Optimized Automatic Lane Assignment In Urban Environment

Nowadays, with the continuously increasing vehicle amount and traffic load, how to fully utilize the road facility to improve the performance of the entire traffic system has raised more and more attention. This dissertation focuses on the feasibility of how to utilize the lateral space, for example assign a specific proper traffic lane for each vehicle travelling on the road, to improve the global traffic efficiency. Corresponding methods and system prototypes are also provided.In order to find out how different lane allocation systems impact on the road traffic, the dissertation start from understanding the driver behavior. A driver behavior model which mainly focuses on the decision model of driver lane changing strategy on single road section is rolled out. In order to validate the model, the author collected some data about lane changing from real traffic scenario at first. On the other hand, for observing and validating different lane allocation algorithms and driver behavior model, a reliable and flexible traffic simulator is required. Thus, the author developed a simulator which bases on the above driver behavior model together with the model of vehicle and road environment. The simulator can reproduce the traffic scenario and also helps to verify the correctness of the driver behavior model by comparing the simulation results with the data collected from the real world.After understanding the driver behavior in the traffic system, analyzing the different actions a driver would make in different situation, the author found there are several human driving behaviors can be improved. Start from these points, the dissertation explored how to improve the traffic performance in different scenario and provided some methods. On top of that, the author proposed a general phase based lane allocation algorithm and a centralized control system, which can be applied on automatic vehicles and covers almost all the traffic scenarios. And the effectiveness of the system is validated by different simulators includes the above simulator with new driver behavior model, the problem and future work direction is also identified.This dissertation mainly analyses how to improve traffic performance by properly allocate each vehicle on a specific lane on a road section, and provides a general control algorithm which can be also used on automatic vehicles. In this research, the author has a deep dive in the driver behavior, and set up a new model to understand human driver and also compare with different automatic lane allocation system. Meanwhile, the author also built a simulator to validate the driver behavior model and lane allocation system, which overcomes the shortage which a prototypic system is not likely to be tested in real traffic situation and on real vehicles. The lane allocation system can significantly improve the traffic performance in most traffic scenarios, which proves the feasibility and usage of dynamic lane allocation system.