] - 2s 51ms/step - loss: 0.0690 - n_outputs0_loss: 0.0672 - n_outputs1_loss: 0.0018 - val_loss: 0.0644 - val_n_outputs0_loss: 0.0637 - val_n_outputs1_loss: 7.8261e-04
Epoch 9/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0693 - n_outputs0_loss: 0.0675 - n_outputs1_loss: 0.0018
Epoch 00009: val_loss did not improve from 0.06324
36/36 [==============================] - 2s 50ms/step - loss: 0.0690 - n_outputs0_loss: 0.0672 - n_outputs1_loss: 0.0018 - val_loss: 0.0653 - val_n_outputs0_loss: 0.0646 - val_n_outputs1_loss: 6.6657e-04
Epoch 10/100
34/36 [===========================&gt;..] - ETA: 0s - loss: 0.0651 - n_outputs0_loss: 0.0633 - n_outputs1_loss: 0.0018
Epoch 00010: val_loss did not improve from 0.06324
36/36 [==============================] - 2s 51ms/step - loss: 0.0657 - n_outputs0_loss: 0.0639 - n_outputs1_loss: 0.0018 - val_loss: 0.0699 - val_n_outputs0_loss: 0.0692 - val_n_outputs1_loss: 6.9729e-04
Epoch 11/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0630 - n_outputs0_loss: 0.0612 - n_outputs1_loss: 0.0018
Epoch 00011: val_loss did not improve from 0.06324
36/36 [==============================] - 2s 51ms/step - loss: 0.0630 - n_outputs0_loss: 0.0612 - n_outputs1_loss: 0.0018 - val_loss: 0.0633 - val_n_outputs0_loss: 0.0619 - val_n_outputs1_loss: 0.0014
Epoch 12/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0643 - n_outputs0_loss: 0.0626 - n_outputs1_loss: 0.0016
Epoch 00012: val_loss improved from 0.06324 to 0.06034, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 52ms/step - loss: 0.0644 - n_outputs0_loss: 0.0628 - n_outputs1_loss: 0.0017 - val_loss: 0.0603 - val_n_outputs0_loss: 0.0598 - val_n_outputs1_loss: 4.9525e-04
Epoch 13/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0611 - n_outputs0_loss: 0.0597 - n_outputs1_loss: 0.0014
Epoch 00013: val_loss improved from 0.06034 to 0.05636, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 52ms/step - loss: 0.0613 - n_outputs0_loss: 0.0599 - n_outputs1_loss: 0.0014 - val_loss: 0.0564 - val_n_outputs0_loss: 0.0557 - val_n_outputs1_loss: 6.3083e-04
Epoch 14/100
34/36 [===========================&gt;..] - ETA: 0s - loss: 0.0593 - n_outputs0_loss: 0.0580 - n_outputs1_loss: 0.0013
Epoch 00014: val_loss did not improve from 0.05636
36/36 [==============================] - 2s 51ms/step - loss: 0.0590 - n_outputs0_loss: 0.0577 - n_outputs1_loss: 0.0013 - val_loss: 0.0580 - val_n_outputs0_loss: 0.0575 - val_n_outputs1_loss: 5.4033e-04
Epoch 15/100
34/36 [===========================&gt;..] - ETA: 0s - loss: 0.0552 - n_outputs0_loss: 0.0540 - n_outputs1_loss: 0.0012
Epoch 00015: val_loss did not improve from 0.05636
36/36 [==============================] - 2s 51ms/step - loss: 0.0555 - n_outputs0_loss: 0.0542 - n_outputs1_loss: 0.0012 - val_loss: 0.0565 - val_n_outputs0_loss: 0.0559 - val_n_outputs1_loss: 5.6679e-04
Epoch 16/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0538 - n_outputs0_loss: 0.0527 - n_outputs1_loss: 0.0012
Epoch 00016: val_loss did not improve from 0.05636
36/36 [==============================] - 2s 50ms/step - loss: 0.0538 - n_outputs0_loss: 0.0526 - n_outputs1_loss: 0.0012 - val_loss: 0.0624 - val_n_outputs0_loss: 0.0619 - val_n_outputs1_loss: 5.0014e-04
Epoch 17/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0536 - n_outputs0_loss: 0.0525 - n_outputs1_loss: 0.0011
Epoch 00017: val_loss improved from 0.05636 to 0.05615, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 53ms/step - loss: 0.0534 - n_outputs0_loss: 0.0523 - n_outputs1_loss: 0.0011 - val_loss: 0.0561 - val_n_outputs0_loss: 0.0556 - val_n_outputs1_loss: 5.2067e-04
Epoch 18/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0502 - n_outputs0_loss: 0.0492 - n_outputs1_loss: 0.0011
Epoch 00018: val_loss improved from 0.05615 to 0.05594, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 52ms/step - loss: 0.0501 - n_outputs0_loss: 0.0490 - n_outputs1_loss: 0.0011 - val_loss: 0.0559 - val_n_outputs0_loss: 0.0555 - val_n_outputs1_loss: 4.4089e-04
Epoch 00018: early stopping
Training completed in 0:00:40.


----------- Best Eval Loss :0.055939 ---------
&lt;Figure size 640x480 with 1 Axes&gt;

显示训练结果的Loss曲线

In [26]:

import matplotlib.pyplot as plt

list_of_png = glob.glob('/content/mycar/models/*png')

latest_png = max(list_of_png, key=os.path.getctime)

image = cv2.imread(latest_png)

/content/mycar/models/mypilot.h5_loss_acc_0.055939.png

Out[26]:

&lt;matplotlib.image.AxesImage at 0x7f867c896be0&gt;

robocarstore/173807730625008222

将训练好的模型放回 Donkey Car.

一旦模型训练完毕,你可以在 mycar/models中找到训练好的模型
1.下载mypilot文件到你到PC或Mac

再从PC或Mac拷贝模型到你到pi

最后用Autopilot mode来驾驶Donkey car

Enjoy！

August 18, 2019
in DonkeyCar
15 min read

使用Colab来加速Donkey Car训练 (Tensorflow GPU)

colab

本notebook助你快速训练你的Donkey car或者自动驾驶小车模型。

参考并改进了@sachindroid8的notebook，先向前人致敬！

先说说使用Colab进行训练的优缺点

效率

使用Colab需要梯子，也要把数据传给google，上传时间因人而异，我上传大概花来2-3分钟，剩下就是执行代码和训练。第一执行代码需要搞懂什么原理，后面执行基本不用花什么时间，然而使用GPU训练时间，我有6k图片，训练30-40个Epoch左右提前结束，训练时间少于1分钟！每个Epoch只需1-2秒。

然后使用我的Macbook Pro训练，因为没有GPU的缘故，每个Epoc需要30-50秒，训练下来，接近40分钟到1小时才能完成。

便利性

如果没有梯子，当然Colab不是一个选择，我知道百度也提供一些有限度免费的服务器，以后可以再做测试，但有梯子的话，Google Colab是最好的选择。下面便开始教你怎样开始训练

载入我的Notebook

打开使用Google_Colab来Donkey_Car

安装 TensorFlow 1.14.0

TensorFlow 2.x和Donkey Car 3.x现在还有兼容性问题，暂时不推荐使用(2019.8.17)In [1]:

