HUMAN CONTROL OF ROBOTIC MECHANISMS: MODELLING AND ASSESSMENT OF ASSISTIVE DEVICES

Abstract

The prescription and use of Assistive Technology, particularly teleprostheses, may be enhanced by the use of standard assessment techniques. For input devices, in particular, existing assessment studies, most of which are based on Fitts' Law, have produced contradictory results. This thesis has made contributions to these and related fields, particularly in the following four areas. Fitts' Law (and background information theory) is examined. The inability of this paradigm to match experimental results is noted and explained. Following a review of the contributing fields, a new method of assessing input devices is proposed, based on Fitts' Law, classical control and the concept of 'profiling'. To determine the suitability of the proposed method, it is applied to the results of over 2000 trials. The resulting analysis emphasises the importance of interaction effects and their influence on general comparison techniques for input devices. The process of verification has highlighted gain susceptability as a performance criterion which reflects user susceptability; a technique which may be particularly applicable to Assistive Technology.