Optimal Trajectory Planning for Human-Like Energy Efficient Motion of Lower Limb Exoskeletons

Motion planning is a critical aspect of lower-limb exoskeleton control. Conventional motion planning approaches compute reference trajectories in task space, leading to infeasible or uncomfortable joint level motions. Conversely, joint space planning methods ensure feasibility but may fail to reproduce natural human gait, limiting their effectiveness in rehabilitation. This paper introduces a novel trajectory planning method that generates human-like gaits while guaranteeing feasible joint paths. A set of task space waypoints is generated for a personalized modulation of gait characteristics. The waypoints are mapped in joint space and serve as boundary conditions for a multi-joint optimization problem that leverages Bézier curves and minimizes joint jerk. The resulting trajectories exhibit a high degree of similarity to natural human gait. Experimental testing emphasizes that the proposed method requires 50% lower maximum hip velocity and 38% lower maximum knee velocity to walk at a walking speed of 0.41 m/s compared to a previous taskspace method. Furthermore, the exoskeleton's average electrical power consumption over a stride is reduced by 34% and peak electrical power by 61%. The results demonstrate that the proposed approach enables exoskeletons to walk for longer times, improving the usability of the system for extended use.