W2W: A Simulated Exploration of IMU Placement Across the Human Body for Designing Smarter Wearable

TL;DR

This paper introduces W2W, a simulation framework that systematically explores optimal IMU sensor placements on the human body for activity recognition, revealing new high-utility regions and challenging traditional norms.

Contribution

W2W provides a novel, high-resolution simulation method for evaluating IMU placement utility, enabling data-driven, task-specific sensor configuration design.

Findings

Strong correlation between synthetic and real IMU data rankings

Identification of overlooked high-utility sensor regions

Potential to optimize sensor placement beyond conventional norms

Abstract

Inertial measurement units (IMUs) are central to wearable systems for activity recognition and pose estimation, but sensor placement remains largely guided by heuristics and convention. In this work, we introduce Where to Wear (W2W), a simulation-based framework for systematic exploration of IMU placement utility across the body. Using labeled motion capture data, W2W generates realistic synthetic IMU signals at 512 anatomically distributed surface patches, enabling high-resolution, task-specific evaluation of sensor performance. We validate reliability of W2W by comparing spatial performance rankings from synthetic data with real IMU recordings in two multimodal datasets, confirming strong agreement in activity-wise trends. Further analysis reveals consistent spatial trends across activity types and uncovers overlooked high-utility regions that are rarely used in commercial systems. These findings challenge long-standing placement norms and highlight opportunities for more efficient, task-adaptive sensor configurations. Overall, our results demonstrate that simulation with W2W can serve as a powerful design tool for optimizing sensor placement, enabling scalable, data-driven strategies that are impractical to obtain through physical experimentation alone.