Powered lower limb exoskeletons have been widely researched and developed for walking rehabilitation and assistance in activities of daily living. However, there are many practical problems that limit the applications of current exoskeletons. To achieve overground walking, gait patterns need to be designed for the pilot-exoskeleton system under the considerations of system parameters (total height and weight, size and inertial parameters of segments, range of joint motion, capacity of joint torque output, etc.) and walking conditions (ground inclination, length and height of stairs, stride lengths, step durations, etc.). Due to the complexity of walking dynamics and behaviors, it is difficult to model and solve the gait generation problem. After the gait patterns are predefined, general...[ Read more ]

Powered lower limb exoskeletons have been widely researched and developed for walking rehabilitation and assistance in activities of daily living. However, there are many practical problems that limit the applications of current exoskeletons. To achieve overground walking, gait patterns need to be designed for the pilot-exoskeleton system under the considerations of system parameters (total height and weight, size and inertial parameters of segments, range of joint motion, capacity of joint torque output, etc.) and walking conditions (ground inclination, length and height of stairs, stride lengths, step durations, etc.). Due to the complexity of walking dynamics and behaviors, it is difficult to model and solve the gait generation problem. After the gait patterns are predefined, generally, a position-based control strategy is performed to realize walking with fixed gait parameters or limited patterns, resulting in poor adaptation to walking tasks. Even though multiple patterns have been performed in some research, the transition between various patterns is addressed by restarting from stand pose.

To solve the above challenges, we propose an individualized gait pattern generation method and DMP-based control strategy for a lower limb exoskeleton in therapy. We first develop an IMU-based (Inertial Measurement Unit) human gait tracking system to obtain motion data from able-bodied people, in which we design an on-line joint angle estimation algorithm with adaptive alignment and drift rejection to address the existing problems in current tracking systems. We then construct an optimization-based gait generation scheme to obtain individualized human-like gait patterns depending on the walking behaviors from motion tracking. Combined with the gait generation method, we design a DMP-based (Dynamical Movement Primitive) trajectory planning and control strategy to achieve stable walking in a fixed gait pattern as well as continuous gait pattern regulation in various walking conditions. Finally, we apply the above methods on an exoskeleton and analyze the overall performance.