Dynamics-Regulated Kinematic Policy for Egocentric Pose Estimation

TL;DR

This paper introduces a novel egocentric pose estimation method that combines kinematic and dynamics models with scene object information to achieve physically plausible 3D human-object interaction estimation from a single wearable camera.

Contribution

It presents a dynamics-regulated training approach that synergizes kinematic and dynamics models for improved pose estimation in egocentric views.

Findings

Achieves physically plausible 3D human-object interaction estimation.

Effective in both laboratory and real-world scenarios.

First to incorporate scene object poses into egocentric pose estimation.

Abstract

We propose a method for object-aware 3D egocentric pose estimation that tightly integrates kinematics modeling, dynamics modeling, and scene object information. Unlike prior kinematics or dynamics-based approaches where the two components are used disjointly, we synergize the two approaches via dynamics-regulated training. At each timestep, a kinematic model is used to provide a target pose using video evidence and simulation state. Then, a prelearned dynamics model attempts to mimic the kinematic pose in a physics simulator. By comparing the pose instructed by the kinematic model against the pose generated by the dynamics model, we can use their misalignment to further improve the kinematic model. By factoring in the 6DoF pose of objects (e.g., chairs, boxes) in the scene, we demonstrate for the first time, the ability to estimate physically-plausible 3D human-object interactions using a single wearable camera. We evaluate our egocentric pose estimation method in both controlled laboratory settings and real-world scenarios.