!pip install tensorflow-gpu==1.14.0

Collecting tensorflow-gpu==1.14.0
  Downloading https://files.pythonhosted.org/packages/76/04/43153bfdfcf6c9a4c38ecdb971ca9a75b9a791bb69a764d652c359aca504/tensorflow_gpu-1.14.0-cp36-cp36m-manylinux1_x86_64.whl (377.0MB)
     |████████████████████████████████| 377.0MB 86kB/s 
Requirement already satisfied: grpcio&gt;=1.8.6 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.15.0)
Requirement already satisfied: absl-py&gt;=0.7.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (0.7.1)
Requirement already satisfied: tensorflow-estimator&lt;1.15.0rc0,&gt;=1.14.0rc0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.14.0)
Requirement already satisfied: wrapt&gt;=1.11.1 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.11.2)
Requirement already satisfied: termcolor&gt;=1.1.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.1.0)
Requirement already satisfied: astor&gt;=0.6.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (0.8.0)
Requirement already satisfied: keras-applications&gt;=1.0.6 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.0.8)
Requirement already satisfied: tensorboard&lt;1.15.0,&gt;=1.14.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.14.0)
Requirement already satisfied: wheel&gt;=0.26 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (0.33.4)
Requirement already satisfied: protobuf&gt;=3.6.1 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (3.7.1)
Requirement already satisfied: six&gt;=1.10.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.12.0)
Requirement already satisfied: keras-preprocessing&gt;=1.0.5 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.1.0)
Requirement already satisfied: google-pasta&gt;=0.1.6 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (0.1.7)
Requirement already satisfied: gast&gt;=0.2.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (0.2.2)
Requirement already satisfied: numpy&lt;2.0,&gt;=1.14.5 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.16.4)
Requirement already satisfied: h5py in /usr/local/lib/python3.6/dist-packages (from keras-applications&gt;=1.0.6-&gt;tensorflow-gpu==1.14.0) (2.8.0)
Requirement already satisfied: markdown&gt;=2.6.8 in /usr/local/lib/python3.6/dist-packages (from tensorboard&lt;1.15.0,&gt;=1.14.0-&gt;tensorflow-gpu==1.14.0) (3.1.1)
Requirement already satisfied: werkzeug&gt;=0.11.15 in /usr/local/lib/python3.6/dist-packages (from tensorboard&lt;1.15.0,&gt;=1.14.0-&gt;tensorflow-gpu==1.14.0) (0.15.5)
Requirement already satisfied: setuptools&gt;=41.0.0 in /usr/local/lib/python3.6/dist-packages (from tensorboard&lt;1.15.0,&gt;=1.14.0-&gt;tensorflow-gpu==1.14.0) (41.0.1)
Installing collected packages: tensorflow-gpu
Successfully installed tensorflow-gpu-1.14.0

检查GPU是否有效

如果显示”Found GPU at: / device: GPU: 0“，则GPU可以正常使用

如果没有以上输出，需要检查Runtime （运行类型）是否选择了GPU硬件加速器In [2]:

device_name = tf.test.gpu_device_name()

if device_name != '/device:GPU:0':

raise SystemError('GPU device not found')

print('Found GPU at: {}'.format(device_name))

Found GPU at: /device:GPU:0

克隆Donkey respository

In [3]:

!git clone https://github.com/autorope/donkeycar.git donkey

Cloning into 'donkey'...
remote: Enumerating objects: 120, done.
remote: Counting objects: 100% (120/120), done.
remote: Compressing objects: 100% (76/76), done.
remote: Total 10558 (delta 65), reused 73 (delta 31), pack-reused 10438
Receiving objects: 100% (10558/10558), 58.74 MiB | 46.70 MiB/s, done.
Resolving deltas: 100% (6528/6528), done.

安装 Donkey car ¶

In [4]:

Obtaining file:///content/donkey
Requirement already satisfied: numpy in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (1.16.4)
Requirement already satisfied: pillow in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (4.3.0)
Requirement already satisfied: docopt in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (0.6.2)
Requirement already satisfied: tornado in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (4.5.3)
Requirement already satisfied: requests in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (2.21.0)
Requirement already satisfied: h5py in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (2.8.0)
Requirement already satisfied: moviepy in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (0.2.3.5)
Requirement already satisfied: pandas in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (0.24.2)
Requirement already satisfied: PrettyTable in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (0.7.2)
Collecting paho-mqtt (from donkeycar==3.1.0)
  Downloading https://files.pythonhosted.org/packages/25/63/db25e62979c2a716a74950c9ed658dce431b5cb01fde29eb6cba9489a904/paho-mqtt-1.4.0.tar.gz (88kB)
     |████████████████████████████████| 92kB 4.2MB/s 
Requirement already satisfied: olefile in /usr/local/lib/python3.6/dist-packages (from pillow-&gt;donkeycar==3.1.0) (0.46)
Requirement already satisfied: urllib3&lt;1.25,&gt;=1.21.1 in /usr/local/lib/python3.6/dist-packages (from requests-&gt;donkeycar==3.1.0) (1.24.3)
Requirement already satisfied: certifi&gt;=2017.4.17 in /usr/local/lib/python3.6/dist-packages (from requests-&gt;donkeycar==3.1.0) (2019.6.16)
Requirement already satisfied: chardet&lt;3.1.0,&gt;=3.0.2 in /usr/local/lib/python3.6/dist-packages (from requests-&gt;donkeycar==3.1.0) (3.0.4)
Requirement already satisfied: idna&lt;2.9,&gt;=2.5 in /usr/local/lib/python3.6/dist-packages (from requests-&gt;donkeycar==3.1.0) (2.8)
Requirement already satisfied: six in /usr/local/lib/python3.6/dist-packages (from h5py-&gt;donkeycar==3.1.0) (1.12.0)
Requirement already satisfied: decorator&lt;5.0,&gt;=4.0.2 in /usr/local/lib/python3.6/dist-packages (from moviepy-&gt;donkeycar==3.1.0) (4.4.0)
Requirement already satisfied: tqdm&lt;5.0,&gt;=4.11.2 in /usr/local/lib/python3.6/dist-packages (from moviepy-&gt;donkeycar==3.1.0) (4.28.1)
Requirement already satisfied: imageio&lt;3.0,&gt;=2.1.2 in /usr/local/lib/python3.6/dist-packages (from moviepy-&gt;donkeycar==3.1.0) (2.4.1)
Requirement already satisfied: pytz&gt;=2011k in /usr/local/lib/python3.6/dist-packages (from pandas-&gt;donkeycar==3.1.0) (2018.9)
Requirement already satisfied: python-dateutil&gt;=2.5.0 in /usr/local/lib/python3.6/dist-packages (from pandas-&gt;donkeycar==3.1.0) (2.5.3)
Building wheels for collected packages: paho-mqtt
  Building wheel for paho-mqtt (setup.py) ... done
  Created wheel for paho-mqtt: filename=paho_mqtt-1.4.0-cp36-none-any.whl size=48333 sha256=9a67d0c95fae2b9495c20980895d9569ef53859cbc22bb57fc2466d7a581af7b
  Stored in directory: /root/.cache/pip/wheels/82/e5/de/d90d0f397648a1b58ffeea1b5742ac8c77f71fd43b550fa5a5
Successfully built paho-mqtt
Installing collected packages: paho-mqtt, donkeycar
  Running setup.py develop for donkeycar
Successfully installed donkeycar paho-mqtt-1.4.0

创建项目

我使用了mycar作为项目名称，你可以改名，但改名后需要相应修改后面但代码In [5]:

!donkey createcar --path /content/mycar

using donkey v3.1.0 ...
Creating car folder: /content/mycar
making dir  /content/mycar
Creating data &amp; model folders.
making dir  /content/mycar/models
making dir  /content/mycar/data
making dir  /content/mycar/logs
Copying car application template: complete
Copying car config defaults. Adjust these before starting your car.
Copying train script. Adjust these before starting your car.
Copying my car config overrides
Donkey setup complete.

准备数据: 上传data.zip并解压

现在你需要把pi采集回来的data目录上需要训练的目录打包，保存成data.zip.

在pi的data目录上运行:

$ zip -r data.zip tub_3_19-08-17/

然后拷贝会电脑，准备上传到Colab

上传data.zip到Colab

运行下面代码，会出现一个上传按钮，点击上传刚才打包的data.zipIn [7]:

from google.colab import files

if(os.path.exists("/content/data.zip")):

os.remove("/content/data.zip")

if(os.path.exists("/content/mycar/data/data.zip")):

os.remove("/content/mycar/data/data.zip")

uploaded = files.upload()

WORK_FOLDER = "/content/mycar/data/"

if(os.path.exists(WORK_FOLDER) == False):

!mv /content/data.zip /content/mycar/data/

清理已经上传的文件

你需要确保content/mycar/data目录下有tub目录，目录里面有图片和对应的json文件

data.zip就不用保留了In [0]:

!rm /content/mycar/data/data.zip

训练模型

In [21]:

!python /content/mycar/manage.py train --model /content/mycar/models/mypilot.h5

using donkey v3.1.0 ...
loading config file: /content/mycar/config.py
loading personal config over-rides

config loaded
WARNING: Logging before flag parsing goes to stderr.
W0818 04:53:15.944086 140399927523200 deprecation_wrapper.py:119] From /content/donkey/donkeycar/parts/keras.py:18: The name tf.ConfigProto is deprecated. Please use tf.compat.v1.ConfigProto instead.

