This directory contains scripts to build the JetBot docker containers.

First, call the scripts/configure_jetson.sh script to configure the power mode and other parameters.

cd jetbot
./scripts/configure_jetson.sh

Next, source the docker/configure.sh script to configure various environment variables related to JetBot docker.

cd docker
source configure.sh

Depending on your Jetson L4T version, you may get a similar warning :

JETBOT_BASE_IMAGE not found for 32.6.1.  Please manually set the JETBOT_BASE_IMAGE environment variable. (ie: export JETBOT_BASE_IMAGE=...)

In the case of 32.6.1, the following container will work :

export JETBOT_BASE_IMAGE=nvcr.io/nvidia/l4t-pytorch:r32.6.1-pth1.9-py3

Refer to https://ngc.nvidia.com/catalog/containers/nvidia:l4t-pytorch for more details.

Finally, if you haven't already, set the default docker runtime to NVIDIA. This is needed to use CUDA related components with the containers.

./set_nvidia_runtime.sh

If needed, you can also set memory limits on the Jupyter container.

export JETBOT_JUPYTER_MEMORY=500m
export JETBOT_JUPYTER_MEMORY_SWAP=3G

If you want to build the containers from scratch, simply call :

./build.sh

Call the following to enable the JetBot docker containers

sudo systemctl enable docker   # enable docker daemon at boot
./enable.sh $HOME   # we'll use home directory as working directory, set this as you please.

Now you can go to https://<jetbot_ip>:8888 from a web browser and start programming JetBot! You can do this from any machine on your local network. The password to log in is jetbot.

Note: The directory you specify to ./enable.sh will be mounted as a volume in the jupyter container at the location /workspace. This means the work you in the /workspace folder inside container is saved. This is set to the root directory of Jupyter Lab. Please note, if you work outside of that directory it will be lost when the container shuts down.