Road Information Perception For Autonomous Vehicle

The autonomous vehicle has to know a completed perception of the surrounding environmental information during the driving.And the road information is one of the important information.This paper first presents the machine learning based method using the accelerometer to collect the vibration information of vehicle suspension.The features of the power spectral density of the profile of the road is extracted.Then the texture features extracted from the road terrain image data leads to the higher classification accuracy and robustness.Furthermore,this paper proposes a method based on environment field to describe the vehicle's environment to detect the drivable area.In the space environment around the vehicle,the point cloud data of the LIDAR are loaded into the space units of a novel spider grid map,respectively.Then the field values of each space unit are calculated.Finally,the connected correlation calculation is carried out to obtain the traveling area.The experiments showed that the proposed method described the road environment information around autonomous vehicle fully and provided clear and efficient detection information for the vehicle motion control stage.