High-Bandwidth Tactile-Reactive Control for Grasp Adjustment
Yonghyeon Lee, Tzu-Yuan Lin, Alexander Alexiev, Sangbae Kim
AI summary
Problem
Vision-only grasping systems are constrained by inevitable perception errors and contact uncertainty, often causing grasp failure. Existing tactile-reactive methods are limited to specific object geometries or planar grasps.
Approach
The method uses high-bandwidth fingertip tactile feedback to compute desired fingertip velocities that minimize a grasp-stability objective, then solves a joint-space quadratic program to track these velocities while avoiding collisions and respecting joint limits.
Key results
- Cross-Finger Gradient Descent consistently converges to stable antipodal grasps
- Achieves >99% convergence rates across diverse object shapes in simulation
- Demonstrates real-time grasp adjustment on a 15-DoF arm-hand system at 200 Hz
- Enables successful grasp refinement from crude initial configurations without prior object geometry
Why it matters
Enables robust, vision-independent grasp refinement for dexterous robots, crucial for real-world manipulation where perception is unreliable.
Abstract
Vision-only grasping systems are fundamentally constrained by calibration errors, sensor noise, and grasp- pose prediction inaccuracies, leading to unavoidable contact uncertainty in the final stage of grasping. High-bandwidth tac- tile feedback, when paired with a well-designed tactile-reactive controller, can significantly improve robustness in the presence of perception errors. This paper contributes to controller design by proposing a purely tactile-feedback grasp-adjustment algorithm. The proposed controller requires neither prior knowledge of the object’s geometry nor an accurate grasp pose, and is capable of refining a grasp even when starting from a crude, imprecise initial configuration and uncertain contact points. Through simulation studies and real-world experiments on a 15-DoF arm–hand system (featuring an 8-DoF hand) equipped with fingertip tactile sensors operating at 200 Hz, we demonstrate that our tactile-reactive grasping framework effectively improves grasp stability. Project page: https://reflexivegrasp.github.io.