Onboard Mission Replanning for Adaptive Cooperative Multi-Robot Systems
Elim Kwan, Rehman Qureshi, Liam Fletcher, Colin Laganier, Victoria Nockles, Richard Walters
AI summary
Problem
Cooperative robotic teams operating in dynamic, remote environments require rapid on-board replanning without relying on fragile ground communication, yet existing methods cannot simultaneously handle variable agent locations, task durations, cooperative tasking, and edge-device constraints.
Approach
The authors formulate mission replanning as a novel variant of the multiple Traveling Salesperson Problem and solve it using a lightweight graph attention reinforcement learning model that sequentially assigns tasks to agents while accounting for variable start locations, task durations, and cooperative work.
Key results
- First formulation of the Cooperative Mission Replanning Problem (CMRP) incorporating flexible start locations, variable task times, and cooperative tasking.
- Development of GATR, a generalized graph attention reinforcement learning replanner suitable for edge deployment.
- Consistently achieves solutions within 10% of the LKH3 heuristic solver while running 85-370 times faster on a Raspberry Pi.
- Demonstrates strong generalization across varying numbers of agents, tasks, and discretization levels in a single model.
Why it matters
This work enables resilient, real-time autonomous decision-making for multi-robot teams in remote or hazardous environments, paving the way for robust edge-AI deployment in aerospace, maritime, and disaster response applications.
Abstract
Cooperative autonomous robotic systems have sig- nificant potential for executing complex multi-task missions across space, air, ground, and maritime domains. But they commonly operate in remote, dynamic and hazardous environ- ments, requiring rapid in-mission adaptation without relying on fragile or slow communication links to centralized compute. Fast, onboard replanning algorithms are therefore essential to enhance resilience for these systems, but do not yet exist. Reinforcement Learning (RL) shows strong promise for efficiently solving mission planning tasks formulated as Travelling Salesperson Problems (TSPs), but existing methods: 1) are unsuitable for replanning, where agents do not start at a single location; 2) do not allow cooperation between agents; 3) are unable to model tasks with variable durations; or 4) lack practical considerations for onboard deployment. Here we address this gap by defining the Cooperative Mission Replanning Problem as a novel adaptation of multiple TSP, and develop a new encoder/decoder-based RL model to solve it effectively and efficiently. Using a simple ex- ample of cooperative drones, we show our replanner consistently (90% of the time) maintains performance within 10% of the state- of-the-art LKH3 heuristic solver, whilst running 85-370 times faster on a Raspberry Pi. This work paves the way for increased resilience in autonomous multi-agent systems.