UC Berkeley

With the predicted increase in worldwide elderly population in the future and already significant populations of disabled people, assistive technologies and rehabilitation devices are demanded significantly. Utilizing a human mechatronic approach results in several advantages, including capability of measuring insightful information for patient's condition and providing proper assistive torque for abnormal movement correction. This dissertation investigates several domains, including (1) human dynamics model, (2) monitoring systems, and (3) design and control of active lower extremity exoskeleton.

The dissertation begins with a study of a human dynamic model and sensing system for diagnosis and evaluation of patient's gait condition as first step of rehabilitation. A 7-DOF exoskeleton equipped with multiple position sensors and smart shoes is developed, so that this system can deliver patient's joint motion and estimated joint torque information. A human walking dynamic model is derived as it consists of multiple sub-dynamic models corresponding to each gait phase. In addition, a 3D human motion capture system is proposed as it utilizes an inertial measurement unit (IMU) sensor for 3D attitude estimation with embedded time varying complementary filter. This sensing system can deliver 3D orientations of upper extremities, and a forward kinematics animation. For the development of a rehabilitation device, an active lower extremity exoskeleton is proposed. A rotary series elastic actuator (RSEA) is utilized as a main actuator of the exoskeleton. The RSEA uses a torsion spring yielding elastic joint characteristics, which is safe for human robot interaction applications. A RSEA controller design is implemented, including a PID controller, a feedforward controller for friction compensation, and a disturbance observer for disturbance rejection. All sensing and actuation systems developed in this dissertation are verified by simulation studies and experiments.