Active ankle-foot orthoses (AAFOs) assist a user unable to dorsiflex the ankle joint. The optimal AAFO is lightweight and aids the user in achieving a nominal gait.

This thesis presents the design and control of a novel discrete nonlinear series elastic actuator for AAFOs. The actuator uses regenerative braking to reduce the peak mechanical input power to the AAFO by 77.2%.

To control the designed actuator and AAFO, two methods are proposed. First, an adaptive extended state observer is developed to reject disturbance and estimate the error in the dynamics model. Second, a nonlinear model predictive controller is presented to track the ankle trajectory while reducing assistance to the user.

The adaptive extended state observer can identify the AAFO and ankle joint parameters. The nonlinear model predictive controller with the adaptive extended state observer reduced the trajectory tracking error by up to 58.3%.