Shape Formation for the Cooperative Transportation of Arbitrary Objects Using Multi-Agent Reinforcement Learning

2026-06-08

AI Summary

This study addresses the challenge of enabling multi-robot systems to autonomously form stable and safe support formations for transporting arbitrarily shaped objects with non-uniform mass distributions. The authors propose a multi-agent reinforcement learning approach that, for the first time, directly applies this technique to adaptive formation generation in complex object transportation without relying on predefined geometric assumptions. By integrating formation control, cooperative navigation, and obstacle avoidance mechanisms, the method enables robot teams to dynamically adjust their positions in real time to balance the load within changing environments. Experimental results demonstrate that the system consistently generates stable formations across varying numbers of robots, environmental layouts, and object geometries, and successfully generalizes to cluttered scenes and previously unseen complex objects, exhibiting strong robustness and generalization capability.

Abstract

Cooperative object transportation is essential in numerous domains, including industrial to domestic services. A popular transportation strategy is to carry objects on top of multi-robot systems. The corresponding task is typically solved by decomposing it into three interconnected subproblems: formation control, cooperative navigation, and collision avoidance. A particular challenge posed by real-world objects is their potentially arbitrary shape and non-uniform mass distribution, necessitating robot formations that securely support the object. In this work, we address the challenge of pattern formation control for transporting such real-world objects by proposing a novel multi-agent reinforcement learning approach. Our approach enables a multi-robot system to autonomously position itself underneath an object to support its weight while avoiding obstacles during the formation process. Our evaluations with diverse environments and varying numbers of robots show that our approach leads to policies that reliably produce balanced formations and generalize to cluttered scenes and objects with complex geometry and non-uniform mass distribution.

Problem

• cooperative transportation
• arbitrary objects
• formation control
• non-uniform mass distribution
• multi-agent systems

Innovation

• multi-agent reinforcement learning
• formation control
• cooperative transportation
• arbitrary-shaped objects
• non-uniform mass distribution