This master thesis describes the initial design and implementation of a controller for an assisting upper body exoskeleton with two degrees of freedom; elbow- and shoulder flexion/extension. This project limits itself to only implementing control on the elbow-joint.
The thesis includes the derivation, parameterization and validation of a dynamic model describing the exoskeleton arm.
Further a model of the human torque output in the elbow joint is derived. The joint torque is modeled using the Hill Muscle Model, in conjunction with electromyography and joint angle measurements.
After an experimental comparison of different controller structures, a controller is implemented using a hierarchical structure that separates exoskeleton controller design into three layers. One that directly controls the actuators, one that handles the interaction between user and exoskeleton, and one that tracks the intention of the user. The three layers are implemented as computed torque control, admittance control and joint torque estimation, respectively.
Quantitative tests have not been performed, but qualitative tests indicate that the implemented controller feels natural, and the user reported that the controller was assisting when lifting a dumbbell disc.
This thesis can be used as basis for further research intro electromyography-based upper body exoskeleton control, as the methods presented also are applicable in estimating and controlling other joints as well.