🍁 MIT Autonomous Pathfinding for Lunar Exploration 🍁

Setup

This project uses uv for package and environment management. After cloning the repository and installing uv run the following commands:

• uv sync: Install dependencies
• uvx pre-commit install: Install pre-commit git hooks

FastSAM requires model weights that are too large to be stored on GitHub. Download the weights using uv run ./scripts/fastsam_checkpoint.py. This will place the weights in the correct location (resources/FastSAM-x.pt).

ORBSLAM requires a vocabulary file that is too large to be stored on GitHub raw. Extract the vocabulary file using uv run ./scripts/orbslam_vocab.py. This will extract the vocabulary file to the correct location (resources/ORBvoc.txt).

Optional Setup

If using Visual Studio Code the following extensions are recommended: python, pylance, ruff.

Running Tests

• uv run pytest: Run tests
• uv run pytest test/test_file.py: Run a specific test file

Running Agents

A python script is included for convenience to automatically start the simulator and run an agent. Due to the large file size of the lunar simulator it is not included in this repository, instead manually copy the contents of LunarSimulator to ./simulator. Alternatively a file path to the LunarSimulator directory can be provided as an optional argument. To start the simulator and evaluate an agent run uv run ./scripts/run_agent.py path_to_agent from the root of this repository. If specifying an alternate location for the simulator use uv run ./scripts/run_agent.py path_to_agent --sim="path_to_lunar_simulator".

Preparing Docker Container

Before building the agent into a docker container, remove any plots or dependencies on OpenCV. I am not sure what will happen if the agent tries to open a window.

Building the docker container requires Leaderboard and LunarSimulator be present in the project directory. To build a container with a specific agent, run uv run ./scripts/build_docker.py ./agents/target_agent.py --name=beaverX. This will build MAPLE and then build a docker container with the agent. If the image was built successfully it should be listed when running docker images To test the docker container start the Lunar Simulator ./RunLunarSimulator and then run ./docker/test_docker.sh --image beaverX:latest. This will run the agent for 10 real-time seconds and then automatically terminate.