Machine Learning for Autonomous Vehicle Environment Perception and Analysis

Abstract

Basically, this paper will cover the perception tasks from both sensor and algorithm perspectives and we will also discuss the current industry and the open problems in AV perception from both research and application perspectives. Based on this study, important views of possible future research topics in camera-radar perception for AVs will be pointed out. This survey is structured in six main sections. The first section is the introduction, the second section includes the surve literature on environmental perception. The surveys are done, in detail, for each sensor such as Cameras, LiDARs, Radars, and. More specifically, this section includes LiDAR surveys and compares the ultimate approach applied for perception. Then, the Radar surveys are presented in detail according to their perception algorithms. Then, Camera-Radar fusion surveys are included. Another sub-section is a very active and beginning trend in this survey. Basic topics are discussed, such as background subtraction for static background especially due to the illumination changes, and the second topic is illumination invariant moving object detection and tracking. Finally, related works and the discussion section is in section four. A very detailed future research direction section is also provided in this section. Section five is the conclusion.