Research On Driver's Starting Intention Recognition Method

The driver's starting intention identification can predict the driving behavior and driving rules of the driver for a period of time,predict the operation of the driver in the next step,and warn the driver of the dangerous behavior.In recent years,major auto manufacturers have begun to research advanced intelligent identification technology.At present,the commonly used methods for driver's starting intention identification include fuzzy reasoning and support vector machine classifier.The above algorithms have problems such as long recognition time,low accuracy and poor reliability in the recognition process.Comparing the advantages and disadvantages of the two algorithms,this paper proposes a method based on fuzzy reasoning-SVM cascading algorithm to identify the driver's starting intention.This algorithm has certain advantages in processing large sample data,and has the advantages of high accuracy and short recognition time.On the basis of analyzing and summarizing domestic and foreign intention recognition results and data collection methods,firstly,the laboratory data acquisition platform and real vehicle data collection scheme are designed.The hardware and software of the laboratory data acquisition platform are introduced.The data is collected in the laboratory using the electronic accelerator pedal,the acquisition experiment box and the host computer.The actual vehicle data is collected and analyzed using the CAN-OBD tool.Secondly,the fuzzy inference algorithm and the support vector machine algorithm are used to identify the driver's starting intention.In the fuzzy reasoning,the fuzzy subset and membership function are selected,and the fuzzy control rules are formulated.The Simulink fuzzy inference model is constructed to classify and identify the driver's intention.In the support vector machine algorithm,the genetic algorithm toolbox is used to optimize the penalty factor C and the kernel parameter g,and the support vector machine classifier is used to identify the driver's starting intention.The two algorithms were trained and tested by laboratory data and real vehicle data respectively,and the corresponding experimental results were obtained.Finally,a Fuzzy inference-SVM cascade algorithm for driver's starting intention identification is proposed.The principle is to input all sample data into the first layer Fuzzy inference model for classification and identification,select the confusing intention,and import it into the second layer support vector machine(SVM)classifier for intent classification identification.The cascading algorithm is verified separately using laboratory data and real vehicle data.The experimental results show that the Fuzzy reasoning-SVM cascading algorithm takes less time and the recognition accuracy is higher than that of a single fuzzy inference algorithm or support vector machine algorithm.The research on the driver's starting intention identification method can not only provide favorable basis and support for the optimal control of the following automobile clutch,optimize the shift curve and driving style recognition,but also help to realize the vehicle assisted driving and intelligent driving,which has certain research significance.