W0818 04:53:15.944333 140399927523200 deprecation_wrapper.py:119] From /content/donkey/donkeycar/parts/keras.py:18: The name tf.Session is deprecated. Please use tf.compat.v1.Session instead.

2019-08-18 04:53:15.954857: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
2019-08-18 04:53:15.959505: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcuda.so.1
2019-08-18 04:53:16.083690: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1005] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-08-18 04:53:16.084227: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x2455100 executing computations on platform CUDA. Devices:
2019-08-18 04:53:16.084262: I tensorflow/compiler/xla/service/service.cc:175]   StreamExecutor device (0): Tesla T4, Compute Capability 7.5
2019-08-18 04:53:16.086194: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 2300000000 Hz
2019-08-18 04:53:16.086460: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x65e6380 executing computations on platform Host. Devices:
2019-08-18 04:53:16.086508: I tensorflow/compiler/xla/service/service.cc:175]   StreamExecutor device (0): &lt;undefined&gt;, &lt;undefined&gt;
2019-08-18 04:53:16.086686: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1005] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-08-18 04:53:16.087039: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1640] Found device 0 with properties: 
name: Tesla T4 major: 7 minor: 5 memoryClockRate(GHz): 1.59
pciBusID: 0000:00:04.0
2019-08-18 04:53:16.087321: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcudart.so.10.0
2019-08-18 04:53:16.088548: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcublas.so.10.0
2019-08-18 04:53:16.089621: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcufft.so.10.0
2019-08-18 04:53:16.089935: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcurand.so.10.0
2019-08-18 04:53:16.091399: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcusolver.so.10.0
2019-08-18 04:53:16.092394: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcusparse.so.10.0
2019-08-18 04:53:16.095470: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcudnn.so.7
2019-08-18 04:53:16.095606: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1005] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-08-18 04:53:16.096009: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1005] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-08-18 04:53:16.096350: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1763] Adding visible gpu devices: 0
2019-08-18 04:53:16.096401: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcudart.so.10.0
2019-08-18 04:53:16.097166: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1181] Device interconnect StreamExecutor with strength 1 edge matrix:
2019-08-18 04:53:16.097190: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1187]      0 
2019-08-18 04:53:16.097201: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1200] 0:   N 
2019-08-18 04:53:16.097482: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1005] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-08-18 04:53:16.097878: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1005] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-08-18 04:53:16.098226: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1326] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 14089 MB memory) -&gt; physical GPU (device: 0, name: Tesla T4, pci bus id: 0000:00:04.0, compute capability: 7.5)
"get_model_by_type" model Type is: linear
W0818 04:53:16.444941 140399927523200 deprecation.py:506] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/init_ops.py:1251: calling VarianceScaling.__init__ (from tensorflow.python.ops.init_ops) with dtype is deprecated and will be removed in a future version.
Instructions for updating:
Call initializer instance with the dtype argument instead of passing it to the constructor
training with model type &lt;class 'donkeycar.parts.keras.KerasLinear'&gt;
Model: "model"
__________________________________________________________________________________________________
Layer (type)                    Output Shape         Param #     Connected to                     
==================================================================================================
img_in (InputLayer)             [(None, 120, 160, 3) 0                                            
__________________________________________________________________________________________________
conv2d_1 (Conv2D)               (None, 58, 78, 24)   1824        img_in[0][0]                     
__________________________________________________________________________________________________
dropout (Dropout)               (None, 58, 78, 24)   0           conv2d_1[0][0]                   
__________________________________________________________________________________________________
conv2d_2 (Conv2D)               (None, 27, 37, 32)   19232       dropout[0][0]                    
__________________________________________________________________________________________________
dropout_1 (Dropout)             (None, 27, 37, 32)   0           conv2d_2[0][0]                   
__________________________________________________________________________________________________
conv2d_3 (Conv2D)               (None, 12, 17, 64)   51264       dropout_1[0][0]                  
__________________________________________________________________________________________________
dropout_2 (Dropout)             (None, 12, 17, 64)   0           conv2d_3[0][0]                   
__________________________________________________________________________________________________
conv2d_4 (Conv2D)               (None, 10, 15, 64)   36928       dropout_2[0][0]                  
__________________________________________________________________________________________________
dropout_3 (Dropout)             (None, 10, 15, 64)   0           conv2d_4[0][0]                   
__________________________________________________________________________________________________
conv2d_5 (Conv2D)               (None, 8, 13, 64)    36928       dropout_3[0][0]                  
__________________________________________________________________________________________________
dropout_4 (Dropout)             (None, 8, 13, 64)    0           conv2d_5[0][0]                   
__________________________________________________________________________________________________
flattened (Flatten)             (None, 6656)         0           dropout_4[0][0]                  
__________________________________________________________________________________________________
dense (Dense)                   (None, 100)          665700      flattened[0][0]                  
__________________________________________________________________________________________________
dropout_5 (Dropout)             (None, 100)          0           dense[0][0]                      
__________________________________________________________________________________________________
dense_1 (Dense)                 (None, 50)           5050        dropout_5[0][0]                  
__________________________________________________________________________________________________
dropout_6 (Dropout)             (None, 50)           0           dense_1[0][0]                    
__________________________________________________________________________________________________
n_outputs0 (Dense)              (None, 1)            51          dropout_6[0][0]                  
__________________________________________________________________________________________________
n_outputs1 (Dense)              (None, 1)            51          dropout_6[0][0]                  
==================================================================================================
Total params: 817,028
Trainable params: 817,028
Non-trainable params: 0
__________________________________________________________________________________________________
None
found 0 pickles writing json records and images in tub /content/mycar/data/tub_3_19-08-17
/content/mycar/data/tub_3_19-08-17
collating 5799 records ...
train: 4639, val: 1160
total records: 5799
steps_per_epoch 36
Epoch 1/100
2019-08-18 04:53:19.663907: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcublas.so.10.0
2019-08-18 04:53:19.984999: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcudnn.so.7
35/36 [============================&gt;.] - ETA: 0s - loss: 0.1681 - n_outputs0_loss: 0.1555 - n_outputs1_loss: 0.0126
Epoch 00001: val_loss improved from inf to 0.14482, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 8s 213ms/step - loss: 0.1677 - n_outputs0_loss: 0.1553 - n_outputs1_loss: 0.0124 - val_loss: 0.1448 - val_n_outputs0_loss: 0.1409 - val_n_outputs1_loss: 0.0039
Epoch 2/100
34/36 [===========================&gt;..] - ETA: 0s - loss: 0.1278 - n_outputs0_loss: 0.1234 - n_outputs1_loss: 0.0044
Epoch 00002: val_loss improved from 0.14482 to 0.08814, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 53ms/step - loss: 0.1257 - n_outputs0_loss: 0.1214 - n_outputs1_loss: 0.0043 - val_loss: 0.0881 - val_n_outputs0_loss: 0.0870 - val_n_outputs1_loss: 0.0012
Epoch 3/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0943 - n_outputs0_loss: 0.0910 - n_outputs1_loss: 0.0033
Epoch 00003: val_loss improved from 0.08814 to 0.07490, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 52ms/step - loss: 0.0938 - n_outputs0_loss: 0.0905 - n_outputs1_loss: 0.0033 - val_loss: 0.0749 - val_n_outputs0_loss: 0.0738 - val_n_outputs1_loss: 0.0011
Epoch 4/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0840 - n_outputs0_loss: 0.0813 - n_outputs1_loss: 0.0027
Epoch 00004: val_loss improved from 0.07490 to 0.07108, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 51ms/step - loss: 0.0835 - n_outputs0_loss: 0.0808 - n_outputs1_loss: 0.0027 - val_loss: 0.0711 - val_n_outputs0_loss: 0.0702 - val_n_outputs1_loss: 8.9668e-04
Epoch 5/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0767 - n_outputs0_loss: 0.0741 - n_outputs1_loss: 0.0026
Epoch 00005: val_loss did not improve from 0.07108
36/36 [==============================] - 2s 50ms/step - loss: 0.0765 - n_outputs0_loss: 0.0739 - n_outputs1_loss: 0.0026 - val_loss: 0.0722 - val_n_outputs0_loss: 0.0717 - val_n_outputs1_loss: 5.4629e-04
Epoch 6/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0750 - n_outputs0_loss: 0.0727 - n_outputs1_loss: 0.0023
Epoch 00006: val_loss improved from 0.07108 to 0.06483, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 52ms/step - loss: 0.0749 - n_outputs0_loss: 0.0725 - n_outputs1_loss: 0.0024 - val_loss: 0.0648 - val_n_outputs0_loss: 0.0641 - val_n_outputs1_loss: 7.5336e-04
Epoch 7/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0711 - n_outputs0_loss: 0.0689 - n_outputs1_loss: 0.0022
Epoch 00007: val_loss improved from 0.06483 to 0.06324, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 52ms/step - loss: 0.0711 - n_outputs0_loss: 0.0689 - n_outputs1_loss: 0.0021 - val_loss: 0.0632 - val_n_outputs0_loss: 0.0626 - val_n_outputs1_loss: 5.9058e-04
Epoch 8/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0693 - n_outputs0_loss: 0.0675 - n_outputs1_loss: 0.0018
Epoch 00008: val_loss did not improve from 0.06324
36/36 [==============================] - 2s 51ms/step - loss: 0.0690 - n_outputs0_loss: 0.0672 - n_outputs1_loss: 0.0018 - val_loss: 0.0644 - val_n_outputs0_loss: 0.0637 - val_n_outputs1_loss: 7.8261e-04
Epoch 9/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0693 - n_outputs0_loss: 0.0675 - n_outputs1_loss: 0.0018
Epoch 00009: val_loss did not improve from 0.06324
36/36 [==============================] - 2s 50ms/step - loss: 0.0690 - n_outputs0_loss: 0.0672 - n_outputs1_loss: 0.0018 - val_loss: 0.0653 - val_n_outputs0_loss: 0.0646 - val_n_outputs1_loss: 6.6657e-04
Epoch 10/100
34/36 [===========================&gt;..] - ETA: 0s - loss: 0.0651 - n_outputs0_loss: 0.0633 - n_outputs1_loss: 0.0018
Epoch 00010: val_loss did not improve from 0.06324
36/36 [==============================] - 2s 51ms/step - loss: 0.0657 - n_outputs0_loss: 0.0639 - n_outputs1_loss: 0.0018 - val_loss: 0.0699 - val_n_outputs0_loss: 0.0692 - val_n_outputs1_loss: 6.9729e-04
Epoch 11/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0630 - n_outputs0_loss: 0.0612 - n_outputs1_loss: 0.0018
Epoch 00011: val_loss did not improve from 0.06324
36/36 [==============================] - 2s 51ms/step - loss: 0.0630 - n_outputs0_loss: 0.0612 - n_outputs1_loss: 0.0018 - val_loss: 0.0633 - val_n_outputs0_loss: 0.0619 - val_n_outputs1_loss: 0.0014
Epoch 12/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0643 - n_outputs0_loss: 0.0626 - n_outputs1_loss: 0.0016
Epoch 00012: val_loss improved from 0.06324 to 0.06034, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 52ms/step - loss: 0.0644 - n_outputs0_loss: 0.0628 - n_outputs1_loss: 0.0017 - val_loss: 0.0603 - val_n_outputs0_loss: 0.0598 - val_n_outputs1_loss: 4.9525e-04
Epoch 13/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0611 - n_outputs0_loss: 0.0597 - n_outputs1_loss: 0.0014
Epoch 00013: val_loss improved from 0.06034 to 0.05636, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 52ms/step - loss: 0.0613 - n_outputs0_loss: 0.0599 - n_outputs1_loss: 0.0014 - val_loss: 0.0564 - val_n_outputs0_loss: 0.0557 - val_n_outputs1_loss: 6.3083e-04
Epoch 14/100
34/36 [===========================&gt;..] - ETA: 0s - loss: 0.0593 - n_outputs0_loss: 0.0580 - n_outputs1_loss: 0.0013
Epoch 00014: val_loss did not improve from 0.05636
36/36 [==============================] - 2s 51ms/step - loss: 0.0590 - n_outputs0_loss: 0.0577 - n_outputs1_loss: 0.0013 - val_loss: 0.0580 - val_n_outputs0_loss: 0.0575 - val_n_outputs1_loss: 5.4033e-04
Epoch 15/100
34/36 [===========================&gt;..] - ETA: 0s - loss: 0.0552 - n_outputs0_loss: 0.0540 - n_outputs1_loss: 0.0012
Epoch 00015: val_loss did not improve from 0.05636
36/36 [==============================] - 2s 51ms/step - loss: 0.0555 - n_outputs0_loss: 0.0542 - n_outputs1_loss: 0.0012 - val_loss: 0.0565 - val_n_outputs0_loss: 0.0559 - val_n_outputs1_loss: 5.6679e-04
Epoch 16/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0538 - n_outputs0_loss: 0.0527 - n_outputs1_loss: 0.0012
Epoch 00016: val_loss did not improve from 0.05636
36/36 [==============================] - 2s 50ms/step - loss: 0.0538 - n_outputs0_loss: 0.0526 - n_outputs1_loss: 0.0012 - val_loss: 0.0624 - val_n_outputs0_loss: 0.0619 - val_n_outputs1_loss: 5.0014e-04
Epoch 17/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0536 - n_outputs0_loss: 0.0525 - n_outputs1_loss: 0.0011
Epoch 00017: val_loss improved from 0.05636 to 0.05615, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 53ms/step - loss: 0.0534 - n_outputs0_loss: 0.0523 - n_outputs1_loss: 0.0011 - val_loss: 0.0561 - val_n_outputs0_loss: 0.0556 - val_n_outputs1_loss: 5.2067e-04
Epoch 18/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0502 - n_outputs0_loss: 0.0492 - n_outputs1_loss: 0.0011
Epoch 00018: val_loss improved from 0.05615 to 0.05594, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 52ms/step - loss: 0.0501 - n_outputs0_loss: 0.0490 - n_outputs1_loss: 0.0011 - val_loss: 0.0559 - val_n_outputs0_loss: 0.0555 - val_n_outputs1_loss: 4.4089e-04
Epoch 00018: early stopping
Training completed in 0:00:40.


