순동역학과 역동역학 접근법에 따른 보행 중 하지 관절 운동역학 결과 비교 분석

다국어 초록 (Multilingual Abstract)

This study compares joint kinematics and kinetics between inverse and forward dynamics methods in gait analysis, using a full-body skeletal model from a healthy male with 25 degrees of freedom.The inverse dynamics method requires the use of residual forces and torques to address the effects of low-pass filter usage and measurement equipment errors. Conversely, forward dynamics simulation and a gait controller calculates joint kinetics without residual forces and torques. The gait controller based on an artificial neural network trained via deep reinforcement learning to mimic the subject’s gait kinematics.The gait controller reproduced a root mean square (RMS) kinematic difference of 2.5 degrees compared to inverse dynamics methods. It also calculated higher joint torque and power. The energy consumption difference between the methods was minimal at 1.8%, with 18% of the energy in the inverse dynamics method supplied by residual forces and torques.