Summary
Robots usually need special sensitive gloves or tactile sensors to learn how to handle objects with the right force from humans. This work shows a new way to teach robots by measuring forces on a human's glove and then estimating similar forces on the robot hand without needing extra sensors on the robot. The researchers created a system where a human wears a force-sensing glove and demonstrates with a camera, and the robot figures out how hard to hold objects by looking at its motor current and a mapping from the glove data. This method helps robots hold items gently and correctly without sharing the same sensors as humans.
What this means in practice
- •For robotic system developers: Integrate force-aware human demonstrations into position-controlled robot hands without installing tactile sensors on the robot.
- •For industrial automation teams: Implement safer, adaptive grasping in robot arms by estimating grip force from motor currents guided by human force data.
Abstract
Human-to-robot skill transfer from sensing gloves has so far relied on shared hardware: the same tactile glove worn by the demonstrator and the robot, or a learned alignment between two tactile sensors. We present GIFT (Glove-Inferred Force Transfer), a pipeline in which the interface between human and robot is a physical unit rather than a shared sensor: fingertip force is measured in newtons on the human side and estimated in newtons on the robot side. A wearable glove records finger flexion, calibrated fingertip force, and wrist orientation, while a head-mounted camera records the demonstration; no robot is present. At deployment, the robot estimates force from actuator-current residuals relative to a free-space baseline, through a calibrated mapping to newtons, so any position-controlled hand that reports motor current can serve as the deployment platform. The policy uses a glove-space state and predicts finger-position targets; the robot enters only through two calibrated adapters, a retargeting decoder and a force estimator. We evaluate GIFT on a cup grasp-and-hold task with two action-chunking policies trained on the same demonstrations, with fingertip-force inputs retained in one and zeroed in the other. In a 50-rollout evaluation with sample size and metrics fixed before scoring, both policies succeeded in all 25 rollouts. The median of the per-rollout hold-phase grip-force estimates was 53% lower with force inputs: 1.20 N versus 2.55 N (one-sided Mann-Whitney U, p<0.0001). In an observation ablation, a vision-only policy achieved 0/15 grasps, policies given hand-command state acquired the grasp, and the force inputs determined how hard the policy held. A force channel measured on the human hand thus transfers to a robot hand with no tactile hardware, through a retargeting map from five glove channels to seven robot actuators, with no sensor shared between the two.