Author(s): WISDISH, S., SMITH, S., OSOFA, J., FARRIS, D., Institution: UNIVERSITY OF EXETER, Country: UNITED KINGDOM, Abstract-ID: 2364

INTRODUCTION:
Monitoring the biomechanics of military personnel during locomotion and physically demanding tasks is essential for understanding movement strategies, injury risk, and the effects of load carriage in operational environments. Wearable inertial measurement units (IMUs) are a candidate approach for field-based biomechanical sensing, with demonstrated validity during steady-state gait in controlled laboratory conditions[1]. However, their performance across varied gait conditions and highly dynamic military tasks remains unclear. Establishing confidence in IMU-derived kinematics under these conditions is critical for biomechanical monitoring in military contexts and downstream applications, including adaptive assistive systems[2]. This work therefore assessed the criterion validity of IMU-derived lower-limb kinematics in the laboratory and evaluated their feasibility during simulated military activities.
METHODS:
Two experimental studies were conducted. 1) 20 healthy adults performed treadmill walking and running in 30 s bouts across multiple speeds (1.1–3.0 m·s⁻¹) and inclines (−10%-+15%), with and without an additional 20 kg load. IMU and optical motion capture-derived lower-limb joint kinematics were determined using OpenSim and compared via root mean squared difference (RMSD) and correlation coefficients. 2) A field-based feasibility study involving 18 actively serving military personnel completing a standardised obstacle course replicating dismounted military tasks. IMUs were worn throughout. Sensor disruptions were recorded and full-body kinematics were compared during standardised functional movements performed pre- and post-course completion.
RESULTS:
1) Hip and knee flexion-extension displayed strongest IMU-motion capture agreement during walking (RMSD = ~10° [~3-13° IQR]; Hip: r=0.91; Knee: r=0.92) irrespective of gradient or load. Differences increased during jogging and running (~20°), while sagittal hip and knee correlations remained high (r = 0.79-0.96). Frontal- and transverse-plane kinematics exhibited largest errors (~30°) and weakest correlations. Removing systematic model pose offsets substantially reduced RMSD values, indicating calibration-related bias.
2) 17 participants (94%) completed post-task assessments with usable IMU data. 26 sensor disruptions were recorded, predominantly affecting lower-limb sensors during prone and crawling tasks. Pre- and post- lower-limb joint range of motion RMSD ranged from 4-16° (r = 0.6-0.9). Post-task walking kinematics generally remained within the pre-task mean ± 2 standard deviations.
CONCLUSION:
While using IMUs for monitoring the biomechanics of military personnel during dynamic tasks seems feasible, our work identifies several key issues that need addressing for successful application. This includes optimising sensor attachment and data-processing strategies that avoid the need for input from other systems.
References: 1) Al Borno M, et al. J NeuroEngin Rehabil. 2022;19:22; 2) Slade P, et al. Nature. 2022;610:277–82.