Coordinating different legged robots to move shared loads smoothly
HA-DiPO: Diffusion Policy Optimization for Heterogeneous Cooperative Transport
Robotics
Summary
Moving a shared object with different types of walking robots is tricky because each robot moves differently. The authors created HA-DiPO, a system that helps these robots work together by splitting the problem into two parts: one that decides the team’s plan and another that handles each robot’s unique movements. This system learns from experience in a simulator without needing examples to copy. It lets each robot figure out how to help move the load using only its own information, without talking to others. Tests show it works better than previous methods, including real-world trials with robots and even a human helping.
What this means in practice
- •For robotics development teams: Create coordination systems that handle diverse legged robots moving objects without direct communication, improving efficiency in mixed-robot teams.
- •For industrial automation engineers: Implement robust cooperative transport solutions for complex payloads using multi-legged and wheeled robots operating together reliably.
Authors
Xuebin Li, Jinpeng Xiao, Zong Chen, Pierluigi Arpenti, Bruno Siciliano, Jiebao Sun, Yiqun Li
Abstract
Cooperative transport with heterogeneous legged robots requires coordinating different motion capabilities while supporting an unfixed shared payload. We present HA-DiPO (Heterogeneous-Agent Diffusion Policy Optimization), a hierarchical multi-agent reinforcement learning framework under centralized training with decentralized execution (CTDE). The framework decouples team-level coordination from embodiment-specific motion control: at the high level, a centralized critic uses team-value estimates to jointly train independent coordination policies, aligning robot-specific action responses with a shared transport objective; at the low level, embodiment-specific robust motion policies track the coordination commands. The high-level diffusion policies offer more flexible action distributions than single-Gaussian parameterizations, allowing potentially multimodal coordination responses to be represented. The coordination policies learn from scratch through reward-driven simulator interaction without demonstrations. At execution, each robot generates coordination commands using only local robot and payload observations, without partner-state inputs or inter-robot messages. Simulation comparisons show that HA-DiPO achieves higher final mean returns and velocity-tracking success rates, as well as lower velocity errors, than the evaluated baselines. Real-world experiments further demonstrate quadrupedal-wheeled-legged and human-robot cooperative transport.