3-D Dynamic Walking Trajectory Generation for a Bipedal Exoskeleton with Underactuated Legs: A Proof of Concept

This paper presents a framework to address three dimensional (3-D) dynamic walking for a bipedal exoskeleton with underactuated legs. To achieve this goal, the framework is constructed via a trajectory generator and an optimized inverse kinematics algorithm that can cope with underactuation. In order to feasibly attain task velocities with underactuated legs, the inverse kinematics algorithm makes use of a task prioritization method via the exploitation of null space. In doing so, the tasks with lower priority, e.g., swing foot orientation, are attained as much as possible without disrupting the higher priority tasks, such as CoM trajectory. Meanwhile, the trajectory generator utilizes the ZMP concept analytically and ensures the acceleration continuity throughout the whole walking period, regardless of the contact and phase changes. The proposed method is verified via a lumped human-bipedal exoskeleton model that is developed and simulated in MSC.ADAMS simulation environment. As a result, we obtained feasible and dynamically balanced 3-D walking motion, in which no oblique foot landing or exaggerated torso orientation variations were observed, despite the underactuated nature of the robot legs.