Evaluation of center of mass energetics during locomotion using computer vision

Abstract

Computer vision techniques have emerged as a reliable approach for estimating gait kinematics. However, the estimation of center of mass (CoM) energetic parameters using markerless methods remains unvalidated. This study assessed the accuracy of this techniques in estimating CoM energetic parameters by comparing them to a gold-standard motion capture system. Six subjects walked at different speeds on a treadmill while being recorded using both methods. CoM energetic parameters were computed and analysed to determine the error between approaches. Results showed relative errors below 5% for energy recovery during walking, suggesting that computer vision techniques hold promise for estimating CoM energetic parameters in gait analysis.