Real‐time estimation of a stroke patient’s motor intent derived from multimodal signals in adapting exoskeleton assistance levels within 200 MS

Impairment of the human upper limb triggered by stroke is a major source of depression linked to independence and quality of life. As a result, the concern prompted the introduction of robotic exoskeletons that can be used to provide intensive and task-specific rehabilitation. Nevertheless, most current systems are based on predetermined or gradually responsive assistance strategies that fail to respond to the changing motor intent of the patient in a satisfactory manner. The paper researches the question of whether real-time multimodal motor intent estimation could be applied in control of exoskeleton assistance during the 200 ms period to enhance the accuracy of movement and the reduction of the effort required of users. A biofeedback-exoskeleton model of the human coupled with an exoskeleton is developed and motor intent is defined, both kinetically and kinematically. Multimodal sensing involves the combination of surface electromyography, joint torque measurements, and interaction forces. Preprocessing and feature extraction are also considered to meet the demands of stringent latency limits. Computationally effective sensor fusion algorithm estimates real-time voluntary joint torque and intent confidence. These estimates are added to an assist-as-needed impedance-based control law that adjusts the support of torque based on voluntary contribution that is detected. The proposed multimodal adaptive controller is evaluated using model-based analysis and the fixed assistance and EMG-only control strategies. Findings emphasize minimizing the error of tracking the trajectory, enhancing the smoothness of interaction and ensuring that muscle activity fits in desirable efforts and does not violate the response latency sub-200 ms. Latency-sensitive multimodal intent recognition then has the potential to improve the responsiveness of upper-limb robotic rehabilitation, make it safer and more engaging to the patient.