Physical Interaction: Reconstructing Hand-object Interactions with Physics

Abstract

Single view-based reconstruction of hand-object interaction is challenging due to the severe observation missing caused by occlusions. This paper proposes a physics-based method to better solve the ambiguities in the reconstruction. It first proposes a force-based dynamic model of the in-hand object, which not only recovers the unobserved contacts but also solves for plausible contact forces. Next, a confidence-based slide prevention scheme is proposed, which combines both the kinematic confidences and the contact forces to jointly model static and sliding contact motion. Qualitative and quantitative experiments show that the proposed technique reconstructs both physically plausible and more accurate hand-object interaction and estimates plausible contact forces in real-time with a single RGBD sensor.

arXiv commentAccepted to SIGGRAPH Asia 2022, Conference Track

References in corpus (5)

• DeepMimic: Example-Guided Deep Reinforcement Learning of Physics-Based Character Skills
• Capturing Hands in Action using Discriminative Salient Points and Physics Simulation
• SFV: Reinforcement Learning of Physical Skills from Videos
• PhysCap: Physically Plausible Monocular 3D Motion Capture in Real Time
• Physically Plausible Pose Refinement using Fully Differentiable Forces

Cited by in corpus (1)

• Adaptive Motion Planning for Multi-fingered Functional Grasp via Force Feedback