Objective: This study aimed to investigate the influence of varying sensor attachment pressure conditions and normalization techniques on surface electromyography (sEMG) signals and maximum voluntary contraction (MVC) measurements in lower-limb muscles. It also intended to assess how these factors affect the interpretation of muscular activity during the operation of wearable lower-limb exoskeletons.
Background: Lower-limb exoskeletons are among the most prominent technologies designed to assist and/or augment muscle load and strength across various industrial
applications. Surface electromyography (sEMG) has been widely employed to evaluate the usability of such exoskeletons. However, due to the structural characteristics of exoskeletons, physical interference between the device and sEMG sensors often leads to unintended pressure on the sensors. This sensor pressure can cause signal distortion or interference. To address this issue, researchers have attempted to reposition EMG sensors, leading to an inaccurate assessment of the exoskeletons.
Method: Twenty healthy adults (10 males and 10 females) participated in this study. Surface EMG electrodes were placed on four lower-limb muscles: the rectus femoris (RF), biceps femoris (BF), tibialis anterior (TA), and gastrocnemius (GAS). Maximum voluntary contraction (MVC) and task-related EMG signals were recorded under three sensor attachment pressure conditions: no pressure (NP), medium pressure (MP), and high pressure (HP). The EMG data were normalized using three corresponding MVC
references: NPMVC, MPMVC, and HPMVC. Participants performed static and dynamic half-squat tasks, as well as exoskeleton-assisted tasks. The data were analyzed to assess the main effects of pressure conditions and normalization method, as well as their interaction effects on muscle activation patterns.
Results: Results showed that there were no significant effects of sensor pressure or normalization methods on MVC or task-related sEMG output. Overall muscle activation trends remained consistent across all sensor pressure conditions. During the half-squat tasks, the anterior muscles - rectus femoris (RF) and tibialis anterior (TA) - exhibited higher activation levels compared to the posterior muscles biceps femoris (BF) and gastrocnemius (GAS). This activation pattern was stable irrespective of sensor pressure condition and normalization technique. In exoskeleton-assisted tasks, overall muscle activation was reduced relative to the half-squat tasks; however, the relative activation hierarchy among muscles was maintained and similarly unaffected by variations in pressure or normalization method.
Conclusion: This study demonstrated that sensor attachment pressure and normalization method had no significant effect on lower-limb muscle activation outcomes across MVC assessments, half-squat exercises, and exoskeleton-assisted tasks. However, as the investigation was limited to anteriorly applied sensor pressure and a young adult population, further research is needed to validate the generalizability of these findings across different sensor placements and demographic groups.
Application: The results provide a methodological basis for reliably evaluating exoskeleton performance without modifying sensor placement or normalization strategies, thereby minimizing unnecessary adjustments in future assessments.