Authors: Wimal Perera, Prasad Wimalaratne
Building a navigation system within a 3-dimensional virtual computer world is challenging. Further, building our navigable virtual world corresponding to real world facts (such as a map or a landscape of a given area) magnifies this challenge. This research investigates on possibilities of modeling a 3D navigable world corresponding to real world facts via mathematical constructs derived using the provided information of a given area. Thereafter, this realism augmented 3D navigation system is transformed into a 3D virtual driving simulator. Our driving simulator is used as a tool for measuring driver behaviour under different types of varied conditions. The final output of this simulator after a single driving session is a sample data set indicating predefined metrics; those can be used to state observations for a typical human factors experiment. Hence the observations after a driving session in a simulated environment are used to derive useful inferences with respect to real driving. The metrics include; number of wall collisions, number of traffic cone collisions, time taken per driving session and speed detected during different times (can be used to obtain a speed graph later). The sample human factors experiment tested on our driving simulator is about inferring the relationship between driver behaviour vs. consumption of alcohol, which is simplified to driver behaviour vs. drowsiness for the sake of demonstration.