Reliability-Gated Fusion of Consumer Head and Foot IMUs for Lower-Body Sparse Inertial Pose Estimation

Paper Overview

• Field: Computer Vision
• Authors: Zhilin Guo, Boqiao Zhang, Oszkár Urbán, Josef Bengtson, Hakan Aktas, Wenzhao Li, Siyu Hong, Kyle Fogarty, Chenliang Zhou, Ali Senguel, Cengiz Oztireli
• Published: 2026-09-28
• arXiv: 2609.35764

Summary

Sparse inertial pose estimation promises camera-free motion capture from consumer devices, but consumer sensors are unreliable: firmware-fused orientations are biased, mounting varies between sessions, and streams drift or drop out. On a new 35-take single-subject benchmark pairing an earbud head inertial measurement unit (IMU) with two smart-insole foot IMUs (SAM-3D-Body pseudo-ground-truth labels), the authors show the reliability problem is channel-level: a channel ablation isolates foot acceleration as the most informative input (66.6 mm vs. 79.0 mm head-only) and the firmware-fused foot orientation as the liability that destroys the gain.

The proposed approach lets the model learn how much to trust each channel of each stream: one temporal gate per stream per channel block, trained with an auxiliary reliability objective using synthetically corrupted pretraining data. The channel-gated model is the most accurate of all learned fusion schemes on clean data (69.4 mm vs. 83.7 static, 86.6 ungated) and performs best under all simulated faults. Its gates suppress the natively biased foot orientation channels on clean real data without test-time supervision, and flag dropout bursts with 0.92–0.999 AUROC. Learning to gate on reliability, rather than relying on sensor count, is a key lever for deployable sparse inertial capture.

Key Findings

• Foot acceleration is the most informative input channel (66.6 mm vs. 79.0 mm head-only).
• Firmware-fused foot orientation is a liability that negates the accuracy gain.
• Per-stream, per-channel temporal gating trained with synthetic corruption achieves 69.4 mm on clean data, outperforming static (83.7 mm) and ungated (86.6 mm) fusion.
• Gates suppress biased foot orientation channels without test-time supervision and detect dropouts with 0.92–0.999 AUROC.