Study On The Intelligent Vehicle Automatic Driving Method Based On Fuzzy Neural Network

Intelligent vehicle is a research frontier in the field of vehicle engineering and new directions for the future development of automobile industry.It is the key problem in the field of motion control that how to recognize the environment and achieve the automatic driving on the basis of acquiring information.As intelligent vehicle is an important part of intelligent transportation system(ITS),the research has important practical influence to improve the vehicle safety,energy saving and environmental protection performance,improve the efficiency of traffic system and reduce traffic accidents as well.The work of this paper can be concluded into the following two aspects:A method of road information identification based on a utomatic threshold,which achieved the corrected image distortion and transformation of image coordinates to world coordinates through camera calibration.The automatic threshold method based on O tsu method enhances the adaptability of machine vision system to environment.This work provided a basis for the following research.Combined with the advantages of fuzzy control and neural network,a fuzzy neural network intelligent vehicle autopilot controller is proposed.According to the results of visual identification,the controller extracts the lane line from the nearest distance of the vehicle as the reference lane.The multi-layer feed forward fuzzy neural network system based on T-S fuzzy model is constructed by taking the reference lane line slope and its distance from the vehicle as the input and the vehicle steering angle as output.Based on the artificial driving experience,the basic rules of fuzzy control and the initial parameters of the fuzzy neural network are formed,so as to improve the convergence speed of neural network.Taking the artificial driving behaviors as the simples,the back-propagation learning algorithm is used to adjust the network parameters,in order to further simulation of artificial driving behavior.Based on W ebots simulation software and unmanned vehicle,the simulation and experimental research are carried out.Compared with the single fuzzy controller and PID controller,the simulation results show that the algorithm has better adaptability and robustness.The actual vehicle experiment shows that in the case of visual conditions,the algorithm can simulate the driving experience of the artificial and has good ride comfort in both the corner and straight line.