Safety evaluation of heterogeneous traffic: Experiments using different models in SUMO
Ladda ner
Typ
Examensarbete för masterexamen
Program
Automotive engineering (MPAUT), MSc
Publicerad
2020
Författare
Xiao, Weicheng
Modellbyggare
Tidskriftstitel
ISSN
Volymtitel
Utgivare
Sammanfattning
With the development of self-driving technology, the day when autonomous vehicles share roads with traditional human-operated vehicles seems to be around the corner. This makes the safety evaluation of this so-called heterogeneous traffic particularly important.
In this thesis, by conducting microscopic heterogeneous traffic simulation on the simulation platform SUMO, the impact of autonomous vehicles on traffic safety and efficiency is studied. In order to obtain more accurate results, the initial car-following model and lane-changing model in SUMO need to be modified and calibrated using real-world data before being implemented in the simulation.
As some previous research have shown, the types of vehicles involved on driving situation have an impact on drivers’ driving behaviours. The neglect of this impact has led to errors when reproducing the realistic driving behaviour with the existing car-following and lane-changing models. In this thesis, the models are modified by setting appropriate value for some related parameters to reflect this impact. Then the models are calibrated using the data extracted from highD dataset. Three performance indicators, namely number of conflicts, number of lane-changing and a speed performance indicator,are proposed to rate the error in terms of car-following, lane-changing and safety aspect. After the calibration, the best set of parameter values is selected and used to represent those human-operated vehicles in the heterogeneous traffic simulation. As for the autonomous vehicles, both zero-error Intelligent Driver Model (IDM) and Cooperative Adaptive Cruise Control (CACC) model are used to present two different types of autonomous vehicles.
After getting all the models needed, the heterogeneous traffic simulation is conducted in SUMO. Several indicators, such as time to collision and number of lane-changing, etc., are used to evaluate the safety and efficiency of traffic. The results are different when using different autonomous car models. For the zero-error IDM case, the results show that traffic safety and efficiency increase as the penetration rate of autonomous vehicles increases. For the CACC model, the traffic efficiency increases with the increase in the penetration rate, but the traffic safety deteriorates when the penetration rate is low, and it slowly improves only after the penetration rate is higher than 0.5. The simulation results help to understand the impact that autonomous vehicles will bring on heterogeneous traffic.
Beskrivning
Ämne/nyckelord
heterogeneous traffic , microscopic simulation , intelligent driver model , highD dataset , traffic safety evaluation , mixed traffic