Learning Tactile-Aware Quadrupedal Loco-Manipulation Policies

2026-04-29

AI Summary

This work addresses the limitations of current quadrupedal robots in contact-intensive manipulation tasks, which predominantly rely on vision and proprioception and struggle with dynamic, uncertain environmental interactions. The paper introduces the first scalable tactile-aware learning framework for quadrupedal loco-manipulation: a high-level policy conditioned on tactile inputs is trained from human demonstrations to predict end-effector trajectories and tactile interaction cues, while a whole-body control policy is optimized via large-scale simulation-based reinforcement learning, enabling zero-shot transfer to the real world. By explicitly modeling tactile signals as policy outputs and integrating tactile-visual perception with hierarchical imitation learning, the approach achieves a 28.54% average performance improvement over vision-only and visuo-tactile baselines in real-world contact-rich tasks, successfully accomplishing challenging operations such as in-hand reorientation and insertion, valve turning, and fine object manipulation.

Abstract

Quadrupedal loco-manipulation is commonly built on visual perception and proprioception. Yet reliable contact-rich manipulation remains difficult: vision and proprioception alone cannot resolve uncertain, evolving interactions with the environment. Tactile sensing offers direct contact observability, but scalable tactile-aware learning framework for quadrupedal loco-manipulation is still underexplored. In this paper, we present a tactile-aware loco-manipulation policy learning pipeline with a hierarchical structure. Our approach has two key components. First, we leverage real-world human demonstrations to train a tactile-conditioned visuotactile high-level policy. This policy predicts not only end-effector trajectories for manipulation, but also the evolving tactile interaction cues that characterize how contact should develop over time. Second, we perform large-scale reinforcement learning in simulation to learn a tactile-aware whole-body control policy that tracks diverse commanded trajectories and tactile interaction cues, and transfers zero-shot to the real world. Together, these components enable coordinated locomotion and manipulation under contact-rich scenarios. We evaluate the system on real-world contact-rich tasks, including in-hand reorientation with insertion, valve tightening, and delicate object manipulation. Compared to vision-only and visuotactile baselines, our method improves performance by 28.54% on average across these tasks.

Problem

• quadrupedal loco-manipulation
• tactile sensing
• contact-rich manipulation
• visuotactile perception
• whole-body control

Innovation

• tactile-aware
• loco-manipulation
• hierarchical policy
• visuotactile learning
• zero-shot transfer