Road Scene Understanding Based On Sensor Data Fusion

Intelligent driving of vehicles is the direction of future traffic development.Autonomous vehicles is a future travel tool for liberating drivers and improving traffic efficiency,among which sensors such as millimeter-wave radar and cameras are an important part of their underlying support.Based on the fusion of radar data and image data,this paper studies the front target as the research object and provides accurate,reliable and robust front environment perception information for the path planning system of autonomous vehicles.It is of great significance.This paper studies three aspects of road scene understanding: predicting the main direction of the road,correcting the position of the target,predicting the moving direction and behavior of the target.The main research work of this paper includes:1.The advantages and disadvantages of millimeter wave radar,camera and lidar are compared and analyzed.The principle and algorithm of acquiring target information by millimeter wave radar are studied.The algorithm of road guardrail and road edge detection based on millimeter wave radar data is studied,and the main direction of road is predicted.2.The effective target decision model of millimeter wave radar is constructed,which includes Kalman filter tracking algorithm and life cycle algorithm.Effective tracking and filtering of target points is realized.3.A data acquisition system of millimeter wave radar and camera is built and a spatiotemporal fusion model of two sensor data is established.4.Research on millimeter wave radar and image fusion algorithm.Several image vehicle detection algorithms based on deep learning are analyzed and compared.The millimeter wave radar target is processed by the effective target decision model,and the target level fusion is realized with the image target.