This repository belongs to Carnegie Mellon University, Masters of Science - Robotic Systems Development (MRSD) Team E - VADER
Team Members: Tom Gao, Abhishek Mathur, Rohit Satishkumar, Kshitij Bhat, Keerthi GV
Special thanks to the Robot Autonomy course (16-662) at CMU, the Docker infrastructure of which inspired this Docker solution.
First Revision: March 2025
This includes a dockerfile that builds the image for VADER's PC. More specifically, it:
- Shares the docker_ws folder to
/home/docker_wsinside the docker container, - Shares the host GPUs and devices,
- And sets up displays so that windows are launched in the host's display.
Make sure you have docker and the docker service is enabled.
Go into vader_docker and run build_docker.sh. Wait for the build to complete.
If you get the error where GPUs are not found during run_docker.sh, try running docker context use default.
Run run_docker.sh which should create a container and give you a terminal. The displays are set up so that the docker container can launch windows in the host's displays.
For additional terminals, run terminal_docker.sh in new terminals. These should attach to the same container.
Get two docker terminals, run roscore in one of them and run rviz in the other. The RViz window should pop up.
Make sure you can call nvidia-smi (if you have a nvidia driver on the host machine) and can use the GPU.
Verify that whatever in the docker_ws folder is mirrored successfully to the container.
To add packages or make modifications to the image, just add new lines to the Dockerfile. Make sure you use build_docker to rebuild the image so that it is up to date afterward.