Functional Force-Aware Retargeting from Virtual Human Demos to Soft Robot Policies

TL;DR

SoftAct is a force-aware retargeting framework that enables soft robotic hands to learn human manipulation skills from virtual reality demonstrations by explicitly modeling contact forces and geometry.

Contribution

It introduces a novel force-aware retargeting algorithm that improves skill transfer to soft robots, accommodating embodiment mismatch and nonlinear compliance.

Findings

Reduces fingertip trajectory RMSE by up to 55% compared to baselines.

Achieves higher success rates in zero-shot real-world deployment.

Effectively reproduces human manipulation skills on a soft robot hand.

Abstract

We introduce SoftAct, a framework for teaching soft robot hands to perform human-like manipulation skills by explicitly reasoning about contact forces. Leveraging immersive virtual reality, our system captures rich human demonstrations, including hand kinematics, object motion, dense contact patches, and detailed contact force information. Unlike conventional approaches that retarget human joint trajectories, SoftAct employs a two-stage, force-aware retargeting algorithm. The first stage attributes demonstrated contact forces to individual human fingers and allocates robot fingers proportionally, establishing a force-balanced mapping between human and robot hands. The second stage performs online retargeting by combining baseline end-effector pose tracking with geodesic-weighted contact refinements, using contact geometry and force magnitude to adjust robot fingertip targets in real time. This formulation enables soft robotic hands to reproduce the functional intent of human demonstrations while naturally accommodating extreme embodiment mismatch and nonlinear compliance. We evaluate SoftAct on a suite of contact-rich manipulation tasks using a custom non-anthropomorphic pneumatic soft robot hand. SoftAct's controller reduces fingertip trajectory tracking RMSE by up to 55 percent and reduces tracking variance by up to 69 percent compared to kinematic and learning-based baselines. At the policy level, SoftAct achieves consistently higher success in zero-shot real-world deployment and in simulation. These results demonstrate that explicitly modeling contact geometry and force distribution is essential for effective skill transfer to soft robotic hands, and cannot be recovered through kinematic imitation alone. Project videos and additional details are available at https://soft-act.github.io/.