A Lateral Control Method Of Intelligent Vehicle Based On Fuzzy Neural Network

Intelligent vehicles played such a significant role in scientific researches, transportation, and military affairs, especially in Intelligent Transportation Systems. Intelligent vehicle is a complex system which combined electron, sensor, and image processing technology. Application of intelligent vehicles can highly reduce traffic accident rate, and improve transportation efficiency simultaneously. This paper is developed based on visual navigational intelligent vehicles; vehicles control and path tracing will be discussed.Vehicle dynamics model is the kernel of vehicle movement simulation. Considering the importance of intelligent vehicle dynamics model, vehicle dynamics models will be built in two different ways. Firstly, simplifying the vehicle models and building an 3-DOF vehicle dynamics mathematical model. Secondly, building a lateral control model as intelligent vehicles’lateral control method is the major research method in this paper. Thirdly, in order to demonstrate the effectiveness in a more realistic environment, another vehicle dynamics model needs to be built by using multibody dynamics simulation software. At the same time, a 3D terrain model is needed to final construct a simulation environment. Under such environment, real time revision of vehicles’kinetic parameters and man-machine interaction can be achieved; it increased the openness and extensibility of the simulation environment.The analysis is carried out under the framework of fuzzy control and neural network considering their different characteristics. Logic reasoning capability of fuzzy control and self-learning, self-adjusting capability of neural network were adequately developed. Based on fuzzy neural network, intelligent vehicle lateral control system was established. The system covered intelligent vehicle model, image processing module, deviation fusion module, and fuzzy neural network lateral controller. The function of image processing module was to obtain intelligent vehicles lateral error and orientation error. Deviation fusion mainly considered the influence of input dimension towards lateral controller, and input of lateral controller was formed by integrated error after fusion and changing rate. In terms of fuzzy neural network lateral controller building, the first step is to build fuzzy neural network structure on account of T-S model by using experimental data oriented fuzzy C-Means Clustering Algorithm, stated the network functions of each layer, and determined the excitation function network should adopt. Next, adjusting the parameters in network by using bp neural networks learning algorithm, and finally optimize fuzzy neural network membership function and fuzzy control rules. Two simulation experiments analyses were carried out correspond to previous vehicle dynamics mathematical model and vehicle multi-body dynamics model respectively. In the first simulation experiment, straight path operating condition and multi-curvature complex path operation condition were established separately to validate controller’s path tracking control. The results demonstrated that the controller had excellent performance in timeliness, smoothness, and robustness. In the second simulation experiment, two path operating condition and 3D terrain environment were established. The results demonstrated that such controlling method had a good control effect.