Principal Investigator
Builds large-scale decentralized and collaborative intelligent systems for robot teams, enabling them to operate cohesively through environmental perception, inter-robot communication, and coordinated action. The work combines learning-based methods — graph neural networks and constrained learning — with approximation algorithms that carry provable guarantees, tested in real-world field experiments.
Earlier work introduced a unified framework for generalized coverage of point, curve, and area features in an environment, formalized as graph-based optimization problems, along with approximation algorithms and fast heuristics for large-scale instances.