Research On Autonomous Driving Of Intelligent Automobile Based On Safe Neighborhood And Fuzzy Control

As an important component of ITS(intelligent transportation system), pilotless automobile technology has been paid more and more emphasis among research community recently, therefore, the research on pilotless automobile technology has important significance undoubtedly. The main work in this paper is the study of the intelligent vehicle automatic driving, and try to give an effective control method.According to the idea of granular computing and the method of neighborhood granulation, we firstly defined the concept of neighborhood information particles in two-dimensional plane, the method of road information state granulation by neighborhood information particles is also given. Furthermore, the concept of two neighborhood information particles vicinity is defined. Based on this concept, road is divided into a column of adjacent particles sequence, the automobile in the whole road driving control is transformed into the neighborhood driving control, the superposition of the overall effect can reach the final control target.Generally, the research on intelligent cars is based on vision at present. That is: computer vision technology is adopted to imitate mechanism of human vision model, the image seized by camera on the car will be disposed to gain information. Furthermore, relevant controlling strategies will be adopted to control the car. We will try to give the real aequilate road by two equidistant curves in two-dimensional plane, and then accurate mathematical model will be obtained. The vehicle will be abstracted to be a practical on the road,fuzzy multi-object optimization control will be adopted to realize automatically travel of the car. To solve the problem, this paper defined safe neighborhood of the road in two-dimensional plane. Various mathematical models of safe neighborhood has given. Furthermore, the searching method of sector safe neighborhood is given and sector safe neighborhood is adopted as the driving neighborhood of intelligent automobile. Multi-object problems of the whole model is analyzed and defined. And the method is obtained to confirm optimal driving neighborhood. Then two double-input single-output fuzzy controllers are designed. The rules of fuzzy controllers are obtained by solving multi-object model. As for different roads and different obstacles information, simulation experiments are made. And the results shows that controlling methods obtained in this paper can control the driving of automobile preferably and evade barriers smoothly. Finally, compared with other controlling methods, the controlling method put forward by this paper is efficient in overcoming the disadvantages of some controlling methods.