TacBPM: A Tactile-conditioned Behavior Prior Model for Dexterous Reorientation

2026-09-16

AI Summary

提出了一种基于触觉条件的行为先验模型TacBPM，用于灵巧的手内重定向任务，通过减少从原始关节命令开始的重新探索来加速训练并实现稳定策略。

Abstract

Dexterous in-hand manipulation requires policies that coordinate high-DoF hand joints through intermittent, contact-rich interaction. Beyond target-orientation tracking, such policies must discover finger gaits that preserve object stability while adapting to geometry, anisotropy, pose, contact, and sensing changes. We propose \method, a tactile-conditioned behavior prior model for dexterous reorientation. \method distills multi-scale sphere specialists into a latent controller and lets downstream policies reuse the fixed tactile prior through residual latent actions, reducing renewed exploration from raw joint commands. The prior conditions on tactile-proprioceptive history so latent behavior reflects the current hand-object interaction. We evaluate arbitrary-pose transfer across anisotropic objects, commanded-axis rotation, and an arm-hand Grasp-to-AnyPose task in which the robot must grasp, lift, transport, and reach goal poses for novel tool geometries and generalized placements. Extensive experiments demonstrate that the proposed method accelerates training and enables stable policies where matched raw-action PPO remains near failure, with successful sim-to-real transfer in in-hand and arm-hand tasks.

Problem

• dexterous in-hand manipulation
• high-DoF hand joints
• finger gaits
• object stability
• geometry and anisotropy

Innovation

• tactile-conditioned behavior prior
• dexterous reorientation
• latent controller
• residual latent actions
• sim-to-real transfer