FACTR 2: Learning External Force Sensing for Commodity Robot Arms without Dedicated Force Sensors

Forum topic · 小凯 · 2026-06-12

Summary

FACTR 2 introduces NEXT (Neural External Torque Estimation), a data-driven method that estimates external joint torques on commodity robot arms without any dedicated force sensors. NEXT trains in just 1 minute from only 10 minutes of free-motion data, yet achieves estimation accuracy comparable to dedicated joint-torque sensors. This capability enables force-feedback teleoperation on low-cost robot arms. Building on NEXT, the paper proposes FIRST (Force-Informed Re-Sampling Training), which improves imitation learning by up-sampling pre-contact and contact segments during behavior cloning. Across five long-horizon manipulation 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 robot arms with no additional sensing hardware, lowering the barrier to contact-rich manipulation. The work is by researchers including Steven Oh, Kenneth Shaw, Ruslan Salakhutdinov, and Deepak Pathak, available as arXiv:2606.12406.

Paper Overview

Field: Machine Learning / Robotics
Authors: Steven Oh, Jason Jingzhou Liu, Tony Tao, Philip Han, Kenneth Shaw, Satoshi Funabashi, Ruslan Salakhutdinov, Deepak Pathak
arXiv: 2606.12406

Abstract

Contact-rich manipulation requires force sensitivity, but many robot arms lack dedicated force sensors due to their high cost. The authors present Neural External Torque Estimation (NEXT), a data-driven method that estimates external joint torques without needing any dedicated force sensors.

Key Contributions

NEXT (Neural External Torque Estimation): Trains in 1 minute from only 10 minutes of free-motion data, yet achieves estimates comparable to dedicated joint-torque sensors.
Force-feedback teleoperation: NEXT enables force-feedback teleoperation on low-cost robot arms.
FIRST (Force-Informed Re-Sampling Training): Improves policy learning by up-sampling pre-contact and contact segments during behavior cloning.
Results: Across five long-horizon tasks, FIRST outperforms prior force-aware policies by over 17% in task progress.

Conclusion

Together, NEXT and FIRST bring force-aware teleoperation and policy learning to commodity, off-the-shelf robot arms without requiring any additional sensing hardware.