Learning Policies for Dynamic Coalition Formation in Multi-Robot Task Allocation

TL;DR

This paper introduces a decentralized, learning-based framework for multi-robot coalition formation that enables robots to adaptively coordinate using local information, improving efficiency in complex, dynamic task environments.

Contribution

It extends MAPPO with spatial action maps, motion planning, and intention sharing, providing a novel approach for adaptive, decentralized coalition formation in multi-robot systems.

Findings

Effective in large robot populations

Enables timely coalition revisions

Handles diverse task sets

Abstract

We propose a decentralized, learning-based framework for dynamic coalition formation in Multi-Robot Task Allocation (MRTA). Our approach extends MAPPO by integrating spatial action maps, robot motion planning, intention sharing, and task allocation revision to enable effective and adaptive coalition formation. Extensive simulation studies confirm the effectiveness of our model, enabling each robot to rely solely on local information to learn timely revisions of task selections and form coalitions with other robots to complete collaborative tasks. The results also highlight the proposed framework's ability to handle large robot populations and adapt to scenarios with diverse task sets.