The Research Of Path Recognition And Control Strategy For Intelligent Car

With the improvement of the automobile industrial technology, the expectation on Intelligent navigation of automobile become higher and higher.Now, the navigation with road is one important direction of intelligent navigation. This paper is based on intelligent car model which is used in intelligent car match and research the car in the terms of navigation with road to realize the aim that intelligent car recognize the road automaticly.The main idea of road recognition is following:First, choose and design a set of image preprocessing methods which is valid for road. The preprocessing methods contains deal ing with redundancy information and removing noise and so on . Then ,in terms of practical experience of drive,the paper classify the road of intelligent car into six types.According to the road types,the characters of road image are getted and the samples space is established. Then, by training and studying, eshtablish BP neural frame to configure relationship between input character and specific road type,so as to reach the aim that recognize road type automatively. At last, According to differient road, different control method should be applied for the intelligent car,at the same time,introducing Fuzzy_PI to regulate car speed,make the speed respond quickly and smoothly.This realization mainly includes 4 steps:1. Road image processingGetting image information is the premise of recognizing road, the paper considers several conditions that maybe occurred in the car running,.The conditions are followed: the indoor daylight environment, indoor light environment and indoor flashlight environment. By using the method of median filter, threshole segment, image slender and connect of break point, the continuous line can be get and the thrend information of road is picked out2. Getting character from road image.The paper classify roads into six typs,they are straight road,ordinary turning , big turning,turning entrance, turning exist and urgency turning.In terms of six characters of road, analyse six character of road.The character is described in the following part:(1) Character1 reflects the landscape span between start and end point.(2) Character2 reflects the portrait span between start and end point.(3) Character3 reflects the max span between the whole line. (4) Character4 reflects the position of middle line of start and end point.(5) Character5 reflects whether there are point of intersection between guide line and middle line of image.(6) Character6 reflect the basic shape of road.Analyse six type of road quantificationally at the base of characters and establish road samples space to use as input of classifier in the next part.3. Designing classifier of roadUsing all the road characters as input of classifier and specific road type value as output, design and train BP network using C. Then, test the network above to see whether it is practical.(1) Put the entire sample in the samples space into BP neural network to train the configuration of network, change the right value and reconfigure network until the error meet demands so as to establish the network configuration that reflect the corresponding relationship between input characters and output type.(2) Use road sample in random to test the network above. Comparing the output value of network and the value desired to test the accuracy of BP neural network.the result show that the output of network is approximately equal to specific value, which means that the neural network is valid and can be used in intelligent car for road recognition.4. Strategy of controllingAccording to six types of road, different strategy of angle and speed should be given,at the same time ,in order to react speed quickly and smoothly,the paper design the fuzzy_PI controller to promote the ablity of speed reaction.The above research provides the theory base for road image processing, road recognition and speed control method and provides theory reference for practice.