Predicting user performance time for hand gesture interfaces

Abstract

User interfaces based on touchless hand gestures have advantages over conventional user interfaces in a variety of scenarios. However, they still have challenging problems to be researched, such as the design and evaluation of them in order to obtain satisfactory results. The classical approach of involving users to choose gestures or analyze interface designs needs to be complemented with predictive evaluations for cases in which those user-based methods are inapplicable or expensive to do. Thus quantitative user models are needed to perform those evaluations. THGLM is a model based on KLM and gesture units, but its first formulation needs to be improved. This paper completes the model by analyzing its performance in several user studies. In particular, we found out that THGLM forecasts performance time in doing tasks on Uls based on touchless hand gestures (THG) in an acceptable way (prediction error = 12%, R-2 > 0.9). The paper also reports a study concerning the model utility to analyze and compare interface designs. Moreover, the model utility was confirmed by independent designers who were invited to participate in a study. Finally, the initial model was extended by introducing several new operators. As a conclusion, the present model has some intrinsic limitations which are discussed, but the results confirm the general hypothesis that it can be used to analyze Uls based on touchless hand gestures.

Relevance to industry: THGLM should become a useful tool for UI designers to perform usability assessments, improve interface designs, and develop good software applications using THG. This is especially useful in situations where it is difficult to conduct tests with users or as a preliminary step in the process of developing software. (C) 2017 Elsevier B.V. All rights reserved.