Phy-Tac: Toward Human-Like Grasping via Physics-Conditioned Tactile Goals

2025-11-03

AI Summary

A fundamental gap exists between rigid, over-squeezing robotic grasping and human-like gentle, force-optimal manipulation. Method: This paper proposes the Force-Optimal Stable Grasping (FOSG) framework, integrating pose optimization, physics-guided contact region analysis, and tactile goal generation to construct a Physics-Conditioned Latent Diffusion Model (Phy-LDM). Phy-LDM jointly encodes mechanical constraints and tactile feedback in a unified latent space, coupled with a latent-space Linear Quadratic Regulator (LQR) controller for closed-loop force regulation. Training data are synthesized via high-fidelity physics simulation, eliminating reliance on costly real-world tactile annotations. Contribution/Results: Experiments across multiple robotic platforms demonstrate that FOSG significantly improves grasp stability (+23.6%) and force efficiency (reducing actuation force by 41.2%) over fixed-force baselines and GraspNet. It achieves, for the first time, human-scale adaptive gentle manipulation, establishing a generalizable visuo-tactile–force synergistic control paradigm for embodied intelligence.

Technology Category

• Intelligent Robots: Manipulation
• Computer Vision: Low Level & Physics-based Vision
• Search and Optimization: Learning to Search

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Humans naturally grasp objects with minimal level required force for stability, whereas robots often rely on rigid, over-squeezing control. To narrow this gap, we propose a human-inspired physics-conditioned tactile method (Phy-Tac) for force-optimal stable grasping (FOSG) that unifies pose selection, tactile prediction, and force regulation. A physics-based pose selector first identifies feasible contact regions with optimal force distribution based on surface geometry. Then, a physics-conditioned latent diffusion model (Phy-LDM) predicts the tactile imprint under FOSG target. Last, a latent-space LQR controller drives the gripper toward this tactile imprint with minimal actuation, preventing unnecessary compression. Trained on a physics-conditioned tactile dataset covering diverse objects and contact conditions, the proposed Phy-LDM achieves superior tactile prediction accuracy, while the Phy-Tac outperforms fixed-force and GraspNet-based baselines in grasp stability and force efficiency. Experiments on classical robotic platforms demonstrate force-efficient and adaptive manipulation that bridges the gap between robotic and human grasping.

Problem

• Developing physics-conditioned tactile method for force-optimal stable grasping
• Bridging gap between robotic over-squeezing and human-like minimal force grasping
• Unifying pose selection, tactile prediction and force regulation in manipulation

Innovation

• Physics-conditioned tactile method for force-optimal grasping
• Latent diffusion model predicts tactile imprint under target
• Latent-space LQR controller minimizes actuation for tactile goals