Effects of target size and distance on kinematic endpoint prediction

Abstract

Because of the ubiquity of the WIMP paradigm, many researchers seek to design new pointing facilitation techniques for Fitts-style pointing tasks. However, many of these pointing facilitation techniques make one of two simplifying assumptions: either salient targets are sparsely placed on the display, or there exists some ability to identify the endpoint, the target, of a user's movement in real time. In this paper we extend previous work on kinematic endpoint prediction (KEP), a technique that uses models of user motion to predict endpoint in Fitts-style pointing tasks. We introduce a simplified algorithm to predict user endpoint. We present a technique to measure the numerical stability of endpoint predictions in real time. We show that the distance of motion has a significant effect on predictor accuracy. Finally, we develop an accurate understanding of the relationship between movement distance and predictor accuracy and show how we can use this understanding to infer accurate, real-time probability distributions on target sets within an interface. Together, these results allow KEP to be applied in new and novel ways to pointing facilitation techniques.