An Online Transition of Speed-dependent Reference Joint Trajectories for Robotic Gait Training

Rehabilitation robots reduce the physical burden on therapists, quantify training and allow greater dose of therapy on individuals with neurological impairments. Robots are also capable of precisely customizing therapy based on the user's physiology and/or needs, for example, customizing a reference trajectory for gait training. While a number of methods for obtaining reference gait patterns have been proposed, these approaches lack the ability of altering the trajectories according to the varying walking speed in real-time. The objective of this paper is to develop an online algorithm that can provide a continuous, speed-dependent reference gait pattern for robotic gait training. We employed Fourier series and profile blending methods to generate natural transitions in gait patterns, and synchronized the gait cycle time according to the given arbitrary walking speed. The simulation results suggest that the algorithm can stably change the gait patterns with the given walking speed in a synchronous manner. We conclude that the method can provide online speed-dependent walking motion that can be used for general robotic gait training applications.