ForceBand: Learning Forceful Manipulation with sEMG

2026-06-24

AI Summary

Current robot imitation learning struggles with force-sensitive manipulation tasks due to the absence of contact force information in demonstrations. This work proposes a multimodal demonstration acquisition method leveraging ForceBand, a low-cost wrist-worn sEMG device, which fuses surface electromyography (sEMG), inertial measurement unit (IMU), and video data to construct the first dataset enabling fine-grained fingertip force estimation. An EMG2Force model is trained to map sEMG signals to individual finger contact forces, thereby generating force-augmented demonstrations without requiring dedicated force sensors and significantly lowering the barrier to force-aware imitation learning. Experimental results show that the proposed approach reduces force prediction error by over 50% compared to vision-only baselines and achieves an 87% success rate across diverse fine manipulation tasks, including grasping, squeezing, and placing various objects.

Abstract

Human demonstrations are a scalable data source for learning robot manipulation policies. However, common sources of human demonstration data, such as motion-capture trajectories and internet videos, capture mostly motion and appearance while missing the contact forces that are critical for force-sensitive manipulation. In this paper, we introduce ForceBand, a low-cost wrist-worn sEMG system that turns human muscle activity into force-enriched demonstrations. We first collect a 10-hour multimodal dataset containing egocentric video, sEMG, IMU, and fingertip force measurements across diverse actions and objects. Using this dataset, we pre-train an EMG2Force model that predicts per-finger forces from sEMG and IMU signals. After a short user-specific calibration, users can collect target-task demonstrations using only ForceBand and video; EMG2Force then labels these demonstrations with per-finger force traces, producing force-augmented demonstrations for robot policy learning. Experiments show that ForceBand recovers fine-grained fingertip interactions with over 50% lower force prediction error than vision-based baselines and achieves an 87% success rate on pick, squeeze, and place tasks that require object-specific force control across objects with diverse shapes, sizes, and weights. Project website: https://forceband-emg.github.io

Problem

• force-sensitive manipulation
• human demonstrations
• contact forces
• robot manipulation
• sEMG

Innovation

• sEMG
• force estimation
• robotic manipulation
• human demonstration
• multimodal learning