Prediction Of Road-tire Adhesion Characteristics Based On HMM

The identification of pavement adhesion characteristics has always been the focus of research on active safety systems for automobiles.With the development of intelligent driving,the recognition requirements for pavement adhesion characteristics of intelligent vehicles are increasing,and it is necessary to predict the road-tire attachment characteristics in front accurately to improve the safety of the intelligent vehicles.In order to solve this problem,this paper haved used the front road image information and the road surface attachment characteristic recognition principle obtained by the intelligent vehicles to estimate the adhesion information of the planned driving area.Firstly,the texture and color features of the common road surface were analyzed,and common road surface images were processed and characterized.The grayscale co-occurrence matrix(GLCM)and HSV color space were used to extract the texture and color characteristic parameters of the typical pavement image and analyzed them.Then taking energy,entropy,moment of inertia,correlation four kinds of texture feature parameters and hue,saturation,brightness three color feature parameters to characterize typical road surface image features as an input parameter for Hidden Markov Model.Then,the principle of the traditional road surface adhesion characteristic estimation algorithm was analyzed.Six typical road surface ?-s curves are fitted using the Burckhardt tire-road model,and the relationship between common road surface and adhesion characteristics was obtained.Based on this,a new type of current pavement peak adhesion coefficient identification algorithm for different road surfaces was designed and carried out in Carsim/Simulink.The simulation results showed that the proposed identification algorithm can effectively estimate the given the peak adhesion coefficient hidden by different road surfaces.Finally,Hidden Markov Model was used to model the characteristics of typical road surface images and the adhesion characteristics of different road surface images.A model for predicting the road-tire adhesion characteristics was established,and the extracted road surface image characteristic parameters and different road surface adhesion coefficient values were used.The parameters wre trained and the model was continuously optimized.Finally,the accuracy of the established model was verified by inputting the test sample data.In order to further verify the feasibility and practicality of the road-tire adhesion characteristics prediction model established in the front road,using the intelligent driving platform of a bus company in Nanjing to conduct real vehicle experiments.The experimental results showed that the established HMM model can achieve an off-line recognition rate of more than 90% for six types of typical pavement road-tire attachment characteristics,and can accurately estimate the frontal adhesion characteristics in real-time on a sudden road surface test,and the model algorithm can be applied to the intelligent emergency braking algorithm of the intelligent vehicle brakes in advance when the road-tire attachment characteristic decreases,effectively shortens the braking distance,and verifies the feasibility of applying the proposed HMM model algorithm on an intelligent vehicle.