A compliant semi-autonomous reactive control architecture applied to robotic holonomic wheelchairs

Abstract

This paper presents a novel intelligent control architecture for semi-autonomous systems. This work has been motivated by the need for a control architecture that can differentiate between users of different perceptual and cognitive capabilities while interacting with semi-autonomous machines. A good demonstrative example for this control system is a robotic wheelchair. The proposed architecture evolves around three variables: degree of autonomy to reflect the user's capabilities, user's level of confidence in commanding the machine, and strength of conflict between the user's command and the machine's autonomous response. A method is proposed to identify the user capabilities and accordingly to set the degree of autonomy. The analogy between this architecture and horseback riding is demonstrated.