Implementation of artificial intelligence to improve novice drivers' hazard perception skills

Abstract

Commercial off-the-shelf video games, such as Grand Theft Auto 5, bring in hundreds of thousands of sales within the first few days. Why not a teenagers' interest in a video game for more educational purposes? Simulation programs are often expensive and rare, but serious games can be made easily available and open for the public. An educational driving simulator in the form of a video game poses an interesting opportunity to reach teenagers, by combining flashy graphics and friendly competition with real-world road hazards. In an attempt to create and test such a program, our team designed artificial vehicles in a driving simulator that generate random traffic and attempt to crash the player vehicle. In this single player game, the player must maneuver the player vehicle to a mission objective. Each of the three players in this study were thirteen year old male students with absolutely no driving experience. These students completed a pre-test, two ten minute missions, and a post-test. Using eye tracker software to analyze horizontal scanning, there did not seem to be significant results to indicate an improvement in hazard perception. However, each student reported an interest in future participation. Therefore, it is recommended that future studies on the current project are continued, with the appropriate software upgrades needed to allow students to operate the game at a normal frame rate (without lag).