Predicting athlete ground reaction forces and moments from motion capture

Area of Science:

• Biomechanics
• Sports Science
• Data Science

Background:

• Understanding athlete ground reaction forces and moments (GRF/Ms) is crucial for calculating net joint forces, moments, and injury risk.
• Traditional force platforms are lab-bound, limiting field-based kinetic data collection.
• Marker-based motion capture offers a potential alternative for estimating GRF/Ms.

Purpose of the Study:

• To investigate the feasibility of using marker-based motion capture data to estimate athlete GRF/Ms.
• To determine if partial least squares (PLS) regression can model the relationship between motion capture and GRF/Ms.
• To reduce reliance on traditional, laboratory-based force platforms.

Main Methods:

• Analysis of 11 different partial least squares (PLS) regression methods.
• Utilizing raw motion capture trajectories as input.
• Comparing estimated GRF/Ms with data from force platforms.

Main Results:

• Achieved average correlation coefficients of 0.9804 for GRFs and 0.9143 for GRMs.
• Demonstrated the feasibility of accurately predicting GRF/Ms from motion capture data.
• Validated the potential for real-time estimation of kinetic parameters.

Conclusions:

• Marker-based motion capture can accurately estimate athlete GRF/Ms in real-time.
• This method overcomes the limitations of laboratory-based force platforms.
• The findings have the potential to revolutionize athlete performance enhancement and injury prevention through non-invasive biomechanical analysis.