This project is focused on multi-robot collaboration in decentralized settings where each robot only has access to its own observations, so successful coordination requires robots to infer what their teammates are doing or know and signal their own intentions or observations to their teammates. Current work includes designing the coordination architecture, training and evaluating it in simulated multi-robot manipulation tasks, and studying how and when robots should explicitly reason about their teammates’ observations and beliefs. The project involves a mix of robot learning, multi-agent systems, and game-theoretic planning.
All active projects
Multi-Agent Systems