----------- Best Eval Loss :0.055939 ---------
&lt;Figure size 640x480 with 1 Axes&gt;

显示训练结果的Loss曲线

In [26]:

import matplotlib.pyplot as plt

list_of_png = glob.glob('/content/mycar/models/*png')

latest_png = max(list_of_png, key=os.path.getctime)

image = cv2.imread(latest_png)

/content/mycar/models/mypilot.h5_loss_acc_0.055939.png

Out[26]:

&lt;matplotlib.image.AxesImage at 0x7f867c896be0&gt;

robocarstore/173807730625008222

将训练好的模型放回 Donkey Car.

一旦模型训练完毕,你可以在 mycar/models中找到训练好的模型
1.下载mypilot文件到你到PC或Mac

再从PC或Mac拷贝模型到你到pi

最后用Autopilot mode来驾驶Donkey car

Enjoy！

August 18, 2019
in DonkeyCar
16 min read

Use Colab to Accelerate Donkey Car Training (Tensorflow GPU)

colab

This notebook helps you quickly train your Donkey car or autonomous car model.

Reference and improved the notebook of @sachindroid8, thanks to the previous generation!

The Pros and Cons of Using Colab for Training

Efficiency

Using Colab requires a VPN, and also needs to upload data to Google, the upload time varies from person to person, I uploaded it for about 2-3 minutes, the rest is executing the code and training. The first execution of the code needs to understand what the principle is, and the subsequent execution basically does not take much time, but the training time using GPU, I have 6k images, training about 30-40 epochs early, the training time is less than 1 minute! Each Epoch only needs 1-2 seconds.

Then using my Macbook Pro training, because of the lack of GPU, each Epoch needs 30-50 seconds, training down, it takes about 40 minutes to 1 hour to complete.

