DonkeyCar Installation using Docker Container

This docker image works on Jetson Xavier NX/AGX, it comes built with OpenCV 4.10 and Tensorflow 2.14 with CUDA 11.4 built with Python 3.11

First clone this git repository, change directory and make the run.sh script executable

git clone https://github.com/Triton-AI/donkeycontainer.git && 
cd donkeycontainer && 
sudo chmod +x run.sh

To Pull a base image without donkey install do:

docker pull ghcr.io/ucsd-ecemae-148/donkeycontainer:base

To start the container run the start script

./run.sh

Now lets install the dependencies for donkey

Lets add some swap space

git clone https://github.com/JetsonHacksNano/installSwapfile
cd installSwapfile
./installSwapfile.sh -s 8
reboot 

Lets install the donkeycar

git clone https://github.com/autorope/donkeycar
cd donkeycar
git checkout main
pip install -e .[nano] && pip install --force-reinstall pillow

Creating a Donkeycar from Template

donkey createcar --path /home/projects/mycars/deep_learning_car

That creates a car using the default deep learning template. You can also create a car that uses the gps path follow template;

donkey createcar --template=path_follow --path /home/projects/mycars/gps_path_follow_car

You can also create a car that uses the computer vision template

donkey createcar --template=cv_control --path /home/projects/mycars/cv_car

You can name your car folder anything you want.

After that change to the directory of the created car and make changes to myconfig.py as per your car's specifications

To Pull an image with donkeycar installed with all types of cars created under mycars

docker pull ghcr.io/ucsd-ecemae-148/donkeycontainer:donkey_built

To start the container run the start script

./run.sh

Once inside the container you will see a directory named mycars

cd mycars