Simulating Distracted Drivers Using IDM and Real-World Driver Properties

Sammanfattning

Traffic simulation is a useful tool when it comes to examining flaws in existing infrastructure or testing and improving projects that are still being planned. However, one factor that is seldom included in these simulations is distracted drivers and the problems they may cause. This report aims to examine the feasibility of adding distracted drivers to existing simulation tools by simply modifying parameter values of the IDM car-following model. This would enable engineers working with traffic simulations, and using IDM, to consider the effect of distracted drivers without having to use new tools and would also be compatible with existing simulation scenarios. This was done by collecting a set of parameter values for both distracted and focused drivers based on studies on real-world driver behaviour. After that the model was tested in two scenarios, one for safety and one for efficiency, with differing proportions of focused and distracted drivers. The performance in the safety scenario was evaluated based on MTTC and DRAC. In the efficiency scenario average speed and average flow were used. The results from the simulations were later compared to previous studies to determine how accurate the model is. The results of the safety scenario were not promising. Both the MTTC and DRAC values indicated that the distracted drivers were safer than the focused ones. The efficiency indicators were more promising and followed trends found in previous studies. This could mean that there are a few use cases of the model where only efficiency is of interest but for most purposes the model can not be expected to accurately simulate distracted drivers. This however does not mean that using IDM to simulate distracted drivers is not possible, only that our naive approach does not work. A more promising method would be using maximum likelihood method or something similar to fit the parameters to trajectory data collected from real drivers.