Research theme
Multi-agent coordination
Decentralised decision-making and emergent cooperation in robot teams under partial observability and constrained communication.
We develop algorithms that let teams of robots reach collective decisions without a central planner. Our recent work studies how learned policies can be made robust to sparse, noisy, or actively adversarial communication, and how to verify that emergent group behaviour stays within safe operating envelopes.
Recent papers
- Controlling Behavioral Diversity in Multi-Agent Reinforcement Learning ICML 2024 · 2024
- BenchMARL: Benchmarking Multi-Agent Reinforcement Learning JMLR · 2024
- The Cambridge RoboMaster: An Agile Multi-Robot Research Platform DARS 2024 · 2024
- Co-Optimizing Reconfigurable Environments and Policies for Decentralized Multi-Agent Navigation IEEE Transactions on Robotics · 2025
- Concrete multi-agent path planning enabling kinodynamically aggressive maneuvers npj Robotics · 2026
- CoViS-Net: A Cooperative Visual Spatial Foundation Model for Multi-Robot Applications CoRL 2024 · 2024
- D4orm: Multi-Robot Trajectories with Dynamics-aware Diffusion Denoised Deformations IROS 2025 · 2025
- DVM-SLAM: Decentralized Visual Monocular Simultaneous Localization and Mapping for Multi-Agent Systems ICRA 2025 · 2025
- Extending robot minds through collective learning Science Robotics · 2025
- Generalised f-Mean Aggregation for Graph Neural Networks NeurIPS 2023 · 2023
- Reinforcement Learning with Fast and Forgetful Memory NeurIPS 2023 · 2023
- Heterogeneous multi-robot reinforcement learning AAMAS 2023 · 2023
- Efficient Model-Based Reinforcement Learning Through Optimistic Thompson Sampling ICLR 2025 · 2025
- When Is Diversity Rewarded in Cooperative Multi-Agent Learning? ICLR 2026 · 2026
- Pairwise is Not Enough: Hypergraph Neural Networks for Multi-Agent Pathfinding ICLR 2026 · 2026
- Remotely Detectable Robot Policy Watermarking ICLR 2026 · 2026
- Language-Conditioned Offline RL for Multi-Robot Navigation ICRA 2025 · 2025
- No-Regret Thompson Sampling for Finite-Horizon Markov Decision Processes with Gaussian Processes NeurIPS 2025 · 2025
- POPGym: Benchmarking Partially Observable Reinforcement Learning ICLR 2023 · 2023
- Provably Safe Online Multi-Agent Navigation in Unknown Environments CoRL 2024 · 2024
- ReCoDe: Reinforcement Learning-based Dynamic Constraint Design for Multi-Agent Coordination CoRL 2025 · 2025
People
- Alex Raymond
- Amanda Prorok
- Antonio Marino
- Eduardo Sebastián Rodríguez
- Guang Yang
- Jan Blumenkamp
- Jasmine Bayrooti
- Keisuke Okumura
- Lorenzo Magnino
- Maksymilian Wolski
- Manon Flageat
- Mateusz Odrowaz-Sypniewski
- Matteo Bettini
- Michael Amir
- Nicolas Pfitzer
- Peter Woo
- Qingbiao Li
- Rishabh Jain
- Ryan Kortvelesy
- Steven Morad
- Zhan Gao