A Study of Realization of Energy Optimal Gait Motion of Exoskeleton Robot for Paraplegic Patients

Abstract

To regain the ambulation ability of the paraplegic patients, a number of overground gait trainer type lower-limb wearable rehabilitation robots have been developed. These robots consist of similar components such as electrical motors, control board, battery and crutches, and they provide a gait motion to the wearer. Although these robots have been developed and improved for the last several years, papers and documents discussing on how these robots generate their gait motion are rarely found. Since the physical status of the wearer differs and the robot portably operates on the ground with its onboard battery module, it is necessary to generate an energy optimal gait trajectory based on the given dynamic conditions of the robot. In this paper, a method to generate an energy optimal gait trajectory of exoskeleton robot using genetic algorithm is proposed. To do this, the optimization problem is formulated. The design variable of the optimization problem is the set of angular positions that represent the joint trajectories of half gait cycle. The cost function is set as a sum of square of joint torques to represent the energy consumption during one step gait motion. Constraints such as double stance pose at the initial and final phase of the gait motion, joint angle limitations and the condition of the swing leg foot staying in the air during the gait are provided. By simulating robotic gait motion and evaluating the result of the optimization process, it is found that the energy optimal gait trajectory of exoskeleton robot can be theoretically generated. To verify the effectiveness of the proposed optimization process, the experiment is conducted by implementing two gait trajectories, normal human gait trajectory and optimized gait trajectory, to the actual lower-limb exoskeleton robot. By comparing the average sum of square of currents delivered to the electrical motors on the robot per one step gait motions based on two different gait trajectories, the decrease on the energy consumption is verified.