DMP-Based Motion Generation for a Walking Exoskeleton Robot Using Divergent Component of Motion

Abstract

The planning of footsteps has proven to be a difficult task for researchers working on walking exoskeleton robots. Over the past years, researchers have proposed many methods. In addition, Some researchers have payed attention to linear inverted pendulum (LIP) model and its divergent component of motion (DCM). In the meantime, dynamic movement primitives (DMPs) has also been applied to solve this motion planning problem and attracting a lot of attention. This paper describes a gait planning strategy using DCM and DMPs. We design a two-level planning in the proposed strategy. In the first level, we calculate the center of mass(COM) using LIP Model and DCM. Then we use the inverse kinematics of the walking exoskeleton robot to solve the joint trajectories. In the second level, we use gaussian mixture regression(GMR) represent the joint trajectories. And we learn and reproduce the joint trajectory with DMPs. Finally, we validate the feasibility of our approach on walking exoskeleton robot.