Convenience

If you don't have a VPN, of course, Colab is not a choice, I know Baidu also provides some limited free servers, which can be tested later, but with a VPN, Google Colab is the best choice. Below is how to start training

Load My Notebook

Open the Using Google_Colab to Donkey_Car

Install TensorFlow 1.14.0

TensorFlow 2.x and Donkey Car 3.x still have compatibility issues, so it is not recommended to use them temporarily (2019.8.17)In [1]:

!pip install tensorflow-gpu==1.14.0

Collecting tensorflow-gpu==1.14.0
  Downloading https://files.pythonhosted.org/packages/76/04/43153bfdfcf6c9a4c38ecdb971ca9a75b9a791bb69a764d652c359aca504/tensorflow_gpu-1.14.0-cp36-cp36m-manylinux1_x86_64.whl (377.0MB)
     |████████████████████████████████| 377.0MB 86kB/s 
Requirement already satisfied: grpcio&gt;=1.8.6 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.15.0)
Requirement already satisfied: absl-py&gt;=0.7.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (0.7.1)
Requirement already satisfied: tensorflow-estimator&lt;1.15.0rc0,&gt;=1.14.0rc0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.14.0)
Requirement already satisfied: wrapt&gt;=1.11.1 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.11.2)
Requirement already satisfied: termcolor&gt;=1.1.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.1.0)
Requirement already satisfied: astor&gt;=0.6.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (0.8.0)
Requirement already satisfied: keras-applications&gt;=1.0.6 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.0.8)
Requirement already satisfied: tensorboard&lt;1.15.0,&gt;=1.14.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.14.0)
Requirement already satisfied: wheel&gt;=0.26 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (0.33.4)
Requirement already satisfied: protobuf&gt;=3.6.1 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (3.7.1)
Requirement already satisfied: six&gt;=1.10.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.12.0)
Requirement already satisfied: keras-preprocessing&gt;=1.0.5 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.1.0)
Requirement already satisfied: google-pasta&gt;=0.1.6 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (0.1.7)
Requirement already satisfied: gast&gt;=0.2.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (0.2.2)
Requirement already satisfied: numpy&lt;2.0,&gt;=1.14.5 in /usr/local/lib/python3.6/dist-packages (from tensorflow-gpu==1.14.0) (1.16.4)
Requirement already satisfied: h5py in /usr/local/lib/python3.6/dist-packages (from keras-applications&gt;=1.0.6-&gt;tensorflow-gpu==1.14.0) (2.8.0)
Requirement already satisfied: markdown&gt;=2.6.8 in /usr/local/lib/python3.6/dist-packages (from tensorboard&lt;1.15.0,&gt;=1.14.0-&gt;tensorflow-gpu==1.14.0) (3.1.1)
Requirement already satisfied: werkzeug&gt;=0.11.15 in /usr/local/lib/python3.6/dist-packages (from tensorboard&lt;1.15.0,&gt;=1.14.0-&gt;tensorflow-gpu==1.14.0) (0.15.5)
Requirement already satisfied: setuptools&gt;=41.0.0 in /usr/local/lib/python3.6/dist-packages (from tensorboard&lt;1.15.0,&gt;=1.14.0-&gt;tensorflow-gpu==1.14.0) (41.0.1)
Installing collected packages: tensorflow-gpu
Successfully installed tensorflow-gpu-1.14.0

Check if GPU is working

If you see “Found GPU at: / device: GPU: 0”, then the GPU is working normally

If you don't see the above output, you need to check if the Runtime (Run Type) has selected the GPU hardware acceleratorIn [2]:

device_name = tf.test.gpu_device_name()

if device_name != '/device:GPU:0':

raise SystemError('GPU device not found')

print('Found GPU at: {}'.format(device_name))

Found GPU at: /device:GPU:0

克隆Donkey respository

In [3]:

!git clone https://github.com/autorope/donkeycar.git donkey

Cloning into 'donkey'...
remote: Enumerating objects: 120, done.
remote: Counting objects: 100% (120/120), done.
remote: Compressing objects: 100% (76/76), done.
remote: Total 10558 (delta 65), reused 73 (delta 31), pack-reused 10438
Receiving objects: 100% (10558/10558), 58.74 MiB | 46.70 MiB/s, done.
Resolving deltas: 100% (6528/6528), done.

Install Donkey car ¶

In [4]:

Obtaining file:///content/donkey
Requirement already satisfied: numpy in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (1.16.4)
Requirement already satisfied: pillow in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (4.3.0)
Requirement already satisfied: docopt in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (0.6.2)
Requirement already satisfied: tornado in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (4.5.3)
Requirement already satisfied: requests in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (2.21.0)
Requirement already satisfied: h5py in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (2.8.0)
Requirement already satisfied: moviepy in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (0.2.3.5)
Requirement already satisfied: pandas in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (0.24.2)
Requirement already satisfied: PrettyTable in /usr/local/lib/python3.6/dist-packages (from donkeycar==3.1.0) (0.7.2)
Collecting paho-mqtt (from donkeycar==3.1.0)
  Downloading https://files.pythonhosted.org/packages/25/63/db25e62979c2a716a74950c9ed658dce431b5cb01fde29eb6cba9489a904/paho-mqtt-1.4.0.tar.gz (88kB)
     |████████████████████████████████| 92kB 4.2MB/s 
Requirement already satisfied: olefile in /usr/local/lib/python3.6/dist-packages (from pillow-&gt;donkeycar==3.1.0) (0.46)
Requirement already satisfied: urllib3&lt;1.25,&gt;=1.21.1 in /usr/local/lib/python3.6/dist-packages (from requests-&gt;donkeycar==3.1.0) (1.24.3)
Requirement already satisfied: certifi&gt;=2017.4.17 in /usr/local/lib/python3.6/dist-packages (from requests-&gt;donkeycar==3.1.0) (2019.6.16)
Requirement already satisfied: chardet&lt;3.1.0,&gt;=3.0.2 in /usr/local/lib/python3.6/dist-packages (from requests-&gt;donkeycar==3.1.0) (3.0.4)
Requirement already satisfied: idna&lt;2.9,&gt;=2.5 in /usr/local/lib/python3.6/dist-packages (from requests-&gt;donkeycar==3.1.0) (2.8)
Requirement already satisfied: six in /usr/local/lib/python3.6/dist-packages (from h5py-&gt;donkeycar==3.1.0) (1.12.0)
Requirement already satisfied: decorator&lt;5.0,&gt;=4.0.2 in /usr/local/lib/python3.6/dist-packages (from moviepy-&gt;donkeycar==3.1.0) (4.4.0)
Requirement already satisfied: tqdm&lt;5.0,&gt;=4.11.2 in /usr/local/lib/python3.6/dist-packages (from moviepy-&gt;donkeycar==3.1.0) (4.28.1)
Requirement already satisfied: imageio&lt;3.0,&gt;=2.1.2 in /usr/local/lib/python3.6/dist-packages (from moviepy-&gt;donkeycar==3.1.0) (2.4.1)
Requirement already satisfied: pytz&gt;=2011k in /usr/local/lib/python3.6/dist-packages (from pandas-&gt;donkeycar==3.1.0) (2018.9)
Requirement already satisfied: python-dateutil&gt;=2.5.0 in /usr/local/lib/python3.6/dist-packages (from pandas-&gt;donkeycar==3.1.0) (2.5.3)
Building wheels for collected packages: paho-mqtt
  Building wheel for paho-mqtt (setup.py) ... done
  Created wheel for paho-mqtt: filename=paho_mqtt-1.4.0-cp36-none-any.whl size=48333 sha256=9a67d0c95fae2b9495c20980895d9569ef53859cbc22bb57fc2466d7a581af7b
  Stored in directory: /root/.cache/pip/wheels/82/e5/de/d90d0f397648a1b58ffeea1b5742ac8c77f71fd43b550fa5a5
Successfully built paho-mqtt
Installing collected packages: paho-mqtt, donkeycar
  Running setup.py develop for donkeycar
Successfully installed donkeycar paho-mqtt-1.4.0

Create a project

I used mycar as the project name, you can change it, but you need to modify the code after renamingIn [5]:

!donkey createcar --path /content/mycar

using donkey v3.1.0 ...
Creating car folder: /content/mycar
making dir  /content/mycar
Creating data &amp; model folders.
making dir  /content/mycar/models
making dir  /content/mycar/data
making dir  /content/mycar/logs
Copying car application template: complete
Copying car config defaults. Adjust these before starting your car.
Copying train script. Adjust these before starting your car.
Copying my car config overrides
Donkey setup complete.

Prepare data: Upload data.zip and unzip

Now you need to package the data directory that you collected from pi, save it as data.zip.

