Lidar Based Recognition Of Uneven Road Surface In Front Of Intelligent Vehicle

The trend of intellectualization in automobile is becoming a major study point nowadays.The intelligent vehicle with advanced sensors gives a new opportunity to the industries related to automobile.Most of the studies about intelligent vehicles focusing on the problems of safety,that’s about avoiding collision with pedestrian and other vehicles.However,the problems of automobile comfortableness are seldom mentioned.The starting point of this paper is to solve the problem of ride comfort which is seldom concerned in the current research of intelligent vehicles,that is,how to ensure the rider’s comfort.In this paper,a method of recognizing the road ahead of intelligent vehicle,based on lidar,is designed to serve the suspension control for vehicle ride comfort,using which,the situation of the road ahead of vehicle can be perceived before passing through.The main contents of this paper are divided into three parts: the construction of hardware and software platform,the method to extract road condition from lidar pointcloud,and optimization of the computing time.Firstly,according to the target of the scheme,the lidar is chosen as the main sensor.To realize real-time suspension control and get satisfied road recognition results,the way to fix lidar on the car is determined.In order to accurately describe the road surface in front of intelligent vehicle,different coordinate systems are established,and a new method of coordinate calibration of lidar is presented.A hardware basis for the development of software system is established,which can also be used in the experiment to test basic functions.Aiming at the problem of extracting road condition information from pointcloud signal of lidar,a complete signal processing flow is presented in this paper,which can extract geometric description of the road surface unevenness factors from original pointcloud data.At the same time,in order to solve the problem of insufficient information of lidar in its remote part of measurement results,an improved ICP matching algorithm is used to fuse the measurement information of two frames of lidar,which improves the information quantity.In order to solve the low calculation speed problem of the fusion algorithm,the road surface pre-judge method are introduced.Only when the road surface is uneven,the two frames of lidar information are fused.In the fourth chapter,a coarse-fine quadratic matching method is proposed to further improve the speed and effect of the fusion algorithm.In this method,the sensor signal of the vehicle itself is introduced into the recognition system through CAN bus,and a state observer to estimate the vehicle position and attitude is designed to estimate vehicle position and attitude,which realized the fusion of vehicle-borne signal and lidar signal.