Skip to content

Latest commit

 

History

History
7 lines (4 loc) · 1.21 KB

README.md

File metadata and controls

7 lines (4 loc) · 1.21 KB

robust_cbf

This code simulates a multi-agent environment in which a controlled agent (blue) navigates from a start to goal position while avoiding collisions with other agents (red). Each of the other agents has a randomized (unknown) goal that they try to reach, but approximately half of them blindly travel from their start to goal position, while the other half exhibit some collision avoidance behavior. Our code implements the algorithm in the paper "Safe Multi-Agent Interaction through Robust Control Barrier Functions with Learned Uncertainties", which maintains safety of the blue agent by using a robust multi-agent Control Barrier Function (CBF) that predicts and accounts for uncertainty about other agents(red).

Run "python game_GP.py" to run the simulation with the robust multi-agent Control Barrier Function (CBF) in a randomized environment. This runs 1000 trials, first with the robust CBF controller, and next with the nominal CBF controller. The CBF code setting up the QP is found in "control.py". The MVG code predicting the uncertainties is found in "GP_predict.py"

The script "GP_train.py" optimizes the MVG hyperparameters, based on entered training data. To execute it, you must load in a set of training data.