Run the following command on the pi data directory:

$ zip -r data.zip tub_3_19-08-17/

Then copy it back to your computer, ready to upload to Colab

Upload data.zip to Colab

Run the following code, you will see an upload button, click to upload the data.zip you just packagedIn [7]:

from google.colab import files

if(os.path.exists("/content/data.zip")):

os.remove("/content/data.zip")

if(os.path.exists("/content/mycar/data/data.zip")):

os.remove("/content/mycar/data/data.zip")

uploaded = files.upload()

WORK_FOLDER = "/content/mycar/data/"

if(os.path.exists(WORK_FOLDER) == False):

!mv /content/data.zip /content/mycar/data/

Clean up the uploaded files

You need to ensure that the content/mycar/data directory has a tub directory, and the directory contains images and corresponding json files

data.zip is not neededIn [0]:

!rm /content/mycar/data/data.zip

Train the model

In [21]:

!python /content/mycar/manage.py train --model /content/mycar/models/mypilot.h5

using donkey v3.1.0 ...
loading config file: /content/mycar/config.py
loading personal config over-rides

config loaded
WARNING: Logging before flag parsing goes to stderr.
W0818 04:53:15.944086 140399927523200 deprecation_wrapper.py:119] From /content/donkey/donkeycar/parts/keras.py:18: The name tf.ConfigProto is deprecated. Please use tf.compat.v1.ConfigProto instead.

W0818 04:53:15.944333 140399927523200 deprecation_wrapper.py:119] From /content/donkey/donkeycar/parts/keras.py:18: The name tf.Session is deprecated. Please use tf.compat.v1.Session instead.

