Research On Highly Detailed Urban Traffic Map For The Autonomous Vehicle

For autonomous vehicles in urban traffic environment,a map which can fully represent the traffic information is indispensable.The common commercial navigation map can provide the traffic topological information for drivers.However,it is not enough for the autonomous vehicle.The autonomous vehicle need not only the traffic topological information,but also the detailed information in some complex environments.Based on these requirements,this thesis makes an effort on highly detailed urban traffic map and its corresponding applications.The main achievements are as follows:1.This thesis designs a novel method to construct the highly detailed urban traffic map,including map labeling,database creating and database restoring.Using this method,we had finished many database for different highly detailed map,including the campus,urban area and highway.This method has been proved to be available.2.This thesis proposes a multi-class path planning method based on highly detailed map,and a dynamic planning method which fuses commercial map and highly detailed map.This method can provide lanemarking data for navigation,and has been proved to be real time and reliable by lots of experiment.3.This thesis proposes a lateral positioning method fusing the lanemarking information in highly detailed map and the result of lanemarking detector for the autonomous vehicle.This method has been proved by the quantitative and qualitative experiments that it can efficiently decrease the lateral positioning error.4.This thesis proposes a method to predict the azimuth of the dynamic vehicles based on Bayes theory.This method extracts the traffic regulation from highly detailed map,and use it to estimate the motion of vehicle,which improves the accuracy of motion estimation.The achievements have been used in the autonomous vehicle platform of our laboratory and get a good result.They can provide important support for the autonomous vehicle in complex environment,and will be also used in the China Future Challenge this year.