TAC-LOCO: Unified Whole-Body Control for Quadrupedal TACtile-Informed LOCO-Manipulation

2026-07-11

AI Summary

This work addresses the challenge of excessive gripping force and object slippage in dynamic loco-manipulation tasks for quadrupedal robots, which arises from the absence of tactile feedback. To overcome this limitation, the authors propose a unified reinforcement learning framework that integrates tactile and proprioceptive sensing to enable whole-body coordination among legs, arms, and grippers. A key innovation lies in compressing high-dimensional tactile array data into an implicit representation embedded directly within the motion policy, allowing tactile cues to dynamically modulate grip strength. Coupled with a specially designed reward function for grasp stability and a zero-shot deployment mechanism, the approach achieves plug-and-play control on the Unitree Go2 platform, reducing required gripping force by 47% and maintaining an object drop rate below 1%.

Abstract

Dynamic loco-manipulation requires legged robots to coordinate whole-body motion while maintaining stable physical interaction with grasped objects under uncertain external forces. While tactile sensing has been widely studied for robotic manipulation, its role in dynamic whole-body control remains largely unexplored. Existing works without tactile feedback commonly grasp firmly rather than regulate the grasp according to the interaction. We propose TAC-LOCO, a tactile-augmented unified reinforcement learning framework that encodes tactile array observations from compliant grippers into a compact latent representation and joins it with proprioception for unified control of the legs, arm, and gripper. With effective grasp stability reward design, the policy learns to simultaneously track body velocity and end-effector trajectories, moderate grasp force, and prevent object slip under both gradual load changes and sudden release events. We deploy the policy zero-shot on a Unitree Go2 with an Interbotix WidowX 250 arm and tactile gripper, demonstrating dynamic tactile-informed loco-manipulation under varying external interactions, achieving a 47% reduction in grasping force and an object drop rate of less than 1%.

Problem

• dynamic loco-manipulation
• whole-body control
• tactile sensing
• grasp stability
• quadrupedal robot

Innovation

• tactile-informed control
• whole-body locomotion-manipulation
• unified reinforcement learning
• grasp stability
• quadrupedal robot