We develop algorithms for multi-agent coordination, learned communication, and vision-based navigation for collaborative robot teams.

Led by Prof. Amanda Prorok at the University of Cambridge, Department of Computer Science and Technology.

FEATURED Mar 16 · npj Robotics

Concrete multi-agent path planning enabling kinodynamically aggressive maneuvers

Keisuke Okumura, Guang Yang, Zhan Gao, Heedo Woo, Amanda Prorok

We present concrete planning, a hybrid approach that captures real-world continuous dynamics while maintaining scalable guaranteed planning via discrete search. The framework integrates advances in robot dynamics learning, optimal control, and anytime complete planning into a modular system deployed with 40 robots — 20 aerial, 8 ground, and 12 obstacle robots — operating in a compact laboratory space.

pdf code video
FEATURED Sep 2025 · Science Robotics

Extending robot minds through collective learning

Amanda Prorok

The current trend toward generalist robot behaviours powered by massive, monolithic AI models is unsustainable. This viewpoint article argues for a paradigm shift toward collective robotic intelligence — a "mixture-of-robots" approach where diverse, specialised agents learn and work together, inspired by natural systems where cooperation among specialised components leads to greater intelligence and resilience.

pdf
FEATURED May 2023 · AAMAS 2023

Heterogeneous multi-robot reinforcement learning

Matteo Bettini, Ajay Shankar, Amanda Prorok

We study cooperative multi-robot tasks where the team is composed of agents with structurally different observations, action spaces, and reward functions. We introduce HetGPPO, a parameter-sharing paradigm enabling heterogeneous behaviours in multi-agent reinforcement learning with graph neural networks, achieving superior performance over role-blind baselines.

pdf code video
Lab activity
Sep 2025
POST LinkedIn

It's finally here! My new Viewpoint Article 'Extending robot

“It's finally here! My new Viewpoint Article 'Extending robot minds through collective learning' is published in Science Robotics. I argue for a paradigm shift: collective robotic intelligence — diverse, specialised agents that learn and work together — instead of massive monolithic 'do-it-all' generalist robots.” read on linkedin →

Showing the 20 most recent items. browse all research →