Prediction of Three-Dimensional Ground Reaction Forces in the Golf Swing Using Wearable Inertial Measurement Units...

Area of Science:

• Biomechanics
• Sports Science
• Wearable Technology

Background:

• Ground reaction force (GRF) is critical for dynamic stability and power generation in activities like the golf swing.
• Traditional GRF measurement using force plates is confined to laboratory settings, limiting real-world performance analysis.
• Estimating GRF from wearable sensors during complex movements like golf swings remains largely unexplored.

Purpose of the Study:

• To investigate the prediction of 3D GRF during golf swings using coupled lower-limb kinematics.
• To evaluate the efficacy of various deep learning architectures and sensor configurations for GRF estimation.
• To determine the feasibility of developing wearable systems for field-based GRF assessment.

Main Methods:

• Collected bilateral hip, knee, and ankle joint angle data using inertial measurement units (IMUs).
• Acquired synchronized 3D GRF data using force plates.
• Evaluated five deep learning architectures across seven sensor configurations, focusing on the TCN-BiGRU model.

Main Results:

• The TCN-BiGRU model demonstrated high accuracy in predicting 3D GRF (R² = 0.94 ± 0.02).
• The full bilateral lower-limb sensor configuration provided the best performance, with the lead leg offering a cost-effective alternative.
• Vertical GRF components were predicted with the highest reliability.

Conclusions:

• Deep learning models can accurately estimate 3D GRF from lower-limb kinematics during golf swings.
• Wearable IMU-based systems show promise for field-based GRF assessment in dynamic sports.
• This approach facilitates advanced analysis of biomechanics in real-world sporting environments.