2019-08-18 04:53:15.954857: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
2019-08-18 04:53:15.959505: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcuda.so.1
2019-08-18 04:53:16.083690: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1005] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-08-18 04:53:16.084227: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x2455100 executing computations on platform CUDA. Devices:
2019-08-18 04:53:16.084262: I tensorflow/compiler/xla/service/service.cc:175]   StreamExecutor device (0): Tesla T4, Compute Capability 7.5
2019-08-18 04:53:16.086194: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 2300000000 Hz
2019-08-18 04:53:16.086460: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x65e6380 executing computations on platform Host. Devices:
2019-08-18 04:53:16.086508: I tensorflow/compiler/xla/service/service.cc:175]   StreamExecutor device (0): &lt;undefined&gt;, &lt;undefined&gt;
2019-08-18 04:53:16.086686: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1005] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-08-18 04:53:16.087039: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1640] Found device 0 with properties: 
name: Tesla T4 major: 7 minor: 5 memoryClockRate(GHz): 1.59
pciBusID: 0000:00:04.0
2019-08-18 04:53:16.087321: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcudart.so.10.0
2019-08-18 04:53:16.088548: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcublas.so.10.0
2019-08-18 04:53:16.089621: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcufft.so.10.0
2019-08-18 04:53:16.089935: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcurand.so.10.0
2019-08-18 04:53:16.091399: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcusolver.so.10.0
2019-08-18 04:53:16.092394: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcusparse.so.10.0
2019-08-18 04:53:16.095470: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcudnn.so.7
2019-08-18 04:53:16.095606: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1005] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-08-18 04:53:16.096009: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1005] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-08-18 04:53:16.096350: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1763] Adding visible gpu devices: 0
2019-08-18 04:53:16.096401: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcudart.so.10.0
2019-08-18 04:53:16.097166: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1181] Device interconnect StreamExecutor with strength 1 edge matrix:
2019-08-18 04:53:16.097190: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1187]      0 
2019-08-18 04:53:16.097201: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1200] 0:   N 
2019-08-18 04:53:16.097482: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1005] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-08-18 04:53:16.097878: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1005] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-08-18 04:53:16.098226: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1326] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 14089 MB memory) -&gt; physical GPU (device: 0, name: Tesla T4, pci bus id: 0000:00:04.0, compute capability: 7.5)
"get_model_by_type" model Type is: linear
W0818 04:53:16.444941 140399927523200 deprecation.py:506] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/init_ops.py:1251: calling VarianceScaling.__init__ (from tensorflow.python.ops.init_ops) with dtype is deprecated and will be removed in a future version.
Instructions for updating:
Call initializer instance with the dtype argument instead of passing it to the constructor
training with model type &lt;class 'donkeycar.parts.keras.KerasLinear'&gt;
Model: "model"
__________________________________________________________________________________________________
Layer (type)                    Output Shape         Param #     Connected to                     
==================================================================================================
img_in (InputLayer)             [(None, 120, 160, 3) 0                                            
__________________________________________________________________________________________________
conv2d_1 (Conv2D)               (None, 58, 78, 24)   1824        img_in[0][0]                     
__________________________________________________________________________________________________
dropout (Dropout)               (None, 58, 78, 24)   0           conv2d_1[0][0]                   
__________________________________________________________________________________________________
conv2d_2 (Conv2D)               (None, 27, 37, 32)   19232       dropout[0][0]                    
__________________________________________________________________________________________________
dropout_1 (Dropout)             (None, 27, 37, 32)   0           conv2d_2[0][0]                   
__________________________________________________________________________________________________
conv2d_3 (Conv2D)               (None, 12, 17, 64)   51264       dropout_1[0][0]                  
__________________________________________________________________________________________________
dropout_2 (Dropout)             (None, 12, 17, 64)   0           conv2d_3[0][0]                   
__________________________________________________________________________________________________
conv2d_4 (Conv2D)               (None, 10, 15, 64)   36928       dropout_2[0][0]                  
__________________________________________________________________________________________________
dropout_3 (Dropout)             (None, 10, 15, 64)   0           conv2d_4[0][0]                   
__________________________________________________________________________________________________
conv2d_5 (Conv2D)               (None, 8, 13, 64)    36928       dropout_3[0][0]                  
__________________________________________________________________________________________________
dropout_4 (Dropout)             (None, 8, 13, 64)    0           conv2d_5[0][0]                   
__________________________________________________________________________________________________
flattened (Flatten)             (None, 6656)         0           dropout_4[0][0]                  
__________________________________________________________________________________________________
dense (Dense)                   (None, 100)          665700      flattened[0][0]                  
__________________________________________________________________________________________________
dropout_5 (Dropout)             (None, 100)          0           dense[0][0]                      
__________________________________________________________________________________________________
dense_1 (Dense)                 (None, 50)           5050        dropout_5[0][0]                  
__________________________________________________________________________________________________
dropout_6 (Dropout)             (None, 50)           0           dense_1[0][0]                    
__________________________________________________________________________________________________
n_outputs0 (Dense)              (None, 1)            51          dropout_6[0][0]                  
__________________________________________________________________________________________________
n_outputs1 (Dense)              (None, 1)            51          dropout_6[0][0]                  
==================================================================================================
Total params: 817,028
Trainable params: 817,028
Non-trainable params: 0
__________________________________________________________________________________________________
None
found 0 pickles writing json records and images in tub /content/mycar/data/tub_3_19-08-17
/content/mycar/data/tub_3_19-08-17
collating 5799 records ...
train: 4639, val: 1160
total records: 5799
steps_per_epoch 36
Epoch 1/100
2019-08-18 04:53:19.663907: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcublas.so.10.0
2019-08-18 04:53:19.984999: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcudnn.so.7
35/36 [============================&gt;.] - ETA: 0s - loss: 0.1681 - n_outputs0_loss: 0.1555 - n_outputs1_loss: 0.0126
Epoch 00001: val_loss improved from inf to 0.14482, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 8s 213ms/step - loss: 0.1677 - n_outputs0_loss: 0.1553 - n_outputs1_loss: 0.0124 - val_loss: 0.1448 - val_n_outputs0_loss: 0.1409 - val_n_outputs1_loss: 0.0039
Epoch 2/100
34/36 [===========================&gt;..] - ETA: 0s - loss: 0.1278 - n_outputs0_loss: 0.1234 - n_outputs1_loss: 0.0044
Epoch 00002: val_loss improved from 0.14482 to 0.08814, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 53ms/step - loss: 0.1257 - n_outputs0_loss: 0.1214 - n_outputs1_loss: 0.0043 - val_loss: 0.0881 - val_n_outputs0_loss: 0.0870 - val_n_outputs1_loss: 0.0012
Epoch 3/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0943 - n_outputs0_loss: 0.0910 - n_outputs1_loss: 0.0033
Epoch 00003: val_loss improved from 0.08814 to 0.07490, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 52ms/step - loss: 0.0938 - n_outputs0_loss: 0.0905 - n_outputs1_loss: 0.0033 - val_loss: 0.0749 - val_n_outputs0_loss: 0.0738 - val_n_outputs1_loss: 0.0011
Epoch 4/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0840 - n_outputs0_loss: 0.0813 - n_outputs1_loss: 0.0027
Epoch 00004: val_loss improved from 0.07490 to 0.07108, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 51ms/step - loss: 0.0835 - n_outputs0_loss: 0.0808 - n_outputs1_loss: 0.0027 - val_loss: 0.0711 - val_n_outputs0_loss: 0.0702 - val_n_outputs1_loss: 8.9668e-04
Epoch 5/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0767 - n_outputs0_loss: 0.0741 - n_outputs1_loss: 0.0026
Epoch 00005: val_loss did not improve from 0.07108
36/36 [==============================] - 2s 50ms/step - loss: 0.0765 - n_outputs0_loss: 0.0739 - n_outputs1_loss: 0.0026 - val_loss: 0.0722 - val_n_outputs0_loss: 0.0717 - val_n_outputs1_loss: 5.4629e-04
Epoch 6/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0750 - n_outputs0_loss: 0.0727 - n_outputs1_loss: 0.0023
Epoch 00006: val_loss improved from 0.07108 to 0.06483, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 52ms/step - loss: 0.0749 - n_outputs0_loss: 0.0725 - n_outputs1_loss: 0.0024 - val_loss: 0.0648 - val_n_outputs0_loss: 0.0641 - val_n_outputs1_loss: 7.5336e-04
Epoch 7/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0711 - n_outputs0_loss: 0.0689 - n_outputs1_loss: 0.0022
Epoch 00007: val_loss improved from 0.06483 to 0.06324, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 52ms/step - loss: 0.0711 - n_outputs0_loss: 0.0689 - n_outputs1_loss: 0.0021 - val_loss: 0.0632 - val_n_outputs0_loss: 0.0626 - val_n_outputs1_loss: 5.9058e-04
Epoch 8/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0693 - n_outputs0_loss: 0.0675 - n_outputs1_loss: 0.0018
Epoch 00008: val_loss did not improve from 0.06324
36/36 [==============================] - 2s 51ms/step - loss: 0.0690 - n_outputs0_loss: 0.0672 - n_outputs1_loss: 0.0018 - val_loss: 0.0644 - val_n_outputs0_loss: 0.0637 - val_n_outputs1_loss: 7.8261e-04
Epoch 9/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0693 - n_outputs0_loss: 0.0675 - n_outputs1_loss: 0.0018
Epoch 00009: val_loss did not improve from 0.06324
36/36 [==============================] - 2s 50ms/step - loss: 0.0690 - n_outputs0_loss: 0.0672 - n_outputs1_loss: 0.0018 - val_loss: 0.0653 - val_n_outputs0_loss: 0.0646 - val_n_outputs1_loss: 6.6657e-04
Epoch 10/100
34/36 [===========================&gt;..] - ETA: 0s - loss: 0.0651 - n_outputs0_loss: 0.0633 - n_outputs1_loss: 0.0018
Epoch 00010: val_loss did not improve from 0.06324
36/36 [==============================] - 2s 51ms/step - loss: 0.0657 - n_outputs0_loss: 0.0639 - n_outputs1_loss: 0.0018 - val_loss: 0.0699 - val_n_outputs0_loss: 0.0692 - val_n_outputs1_loss: 6.9729e-04
Epoch 11/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0630 - n_outputs0_loss: 0.0612 - n_outputs1_loss: 0.0018
Epoch 00011: val_loss did not improve from 0.06324
36/36 [==============================] - 2s 51ms/step - loss: 0.0630 - n_outputs0_loss: 0.0612 - n_outputs1_loss: 0.0018 - val_loss: 0.0633 - val_n_outputs0_loss: 0.0619 - val_n_outputs1_loss: 0.0014
Epoch 12/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0643 - n_outputs0_loss: 0.0626 - n_outputs1_loss: 0.0016
Epoch 00012: val_loss improved from 0.06324 to 0.06034, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 52ms/step - loss: 0.0644 - n_outputs0_loss: 0.0628 - n_outputs1_loss: 0.0017 - val_loss: 0.0603 - val_n_outputs0_loss: 0.0598 - val_n_outputs1_loss: 4.9525e-04
Epoch 13/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0611 - n_outputs0_loss: 0.0597 - n_outputs1_loss: 0.0014
Epoch 00013: val_loss improved from 0.06034 to 0.05636, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 52ms/step - loss: 0.0613 - n_outputs0_loss: 0.0599 - n_outputs1_loss: 0.0014 - val_loss: 0.0564 - val_n_outputs0_loss: 0.0557 - val_n_outputs1_loss: 6.3083e-04
Epoch 14/100
34/36 [===========================&gt;..] - ETA: 0s - loss: 0.0593 - n_outputs0_loss: 0.0580 - n_outputs1_loss: 0.0013
Epoch 00014: val_loss did not improve from 0.05636
36/36 [==============================] - 2s 51ms/step - loss: 0.0590 - n_outputs0_loss: 0.0577 - n_outputs1_loss: 0.0013 - val_loss: 0.0580 - val_n_outputs0_loss: 0.0575 - val_n_outputs1_loss: 5.4033e-04
Epoch 15/100
34/36 [===========================&gt;..] - ETA: 0s - loss: 0.0552 - n_outputs0_loss: 0.0540 - n_outputs1_loss: 0.0012
Epoch 00015: val_loss did not improve from 0.05636
36/36 [==============================] - 2s 51ms/step - loss: 0.0555 - n_outputs0_loss: 0.0542 - n_outputs1_loss: 0.0012 - val_loss: 0.0565 - val_n_outputs0_loss: 0.0559 - val_n_outputs1_loss: 5.6679e-04
Epoch 16/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0538 - n_outputs0_loss: 0.0527 - n_outputs1_loss: 0.0012
Epoch 00016: val_loss did not improve from 0.05636
36/36 [==============================] - 2s 50ms/step - loss: 0.0538 - n_outputs0_loss: 0.0526 - n_outputs1_loss: 0.0012 - val_loss: 0.0624 - val_n_outputs0_loss: 0.0619 - val_n_outputs1_loss: 5.0014e-04
Epoch 17/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0536 - n_outputs0_loss: 0.0525 - n_outputs1_loss: 0.0011
Epoch 00017: val_loss improved from 0.05636 to 0.05615, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 53ms/step - loss: 0.0534 - n_outputs0_loss: 0.0523 - n_outputs1_loss: 0.0011 - val_loss: 0.0561 - val_n_outputs0_loss: 0.0556 - val_n_outputs1_loss: 5.2067e-04
Epoch 18/100
35/36 [============================&gt;.] - ETA: 0s - loss: 0.0502 - n_outputs0_loss: 0.0492 - n_outputs1_loss: 0.0011
Epoch 00018: val_loss improved from 0.05615 to 0.05594, saving model to /content/mycar/models/mypilot.h5
36/36 [==============================] - 2s 52ms/step - loss: 0.0501 - n_outputs0_loss: 0.0490 - n_outputs1_loss: 0.0011 - val_loss: 0.0559 - val_n_outputs0_loss: 0.0555 - val_n_outputs1_loss: 4.4089e-04
Epoch 00018: early stopping
Training completed in 0:00:40.


----------- Best Eval Loss :0.055939 ---------
&lt;Figure size 640x480 with 1 Axes&gt;

Display the Loss curve of the training result

In [26]:

import matplotlib.pyplot as plt

list_of_png = glob.glob('/content/mycar/models/*png')

latest_png = max(list_of_png, key=os.path.getctime)

image = cv2.imread(latest_png)

/content/mycar/models/mypilot.h5_loss_acc_0.055939.png

Out[26]:

&lt;matplotlib.image.AxesImage at 0x7f867c896be0&gt;

robocarstore/173807730625008222

Put the trained model back into Donkey Car.

Once the model is trained, you can find the trained model in the mycar/models directory
1. Download the mypilot file to your PC or Mac

Copy the model from your PC or Mac to your pi

Finally, use the Autopilot mode to drive Donkey car

Enjoy!

