Upper Limb EMG-Based Fatigue Estimation During End Effector Robot-Assisted Activities

This study shows the evaluation of muscle fatigue by applying advanced signal processing techniques to electromyographic (EMG) signals. EMG segmentation methods were employed followed by the evaluation of 41 metrics to correlate fatigue to EMG activation during upper limb movements. By averaging the top 10 most correlated metrics, the impact of each channel on muscle fatigue analysis was also assessed. Notably, the posterior deltoid, the middle deltoid and the extensor ulnar exhibited the highest correlations, while the right brachioradialis showed the lowest correlation. Additionally, a Random Forest model was applied to the obtained metrics to predict muscle fatigue with an average accuracy of 99.50%.