Using Docker

Docker provides an easy way to running the repo code. We build our images using the NVidia Container Toolkit.

Please refer to that page to configure your Docker host. You can verify the proper behavior of NVidia Docker with

sudo docker run --rm --gpus all nvidia/cuda:11.7.0-base-ubuntu20.04 nvidia-smi

Getting kr_autonomous_flight Images

We build and push automatically our docker images to the KumarRobotics DockerHub.

To get the images, you just need to do:

docker pull kumarrobotics/autonomy:state_machine
docker pull kumarrobotics/autonomy:client
docker pull kumarrobotics/autonomy:control
docker pull kumarrobotics/autonomy:map_plan
docker pull kumarrobotics/autonomy:sim

How to build Docker images

Core Components

If you need to build a Docker image locally, you may do it as follows. Replace img_name with the name of the image you would like to build (base, calibration, client,...).

IMG=base; docker build -t "kumarrobotics/autonomy:$IMG" -f "autonomy_core/$IMG/Dockerfile" .

Gazebo Sim

IMG=sim; docker build -t "kumarrobotics/autonomy:$IMG" -f "autonomy_sim/Dockerfile" .