May 12, 2019
in Jetbot
2 min read

Play with JetBot Autonomous Driving （1）Prepare the DIY BOM list

robocarstore/173807773325122723

JetBot is an easy-to-use machine learning autonomous car, I feel it's easier to get started than Donkey Car.
This article lists all the electronic components, 3D printed parts, and hardware that I needed when assembling JetBot.
To avoid everyone taking detours, I will also list the detailed model parameters.

First, let's take a look at the "零件全家福" to get an overview:

robocarstore/173807775925330624

Here is the list of all parts and model parameters:

robocarstore/173807779624892425

In the installation process, I also used some tools, such as: M3 wrench, M2 wrench, M3 hex nut wrench, cross screwdriver, flat screwdriver, high temperature tape, soldering iron (some parts need to be soldered).

以上，就是我组装的这台JetBot用到的所有零件和工具。

Of course, you can also refer to the detailed Nvidia official Git part list, which proposes some other part solutions:
https://github.com/NVIDIA-AI-IOT/jetbot/wiki/bill-of-materials

Next, I will introduce how to install JetBot step by step.

to be continued…

May 12, 2019
in Jetbot
8 min read

Play with JetBot Autonomous Driving (2) Assemble Jetbot Car

robocarstore/173807852624587444

This article will detail the hardware installation process of JetBot, and provide a full installation video.

Due to the main camera malfunction during filming, the video content is unavailable, so only the material from the secondary camera is used, and the quality is somewhat unsatisfactory. Please forgive me.

You can also refer to the installation process of the official Git to complement the shortcomings (some pictures in this article are also from here):
https://github.com/NVIDIA-AI-IOT/jetbot/wiki/hardware-setup

Required Tools

M3 wrench, M2 wrench, M3 hex nut wrench, cross screwdriver, flat screwdriver, insulating tape, high temperature tape, 3M double-sided tape, soldering iron (some parts need to be soldered).

Start Assembling

Step – 1 Install Wireless Network Card

1，Remove the chip module (the entire module with the heatsink) from the Jetson Nano

First, unscrew the 2 screws, then pull the side lock (侧锁) apart, the chip module will automatically pop up at an angle, slide the chip module out along this angle. robocarstore/173807821424761926 robocarstore/173807825424765927 robocarstore/173807827724901728 robocarstore/173807828824814429

2，Install WiFi Module AC8265

Connect the Intel WiFi module AC8265 to the antenna, wrap it with high temperature tape to reinforce it, avoid falling off, and cut the high temperature tape on the groove. Screw the screws of the Jetson Nano base, plug the WiFi module into the WiFi slot, and screw the screws back in to fix the AC8265 WiFi module. robocarstore/173807831124775830 robocarstore/173807831924893431 robocarstore/173807832725044032 robocarstore/173807836024830233

3，Install the chip module back to the Jetson Nano base, and fix the antenna

Plug the chip module back into the slot, press flat, the side lock (侧锁) will automatically snap back in place, screw the 2 screws back in to fix the chip module, and then use high temperature tape to fix the antenna to the heatsink. robocarstore/173807836924645634

robocarstore/173807838324670435

robocarstore/173807839225296536

Step – 2 Connect Power Line to Motor Driver

4，Solder the Motor Driver Module

Here you need to solder, when you buy this motor driver module, some accessories will be included, you only need to solder the interface shown in Figure 1, and the final processing is shown in Figure 4.

robocarstore/173807841224729437

robocarstore/173807842224823038

robocarstore/173807843724698939

robocarstore/173807844433350340

5，Unwrap the Positive and Negative Poles of the MicroUSB Power Line

As shown below, peel off the positive and negative wires, and fix the other wires with insulating tape to prevent short circuits. Generally, it is red positive and black negative, if you are not sure, please refer to the MicroUSB data line connection diagram in the picture, use a multimeter to measure. You can also connect an LED light, use the characteristics of the diode, and turn on the power to test whether the positive and negative poles are correct.

robocarstore/173807845024671941

Note: The positive and negative poles must be clearly identified, if you get it wrong, it may burn your Jetson Nano, or even cause a power explosion.

Make sure to get it right!

Important things are said three times! If you have no experience, you must ask someone with experience for help!

6，Motor Driver Module Power Interface Connection to MicroUSB Data Line

First, look at the interface of this motor driver module, it is actually an integrated board of PCA9685 and TB6612.

robocarstore/173807849325676742

However, assembling JetBot only needs a few interfaces. This step only connects the MicroUSB data line as an external power line (3v3接_正极_、GND接_负极_).

robocarstore/173807851424798543

Step – 3 Install TT Motor

7，Install TT Motor

Prepare the 3D printed parts "chassis" (chassis.stl) and TT motor. Install the TT motor on the "chassis", fix it with M3 screws and M3 hex nuts, please be gentle, the 3D printed chassis is not as strong as you might think.

robocarstore/173807852624587444

robocarstore/173807854224561745

robocarstore/173807855324679046

Step – 4 Install Motor Driver Module

8，Install Motor Driver Module on the "chassis"

robocarstore/173807856624662147

First, use M2*6 self-tapping screws to install the motor driver module on the "chassis" (as shown in Figure 1).

Step – 5 TT Motor and Motor Driver Module Connection

9，TT Motor and Motor Driver Module Connection

According to the wiring method shown in the figure, connect the TT motor to the driver module (even if you connect it incorrectly, it is not a big deal, you can check whether the direction is correct in the subsequent application instance, and then adjust it).

robocarstore/173807858324621348

Note: In the installation video, the motor driver installation direction is different from the picture, please align with the direction shown in the article.

Step – 6 Install Wheels and Pre-connect Dupont Lines

10，Install Omnidirectional Wheels

Prepare the "caster shroud 60mm" (caster_shroud_60mm.stl), "caster base 60mm" (caster_base_60mm.stl), and polyoxymethylene ball (POM ball). Place the "caster shroud 60mm" -> polyoxymethylene ball -> "caster base 60mm" in the groove of the chassis, and then use "M2*8 self-tapping screws" to fix it. robocarstore/173807863324608149

robocarstore/173807865324939550

robocarstore/173807866124643751

robocarstore/173807867024781552

11，Motor Driver Module Pre-connect Dupont Lines

First, use Dupont lines to pre-connect the four pins on the motor driver module, which are: 3.3v, GND, SDA, SCL. Pre-connect it, wait for later use.

robocarstore/173807868126285253

12，Install the Left and Right Wheels

Please install the wheels carefully, so as not to exert too much force and crush the "chassis".

robocarstore/173807871924726254

Step – 7 Jetson Nano、OLED、电机驱动模块接线

13,Fix the Jetson Nano on the "chassis"

Use M2*6 self-tapping screws to fix the Jetson Nano on the "chassis". robocarstore/173807872924554455 robocarstore/173807875224607056

14，OLED Display Pre-wiring

Before connecting the wires, we need to use long double-row bent pins to weld the OLED Display (of course, you can also directly weld Dupont lines). robocarstore/173807876024663757 robocarstore/173807877324632258

15，OLED Display and Motor Driver Module Connection

First, understand the pin of the OLED Display.

robocarstore/173807879224744959

Then refer to the wiring diagram shown below, connect the OLED Display and the motor driver module.

robocarstore/173807879924605060

16，OLED and Jetson Nano Connection

Plug the OLED into the pins shown in the figure (the pins of the OLED Display are one-to-one corresponding), and install it according to Figure 2. robocarstore/173807881124578461 robocarstore/173807883024886862

Step – 8 Install Pi V2 Camera

17,Fix the Camera on the "chassis"

First, use M2*6 self-tapping screws to fix the Pi V2 camera on the "camera mount" (camera_mount.stl), then connect the video line to the Jetson Nano, and finally use M2*6 self-tapping screws to fix the "camera mount" on the "chassis". robocarstore/173807883824550163 robocarstore/173807885124829464 robocarstore/173807885824666465

Step – 9 Install Mobile Power

18,Install the Mobile Power in the "chassis"

This is the last step, and also the simplest step. Put the mobile power into the power slot, and fix it with tape or 3M double-sided tape.

robocarstore/173807886524635366

The next article will introduce how to burn the Jetson Nano system onto the MicroSD card.

to be continue……