On a Moving Target Selection Model in Virtual Reality Based on Decision Trees

Abstract

Virtual reality (VR) systems have been used in various industries. Highly effective humancomputer interaction (HCI) and natural HCI experience have become the key indicators for evaluating a VR system, where target selection is the key for interaction efficiency and experience. In this paper, we propose a moving target selection prediction model, based on the probabilistic Fitts's Law and in combination with decision trees, for moving target selection in VR systems. Firstly, we verified the feasibility of predicting the user intention based on the size and distance of moving targets in VR scenarios through two selection task experiments with a sphere as the target. Then, also through two experiments, we proposed an improved moving target prediction model by factoring in head posture with target size and distance and taking into account the influence of head orientation. Based on the decision tree algorithm, we calculated its prediction accuracy and compared it with the distance scoring function. The results show that the improved prediction model has significantly better accuracy and can accurately predict the user's moving target selection intention in a VR system.