Learning Tactile-Aware Quadrupedal Loco-Manipulation Policies

TL;DR

This paper introduces a hierarchical tactile-aware learning framework for quadrupedal loco-manipulation, combining real-world demonstrations and simulation-based reinforcement learning to improve contact-rich task performance.

Contribution

It presents a novel hierarchical pipeline that integrates tactile sensing with visual perception, enabling zero-shot transfer of learned policies to real-world contact-rich tasks.

Findings

Achieved a 28.54% average performance improvement over vision-only baselines.

Successfully transferred tactile-aware policies from simulation to real-world tasks.

Enhanced coordination in locomotion and manipulation under contact-rich scenarios.

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.