Multi-Agent path planning in Python

Dependencies

Install the necessary dependencies by running the following:

python3 -m pip install -r requirements.txt
brew install ffmpeg

Modified KKT System

Execution

Run the following to test the new KKT code:

cd ./decentralized
python3 decentralized.py -f kkt/new_kkt.mp4 -m kkt

Results

Test (With Optimization)

Parallelized Decentralized Code

Execution

python3 ./robotics_project/decentralized/decentralized.py -m nmpc -f output_animation_file_name.gif

Parallelized Total Cost Computation

Results

4 Obstacles	100 Obstacles

Parallelized Obstacle Prediction

Results

Test (With Parallel Obstacle Prediction)

Execution

To run multi-agent NMPC with the parallelized algorithms, run the notebook decentralized/MultiAgentNMPCParallel.ipynb in a GPU environment (e.g. Google Colab).

Results

Multi-Agent Decentralized NMPC w/ Parallelization