Content area
Abstract
In order to facilitate more natural and intuitive interaction for human users, robots need to move in a more human-like manner as compared to current robots. This change would enable humans to better anticipate robot movements (which would allow humans to better avoid collisions if necessary) and also improve safety in the context of a collision between a robot and a human.
The goal of this thesis was to analyze experimental data of human motion to gain an understanding of how human motion and robot motion differ. From this understanding, a neuro-motor model (NMM) of a human elbow (previously established by Beardsley et al.) was augmented by the addition of a variable stiffness quality. The work in this thesis developed and tested a predictive stiffness model that attempts to recreate the stiffness values used by humans in the context of a disturbance rejection task.
