Study On Driver Modeling For Speed Control

Company with the convenience in the life and work which is brought by the vehicle technology, road safety becomes a more and more important social problem. Too many traffic accidents and too much loss of people and assets make us gradually recognize the importance of study on the automobile safety, especially the active safety. How to design the handling and stability in order to achieve the desired active safety has been one of the main researches for automobile technology in the last several decades. The closed-loop analysis is one of the most important methods for the research on the vehicle handling and stability performance, in this analysis the vehicle is seen as the part of being controlled, and the driver is the subsystem. So building a reasonable driver behavior model is the base of the research. And in some ways the driving auxiliary system can be seen as the simulation of the drivers. So the characteristic of the driver's behavior has already become one of the important research focuses. Thus, a driver model that can accurately describe the driver's performance is necessary for the closed-loop research on the automobile handling and stability.Consequently there stems out the goal of this paper—to establish an efficient and exact driver model which controls the longitudinal speed of vehicle. Then we combined this model with the directional driver model which was built by our group .So we built a new and comprehensive Driver Model and simulated the driver-vehicle closed-loop system based on the driver model and the vehicle model in CarSim. At last we did the research on its engineering application.So the whole work in this paper can be summarized into the following contents:Firstly, the method of Single Neural Adaptive PID Driver Model was introduced on the basis of the theory of PID Control and Neural Network, and a new Driver Model on the vehicle's longitudinal speed control was presented. A single neural adaptive PID controller can do the best description with the strong non-linear system that the vehicle's dynamic system is. After we analyzed the Preview-Follower Theory, the controller was applied on the vehicle's longitudinal speed control. So a new longitudinal Driver Model was presented. The vehicle model and the road sample used in the simulation were then introduced. Simulations show that the performance of the longitudinal Driver Model matches that of the human driver quite well.Secondly, a synthetical Driver Model was presented after we combined the longitudinal Driver Model with the directional Driver Model which was built by our group. The longitudinal and directional coupling was resolved: we have found the relationship of the longitudinal speed and the steady Gain. We spread the model that could be applied into the arbitrary path and the arbitrary changed velocity. The simulation validated the Driver Model comprehensive and integrated.Finally, because the auto-driving is needed, we developed the comprehensive Driver Model and a time optimization Driver Model was presented. We resolved the curvature and the maximal velocity on the arbitrary path. At last we did the closed-loop simulation using the optimization Driver Model and the Vehicle Model built in CarSim. The result could provide us correct guidance for driving.