# How Noisy Poses Break Inverse Dynamics: Analysis and Mitigation for Video-Based Joint Torque Estimation

arXiv: https://arxiv.org/abs/2605.24776 · HTML: https://mlsanity.com/paper/2605.24776/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 6 of its 15 references that are in the map.
- First posted on arXiv 23 May 2026.

## Builds on (most cited first, 6 of 6)

- [Humans in 4D: Reconstructing and Tracking Humans with Transformers](https://mlsanity.com/paper/2305.20091/) (2023)
- [WHAM: Reconstructing World-grounded Humans with Accurate 3D Motion](https://mlsanity.com/paper/2312.07531/) (2023)
- [TokenHMR: Advancing Human Mesh Recovery with a Tokenized Pose Representation](https://mlsanity.com/paper/2404.16752/) (2024)
- [PhysPT: Physics-aware Pretrained Transformer for Estimating Human Dynamics from Monocular Videos](https://mlsanity.com/paper/2404.04430/) (2024)
- [ImDy: Human Inverse Dynamics from Imitated Observations](https://mlsanity.com/paper/2410.17610/) (2024)
- [Optimal-state Dynamics Estimation for Physics-based Human Motion Capture from Videos](https://mlsanity.com/paper/2410.07795/) (2024)

## Abstract (arXiv)

Recent advances in monocular 3D human pose estimation enable accurate body tracking from video. However, translating these kinematic estimates into physical quantities, such as joint torques, remains challenging due to noise amplification through inverse dynamics. In this work, we provide a systematic analysis of how pose estimation noise propagates through the inverse dynamics pipeline. We present three key findings: (1) pose noise is amplified by approximately 1,000x when computing joint torques via numerical differentiation, (2) proximal joints (spine, hips) are up to 10x more sensitive to noise than distal joints (wrists, hands), and (3) low-pass filtering before differentiation substantially reduces this amplification. To enable this analysis, we develop SMPL-Dynamics, a fully differentiable inverse dynamics module for the SMPL body model that requires no external physics simulators. Our module supports end-to-end gradient computation, and we demonstrate this through differentiable pose refinement, which reduces torque error by 93% with negligible change in pose.