Abstract
We developed a novel control architecture for robotic leg prostheses that adapts online at each step to provide biological accurate function based on the user gait pattern. The user-adaptive control can restore physiological joint energetics and does not require any tuning–the user can don the robotic leg and go. The movement of the prosthesis during the swing phase is adapted to the user online using the minimum-jerk principle. Minimum-jerk provides an optimal way to program a smooth movement, allowing amputees to obtain physiological gait symmetry at variable cadences, and, potentially, to handle unexpected perturbations such as tripping.