Research map / 3D vision / Human pose estimation
Ego-Pose Estimation and Forecasting As Real-Time PD Control
Ye Yuan, Kris Kitani
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add full reward (r_e, r_p, r_v terms) over pose-only reward r_q (ablation) | ego-pose estimation policy · Cross-subject MoCap | E_pose: 1.236 -> 1.183 | Improved |
| add end-effector reward to pose reward only (ablation) | ego-pose estimation policy · Cross-subject MoCap | number of resets: 14 -> 4 | Improved |
| add root rewards on top of r_q + r_e (ablation) | ego-pose estimation policy · Cross-subject MoCap | number of resets: 55 -> 4 | Improved |
| use proposed local-frame reward instead of DeepMimic reward (ablation) | ego-pose estimation policy · Cross-subject MoCap | E_pose: 1.515 -> 1.183 | Improved |
| use proposed local-frame reward instead of DeepMimic reward (ablation) | ego-pose estimation policy · Cross-subject MoCap | average acceleration: 17.504 -> 5.260 | Improved |
| add fail-safe reset mechanism (ablation) | ego-pose estimation policy · Cross-subject MoCap | E_pose: 1.206 -> 1.183 | Improved |
| add fail-safe reset mechanism (ablation) | ego-pose estimation policy · Cross-subject MoCap | number of resets: 4 -> 4 | No clear effect |
| use control-based PD policy instead of direct kinematic regression (PoseReg) | ego-pose estimation · Single subject MoCap | E_pose: 0.833 -> 0.640 | Improved |
| use PD action space with target joint angles instead of joint torques (VGAIL) | ego-pose estimation · Single subject MoCap | number of resets: 94 -> 1.4 | Improved |
| use physics-based forecasting policy instead of acLSTM | ego-pose forecasting · Cross-subject MoCap (3s) | E_pose: 1.511 -> 1.339 | Improved |
| use physics-based forecasting policy instead of ERD | ego-pose forecasting · Cross-subject MoCap (1s) | E_pose: 1.374 -> 1.179 | Improved |
Closest papers by meaning
- Whole-Body Conditioned Egocentric Video Prediction (NeurIPS 2025 · similarity 0.89)
- EgoGen: An Egocentric Synthetic Data Generator (CVPR 2024 · similarity 0.86)
- 3D Human Pose Perception from Egocentric Stereo Videos (CVPR 2024 · similarity 0.86)
- Optimal-state Dynamics Estimation for Physics-based Human Motion Capture from Videos (NeurIPS 2024 · similarity 0.86)
- AC3D: Analyzing and Improving 3D Camera Control in Video Diffusion Transformers (CVPR 2025 · similarity 0.86)
- Object-centric 3D Motion Field for Robot Learning from Human Videos (NeurIPS 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
