Developing intelligent agents for traffic education

This study includes the developments we made with the goal of using a driving simulator in developing the hazard perception skills of novice drivers. This study has been specifically oriented to create hazardous traffic situations in the simulator, which are common in Turkish urban traffic environment and let the novice drivers to practice them in order to improve their hazard perception skills. For the purpose of this study, we developed a driver behaviour model, which emulates aberrant driving behaviours and driving styles (i.e. behaviours) for different categories of driver. The model was developed based on a two layer Hierarchical Concurrent State Machines (HCSM) modelling framework. A graphical user interface (GUI) software is developed by using Microsoft Visual Studio programming framework in order to prepare driving style configuration files. In our study, we used TRAFİKENT driving simulator as a test-bed. Validation experiments and evaluations have demonstrated satisfactory results regarding developed models. It is concluded that, the presented system can contribute to improve the hazard perception skills of novice drivers and to a some amount decrease the number of traffic accidents that occurs in our country.