用于 Sim-to-Real 操作的解耦触觉表征与控制
Factorized Tactile Representation and Control for Sim-to-Real Manipulation
Siqi Shang · Bianca Aumann · Tye Brady · Joshua Migdal · Taskin Padir
中文摘要
触觉 sim-to-real 学习需要在桥接仿真接触与设备专属传感器响应的同时,保留控制所需信息。提出一种解耦触觉表征与控制框架,将法向力与接触区域映射到可从传感器信号恢复的有效接触响应。该响应被分为接触几何、力分布与时序接触变化三部分,各部分采用专属编码与随机化策略。触觉门控策略(Tactile Gated Policy)在控制过程中分别保留这些表征,并可在全部掩码配置下运行而无需重训。通过响应重建、空间对齐、力调节以及接触密集的对抗性插 peg 任务,在仿真与真实环境中评估该方法,从而可独立评估不同触觉表征的可用性与迁移可靠性。方法在未见过的几何上实现了 <1 mm 的接触定位精度与 1.69 N 的力跟踪误差,在真实环境对抗性插 peg 任务上较未解耦响应提升 35%,且不同触觉表征对不同交互任务带来差异化收益。
关键要点
- 01问题:触觉 sim-to-real 需同时桥接仿真接触与设备专属传感器响应,并保留控制信息
- 02方法:将接触响应解耦为接触几何、力分布、时序变化,各部分独立编码与随机化,并用 Tactile Gated Policy 分别保留表征
- 03结果:未见几何上 <1 mm 接触定位、1.69 N 力跟踪误差,真实对抗插 peg 任务较未解耦方法提升 35%
- 04局限:各解耦表征对不同交互任务各有所长,需按任务选用,无单一通用最优表征
解读
尚无解读。
原始英文摘要
arXiv:2610.10510v1 Announce Type: new Abstract: Tactile sim-to-real learning must bridge simulated contact and device-specific sensor responses while preserving information needed for control. We propose a factorized tactile representation and control framework that maps normal force and contact patch to an effective contact response recoverable from sensor readings. The response is separated into contact geometry, force distribution, and temporal contact change, with representation-specific encoding and randomization. A Tactile Gated Policy preserves these representations separately through control and operates over all mask configurations without retraining. We evaluate the approach through response reconstruction, spatial alignment, force regulation, and contact-rich adversarial peg insertion in simulation and the real world, enabling the utility and transfer reliability of different tactile representations to be assessed independently. The approach achieves <1 mm contact localization, 1.69 N force-tracking error on unseen geometries, and a 35% improvement in real-world adversarial peg insertion over the unfactorized response, with different tactile representations benefiting different interactions.