Advancing Prosthetic Arm Control: An EMG-Based Grasp Force Estimation Towards Anthropomorphic Movement

Recent advancements in prosthetic arm technology revolutionized the lives of upper limb amputees, offering restored functionality and better quality of life. However, achieving precise control over grasp force modulation remains a pressing research challenge to the body-sensor network research community. Electromyography (EMG) signals have emerged as a potential solution for force estimation, mimicking natural grasp force modulation. This study investigates EMG signals associated with distinct finger movements for five grasp patterns and diverse force levels to enhance prosthetic arm control. Machine learning techniques were used to classify EMG signals from 15 individuals- findings suggest that variations in EMG responses to different finger movements and force levels demonstrate a complex relationship. Our model demonstrates accuracy above 95%, indicating reliable classification. Our study reveals unique muscle activity trends for mild and moderate force levels based on a comprehensive force and EMG data analysis across multiple subjects. These findings contribute to developing prosthetic systems closely mimicking natural limbs, enabling effortless object manipulation with varying force levels leveraging insights on dynamics of EMG signals, human motor control, and intuitive control mechanisms in prosthetic arms.