Real-Time Muscle-Level Haptic Feedback for Enhanced Welding Learning: An sEMG-Based Approach

Abstract

Engaging muscle groups in proper order is an essential skill for human motor tasks such as welding, contributing to musculoskeletal health and task performance. However, learning the correct muscle engagement strategy for a dedicated human motion task like welding can be challenging due to the difficulty of acquiring the implicit information and carrying it over to novice trainees, which is becoming the bottleneck of workforce training. This paper proposes to monitor and model the implicit muscle engagement strategies using surface electromyographic (sEMG) sensors and provide real-time feedback to novice trainees via vibrotactile devices according to the muscle engagement models. The differences between expert trainers’ and novice trainees’ muscle engagement strategies are processed to control the vibrotactile patterns and magnitudes. A human-subject experiment (N=25) was performed to validate the system design in welding training. Our results, which demonstrated the effectiveness of the proposed method in capturing motor control patterns and improving motor skill learning, provide a strong foundation for its application in real-world training scenarios. Furthermore, it was found that providing haptic feedback according to muscle engagement was more effective than visual feedback, especially in force control. It was also noticed that haptic feedback could alleviate the reliance on external feedback compared with visual feedback. This paper presented a novel method and its implementation for welding training, contributing to innovative training methods for the workforce and beyond.