Multi-Quadruped Cooperative Object Transport: Learning Decentralized Pinch-Lift-Move

2025-09-17

AI Summary

This work addresses the decentralized collaborative transport problem for multi-agent systems: a team of quadrupedal robots—operating without communication, centralized control, or mechanical coupling—cooperatively grasp, lift, and move ungraspable objects solely through physical contact forces. We propose a hierarchical policy architecture that decouples base locomotion from end-effector control, coupled with a constellation reward mechanism. By sharing policy parameters and leveraging implicit synchronization cues, the approach enables emergent group coordination, endowing the loosely coupled contact system with rigid-body-like dynamical behavior. The method supports arbitrary team scalability without retraining. Extensive simulations with 2–10 robots successfully transport objects of diverse shapes and masses; sim-to-real transfer is validated on lightweight objects. Our approach significantly enhances robustness and scalability in communication-free, physics-based multi-robot collaboration.

Technology Category

• Intelligent Robots: Multi-Robot Systems
• Multiagent Systems: Teamwork
• Humans and AI: Planning and Decision Support for Human-Machine Teams

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Decentralized Web and Fediverse systems
• Economics, Online Markets and Human Computation: The sharing economy
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

We study decentralized cooperative transport using teams of N-quadruped robots with arm that must pinch, lift, and move ungraspable objects through physical contact alone. Unlike prior work that relies on rigid mechanical coupling between robots and objects, we address the more challenging setting where mechanically independent robots must coordinate through contact forces alone without any communication or centralized control. To this end, we employ a hierarchical policy architecture that separates base locomotion from arm control, and propose a constellation reward formulation that unifies position and orientation tracking to enforce rigid contact behavior. The key insight is encouraging robots to behave as if rigidly connected to the object through careful reward design and training curriculum rather than explicit mechanical constraints. Our approach enables coordination through shared policy parameters and implicit synchronization cues - scaling to arbitrary team sizes without retraining. We show extensive simulation experiments to demonstrate robust transport across 2-10 robots on diverse object geometries and masses, along with sim2real transfer results on lightweight objects.

Problem

• Decentralized coordination of quadruped robots without communication
• Transporting ungraspable objects through contact forces alone
• Scaling cooperative manipulation to arbitrary team sizes

Innovation

• Decentralized hierarchical policy architecture
• Constellation reward for rigid contact
• Shared parameters enable scaling