Research And Application Of Automobile Driving Training System

In recent years,the number of cars in our country has been growing,Along with this is the increase in the number of car drivers.Taking the test for a driver’s license is inevitable for people.However,in the process of driving school learning,on the one hand,there are few coaches and many students,which leads to that some students cannot be taken care of by the coach when practicing cars,which directly affects the training effect and license obtaining speed of students;On the other hand,students cannot intuitively understand the training skills taught by the coach.Therefore,It is very necessary to study the driving assistant system of the training car using machine vision.The main research contents of this thesis are:(1)This thesis provides an overall design idea for the research of the driving assistant system.This design idea describes two parts in detail,the system interface design part and the system functional structure building part.(2)In order to accurately identify the initial position of the training car,this thesis selects SURF algorithm to identify the initial position of the training car by comparing ORB algorithm,SIFT algorithm and SURF algorithm.The image matching experiments were carried out on the three algorithms under different light intensities,and the matching accuracy and matching speed were taken as the evaluation indicators of the experiment.The results showed that SURF algorithm had higher matching accuracy and faster matching speed.(3)Because the tracking box of the fDsst tracking algorithm does not have an inclination angle,it is unable to accurately locate the position of the training car.In order to better realize the dynamic tracking of the training car,this thesis proposes an improved fDsst algorithm.The algorithm combines contour detection algorithm and image processing algorithm on the basis of fDsst algorithm.It is tested on self-made video sequences and compared with Boostingbox,KCFbox,MILbox and other algorithms.The results show that the algorithm has higher tracking success rate and tracking accuracy,and better tracking effect.(4)In order to realize the path planning of the five items in Subject 2,the vehicle parameters are selected according to the actual vehicle,the coordinate system is established,and the vehicle kinematics model is established.The model completes path planning through two constraints.The two constraints condition include: vehicle kinematics constraints and the kinematic constraints that the planned path curve should meet.MATLAB is used to simulate and verify the paths of the five venue,and the feasibility analysis of the path simulation renderings is carried out to further verify the rationality of the path design.