FACTR 2: Learning External Force Sensing for Commodity Robot Arms Improves Policy Learning

2026-06-10

AI Summary

This work addresses the challenge of performing contact-rich dexterous manipulation on low-cost robotic arms, which typically lack dedicated force sensors. The authors propose Neural External Torque Estimation (NEXT), a method that leverages only ten minutes of free-motion data to train, within one minute, a high-accuracy model for estimating external joint torques. Coupled with Force-Informed Re-Sampling Training (FIRST)—a strategy that emphasizes learning during contact phases in behavior cloning—the approach enables standard manipulators to achieve force perception capabilities comparable to those afforded by specialized torque sensors, without any additional hardware. Evaluated across five long-horizon manipulation tasks, the method improves task success rates by over 17% compared to existing approaches, significantly advancing force-feedback teleoperation and efficient policy learning for cost-effective robotic systems.

Abstract

Contact-rich manipulation requires force sensitivity, but many robot arms lack dedicated force sensors due to their high cost. We present Neural External Torque Estimation (NEXT), a data-driven method that estimates external joint torques without needing any dedicated force sensors. NEXT trains in 1 minute from only 10 minutes of free-motion data, yet achieves estimates comparable to dedicated joint-torque sensors. NEXT enables force-feedback teleoperation on low-cost arms and improves policy learning through Force-Informed Re-Sampling Training (FIRST), which up-samples pre-contact and contact segments during behavior cloning. Across five long-horizon tasks, FIRST outperforms prior force-aware policies by over 17% in task progress. Together, NEXT and FIRST bring force-aware teleoperation and policy learning to off-the-shelf robots without additional sensing hardware. Video results and code are available at https://jasonjzliu.com/factr2

Problem

• force sensing
• robot arms
• contact-rich manipulation
• policy learning
• torque estimation

Innovation

• Neural External Torque Estimation
• Force-Informed Re-Sampling Training
• force sensing
• policy learning
• commodity robot arms