A major challenge in enabling a wearable robotic system to assist the user in an outdoor setting is the adaptability of the control system to varying walking conditions. To maximize the benefit of the assistance using an exoskeleton, the assistance should be able to modulate the assistance level based on the user’s biological demand. The following study aimed to develop a new controller that can modulate assistance based on dynamically changing walking conditions, Science Advances (Published on Dec 18th, 2024).
Human-Robot Interactive Upper Limb Training Using Real-time Muscle Dynamics Tracking
The following research study aims to develop advanced real-time muscle dynamics informed Artificial Intelligence (AI) models to accurately track online and in real-time the dynamics and deformation of skeletal muscles of interest. Then, the study aims to develop a user-specific muscle-in-the-loop strength training with a robot for able-bodied adults. This study will allow monitoring of enable user-specified muscle-in-the-loop assistive/training strategies in real-time.
Effectiveness of Robotic Assistance in Patient Population using a Powered Knee Exoskeleton
The following study targets patients with genu recurvatum. The gait abnormality primarily associated with the knee joint caused by walking disability hugely affects the mobility of the patient population. The following study aimed to investigate the changes in the biomechanics of the patient population with assistance from the knee exoskeleton. The impedance controller, which works as an assist-as-needed paradigm, was used to control the knee exoskeleton, which leads users to a set gait trajectory. The data from the patients showed an increase in peak knee flexion and improvements in hyperextensions of the affected leg.
Rehabilitation Effectiveness of Robotic Assistance and Visual feedback System using a Powered Knee Exoskeleton and Visual Game Display
The following project aims to analyze the rehabilitation effect of the feedback system that provides both visual feedback from the game interface and sensory feedback from the exoskeleton assistance. Earlier pilot experiments with a stroke patient with just exoskeleton assistance feedback showed no rehabilitation effect. Therefore, the visual feedback interface was added to provide additional feedback for the user. In the pilot experiment with a genu recurvatum patient, the user had five different visits with 30 minutes training session in each visit. The comparison in no-exoskeleton condition after each training session showed improvements in knee flexion angle and hyperextension of the knee.
Real-time Joint Angle and Torque Estimator using Wearable Ultrasonic Sensors
Skeletal muscles are the primary drivers of joint movement, force generation, and regulation. Even when performing identical movements, muscle activity and synergy can vary significantly across different individuals. Providing appropriate and adaptive assistance tailored to various individual conditions and tendencies is a major hurdle in human-robot interaction (HRI) research. I aimed to address these challenges by developing real-time, muscle dynamics-informed, AI-driven models that can accurately track the dynamics of skeletal muscles using wearable ultrasound sensors in real-time. As a first step towards my goal, I proposed a novel real-time approach in user-dependent, joint angle and torque estimator using B-mode ultrasound sensors.