Intelligent driving vehicle laser radar overview
The rapid development of intelligent driving technology has greatly promoted the research of environmental sensing sensors. Vehicle light detection and ranging (LiDAR) plays an important role in intelligent driving environment sensing system because of its advantages such as accurate acquisition of 3D information, high resolution, strong anti-interference ability, wide detection range and near-all-weather work. The specific situation of intelligent driving puts forward high requirements for the hardware technology and the performance of the algorithm. Meanwhile, the volume, weight, cost and driving voltage of the LiDAR should meet the installation level of the vehicle system.
In order to meet the requirements of intelligent driving, the research on key technologies and algorithms of LiDAR is continuously expanded. In laser transmitting, receiving, scanning, and many other LiDAR hardware technology, it is generally accepted that the scanning technology is the key to the car laser radar. It directly determines the LiDAR scanning frequency, scanning scope, amount of data, and other important parameters, it is the key to reduce LiDAR volume and cost, and it also directly affects the realization of the production. In terms of algorithm development, based on different application scenarios selected by developers, different algorithms have different characteristics. There is no algorithm with perfect effect and strong adaptability; but all kinds of applied algorithm research aim to have real-time and high precision at the same time, so as to provide effective technical support for the control and decision-making of intelligent driving system.
The team of Professor Xiaodong Chenand Huaiyu Cai from Tianjin University affiliated to Tianjin University Unmanned Vehicle Cross Platform, is committed to the research of intelligent driving environment perception technology, has designed environment perception system, completed multi-sensor joint calibration, implemented obstacle detection algorithm based on multi-sensor data, etc. This paper reviews the key technologies and application algorithms of intelligent driving LiDAR in recent years.
This paper introduces the key hardware technologies of intelligent driving LiDAR by means of LiDAR scanning mode and related technologies, and discusses the principle, characteristics and status quo of mechanical, hybrid and all solid-state vehicle-mounted LiDAR. Based on the intelligent driving application task, this paper summarizes three kinds of LiDAR application algorithms: point cloud segmentation, target tracking and recognition, simultaneous localization and mapping (SLAM).
Mechanical LiDAR is first products applied to intelligent driving. With advantages of simple principle, easy to drive, easy to achieve horizontal 360 ° scan, it is still widely used in intelligent driving system, but the high cost and low reliability in the driving environment for long-term use hinder its promotion. Hybrid LiDAR designs the mechanical structure in a miniaturized and electronic way. The main components are produced by using chip technology, and the performance still needs to be optimized. It has been applied in intelligent driving solutions. All solid-state LiDAR does not have any macroscopic or microscopic moving parts, which is reliable and durable. However, it is still in the early stage of development due to its high difficulty in production process. It is recognized as the ideal target of low-cost and miniaturized LiDAR.
The purpose of lidar algorithm is to accurately perceive the surrounding environment in time, track and identify obstacles, complete intelligent vehicle positioning and navigation, and ensure its safe and efficient driving. Among them, the point cloud segmentation algorithm is the basis of target tracking and recognition, which will implement the judgment of the motion state and geometric characteristics of obstacles around the car, and SLAM will complete the accurate positioning and passable path planning of the car. The existing application algorithms of LiDAR have different degrees of limitations. Firstly, the accuracy and real-time performance of the algorithm are mutually restricted, which cannot be satisfied at the same time. Secondly, most of the algorithms are developed for a specific scenario, which is difficult to guarantee portability and stability. The complexity and diversity of the scene make the research of algorithm colorful, showing a multilevel and multi-angle situation.
Based on the above analysis, it can be seen that with the global intelligent driving entering into the preparation period of industrialization and commercialization, LiDAR has become an indispensable environmental sensing sensor with its excellent performance and has been rapidly developed in hardware technology and application algorithm. In order to reduce cost, improve performance and meet the demand of intelligent driving, LiDAR will further move towards solid state, intelligent and networked. The goal of applied algorithm research is real-time, efficient and reliable. It is an urgent problem to optimize and package typical algorithms and provide them to developers as mature modules.
Figure 1 Left: Luminar co-founder Austin Russell watches a LiDAR point cloud image inside a intelligent car. Right: Luminar lidar image of surrounding point cloud, color expresses the distance information of objects in the environment. Picture quoted from Jeff Hecht, "Lidar for self-driving Cars”, Optics & Photonics News 29(1), 26-33 (2018).
Figure 2 Left: famous Velodyne HDL-64E; Right: Velodyne VLS-128 LiDAR proposed in 2017
Picture quoted from “Velodyne officially launched the 128-line LiDAR VLS-128, with a detection range of 300 meters and a resolution which is 10 times higher than the previous generation” [EB/OL]. (2017-11-30)
Figure 3 Point cloud target detection and tracking
Picture quoted from "Ye Yutong, Li Bijun, Fu Liming. Rapid detection and tracking of point cloud
targets in intelligent driving [J]. Journal of Wuhan University (Information Science Edition), 2019, 44(01):142-147+155."
About the Group
Unmanned Vehicle Cross Research Center of Tianjin University was founded in March 2016, and is a multidisciplinary cross research platform composed of research teams from vehicle power, photoelectric engineering, image processing, machine learning, automatic control, mathematical optimization and nonlinear systems. The main research direction in intelligent driving including perception, planning, decision-making, control system design, core algorithm research and intelligent vehicle multidimensional test and evaluation technology, etc. As one of the core team in cross research center, research team led by Professor Huaiyu Cai, Xiaodong Chen from School of Precision Instruments & Opto-Electronics Engineering team main research direction for the opto-electric detection and imaging technology, responsible for environment perception hardware system design, sensor calibration, and obstacle detection and tracking algorithm research based on the LiDAR, visible/infrared cameras. The center has more than 20 professors, associate professors and lecturers, more than 60 doctoral and master students, including 1 National Excellent Youth, 1 Young Yangtze River Scholar and 5 New Century Talents from the Ministry of Education. The center has published more than 160 SCI/EI papers and obtained more than 40 invention patents authorized by the state. Related research achievements have won the first prize of natural science of China society of electronics (ranked second), the third prize of Tianjin natural science and other awards. Participated in the "World Intelligent Driving Challenge" for two consecutive years, and won many awards, such as the excellence award of unmanned driving group cross-country race, the excellence award of urban block scene race, the best rural road access award, and the leading award of virtual scene group.
Chen Xiaodong, Zhang Jiachen, Pang Weisong, et al. Key technology and application algorithm of intelligent driving vehicle LiDAR[J]. Opto-Electronic Engineering, 2019, 46(7): 190182.