Emg and joint angle-based machine learning to predict future joint angles at the knee

Powered exoskeletons could be used to greatly improve a patient’s functionality, and this is especially true if they are able to accurately predict user intent. Electromyogram (EMG) and motion capture data was recorded from three subjects (two female, one male) while performing 15 walking trials. A Neural Net algorithm was trained using the EMG and motion capture data gathered to predict knee flexion angles at various times into the future. Results suggest that predictions made at times closer to the measured signal lead to a higher level of accuracy, and accuracy was greater when both EMG and motion capture data was used compared to each individually.