NeckErgonomics: Adapting Head Rotation in HMDs with Predicted Neck Muscle Activity

Abstract

Interaction with head-mounted displays (HMDs) places a significant workload on neck muscles, exacerbated by the extra weight of the HMD. This paper introduces NeckErgonomics, an interaction approach that adapts HMD interfaces according to estimated neck muscle electromyography (EMG) signals to improve neck ergonomics. Leveraging an open-source dataset, we train a lightweight neural regression model to predict relative neck muscle activity from head yaw, head pitch, and fixation duration. We demonstrate NeckErgonomics through three model-informed applications: (1) control–display gain for head-based pointing, (2) dynamic recentering of virtual windows during knowledge work, and (3) adaptive viewport amplification for seated 360° video viewing. We evaluated the applications in a controlled within-subjects user study (N=18) to demonstrate the viability of NeckErgonomics. Across the three tasks, NeckErgonomics led to lower accumulated head movement, sustained head rotation, and perceived physical demand. The results also reveal task-specific trade-offs with speed, precision, visual stability, and sickness. Taken together, this work indicates a path toward ergonomics-by-design XR techniques in which adaptation is grounded in the estimated muscle activity rather than ambiguous behavioural cues.