MORE | Fall 2020

Control and Coordination of Multi-Robot Swarm Systems

Robot swarms show great promise in performing complex tasks in areas ranging from search and rescue to interplanetary exploration and yet controlling the behaviors of these swarms effectively is an open research problem. This research investigates the control of robot behaviors in a swarm system where robots learn to act in their environment using demonstrations obtained from artificially generated swarm motion data along with Reinforcement Learning (RL). The use of RL ensures that the swarm system can adapt to uncertainty in its environment. This work also uses Graph Neural Networks (GNNs) in the controller for the swarm system that provides a robot with contextual information about its environment and serves as an inductive bias in the learning process.