Learning Tactile-Aware Quadrupedal Loco-Manipulation Policies

arXiv cs.RO / 5/1/2026

📰 NewsIdeas & Deep AnalysisModels & Research

共有:

Key Points

The paper addresses a key challenge in quadrupedal loco-manipulation: vision and proprioception alone struggle to handle uncertain, evolving contact interactions, while tactile sensing can provide direct contact observability.
It proposes a hierarchical, tactile-aware policy learning pipeline that first trains a tactile-conditioned visuotactile high-level policy using real-world human demonstrations.
The high-level policy jointly predicts manipulation end-effector trajectories and time-evolving tactile interaction cues that specify how contact should develop.
It then uses large-scale reinforcement learning in simulation to learn a tactile-aware whole-body control policy that can follow diverse commanded trajectories and tactile cues and transfer zero-shot to real hardware.
Experiments on real contact-rich tasks (reorientation with insertion, valve tightening, and delicate object manipulation) show an average 28.54% performance improvement over vision-only and visuotactile baselines.

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.