Partial Motion Imitation for Learning Cart Pushing with Legged Manipulators

2026-03-27

AI Summary

This work addresses the challenge of balancing stability and precision in quadrupedal robots during mobile manipulation tasks such as pushing a cart, where locomotion and manipulation are tightly coupled. To decouple these behaviors, the authors propose a partial adversarial motion prior that applies imitation learning exclusively to the lower limbs. A robust locomotion policy is first acquired through domain and terrain randomization, which is then integrated with partial imitation to form a coordinated mobile manipulation strategy. Evaluated in both IsaacLab and MuJoCo simulation environments, the proposed approach significantly outperforms baseline methods across diverse cart-pushing trajectories, demonstrating superior stability, accuracy, and generalization capability.

Abstract

Loco-manipulation is a key capability for legged robots to perform practical mobile manipulation tasks, such as transporting and pushing objects, in real-world environments. However, learning robust loco-manipulation skills remains challenging due to the difficulty of maintaining stable locomotion while simultaneously performing precise manipulation behaviors. This work proposes a partial imitation learning approach that transfers the locomotion style learned from a locomotion task to cart loco-manipulation. A robust locomotion policy is first trained with extensive domain and terrain randomization, and a loco-manipulation policy is then learned by imitating only lower-body motions using a partial adversarial motion prior. We conduct experiments demonstrating that the learned policy successfully pushes a cart along diverse trajectories in IsaacLab and transfers effectively to MuJoCo. We also compare our method to several baselines and show that the proposed approach achieves more stable and accurate loco-manipulation behaviors.

Problem

• loco-manipulation
• legged robots
• cart pushing
• stable locomotion
• mobile manipulation

Innovation

• partial imitation learning
• loco-manipulation
• legged manipulators
• adversarial motion prior
• domain randomization