Simulation of a passive knee exoskeleton for vertical jump using optimal control.

Research on exoskeletons designed to augment human activities and the attendant exoskeleton industry are both rapidly growing areas of endeavor. However, progress in the field is currently being hindered by a lack of understanding of human-exoskeleton interactions. At present, the main method applied to reach such an understanding is to build and test prototypes or end-effectors (that simulate the devices), but this is a very time-consuming and costly process. In this study, we aimed to address this problem by simulating passive exoskeleton-human interactions during a vertical jump. The simulation is based on theoretical and empirical models. Using the simulation, we performed a numerical optimization procedure to determine the muscle excitations and starting postures that would give the maximum jump height. The simulation used a planar 4-DOF dynamic model. The muscles at the joints were modeled as torque actuators, with a flexor and an extensor for each joint and passive torque representing the tendon and muscle properties. We then simulated jumps with a passive knee exoskeleton with five different values of stiffness with the aim to study their effect on the jump height. The optimal excitation for the maximum jump height was found by using a genetic algorithm (GA). To improve our optimization performance and to test the convergence of the GA, the GA optimization was performed several times. For each exoskeleton condition, the GA found the optimal jump more than 400 times, and out of these solutions the one that achieved the highest jump was chosen. The result revealed an increase in jump height as the spring became stiffer. In addition, it was found that the energy that was stored in the spring of the exoskeleton was not fully converted to jump height.