Overview | research on edge computing technology for autopilot
2021-05-04 11:09:49 【Sharer: Xiao Ke】
Edge computing plays an important role in environment perception and data processing of autonomous driving . Autopilot can extend its sensing range by obtaining environmental information from the edge nodes. , It can also unload computing tasks to edge nodes to solve the problem of insufficient computing resources . Compared to cloud computing , Edge computing avoids the high delay caused by long distance data transmission , It can provide more rapid response for autonomous vehicles , And reduce the load of the backbone network . Based on this , Firstly, the technology of collaborative sensing and task unloading for autonomous driving vehicles based on edge computing is introduced. , Then the research status of collaborative sensing and task offloading technology is analyzed and summarized , Finally, the problems to be further studied in this field are discussed .
With the development of computer technology and sensor technology, the automobile industry is gradually developing Become a more intelligent autopilot car . The emergence of autopilot will improve the efficiency of traffic flow. , Reducing road traffic accidents . The U.S. highway traffic safety administration divides automatic driving into L0～L5 common 6 Level , from L0 To L5, The intelligence of the car The level of education has gradually improved .L5 A class of self driving cars can perform all the driving operations in any environment without the need of human drivers. Intervention . In order to provide safe automatic driving service , Autopilot You need to get complete information about your environment , And it is processed in real time To make driving decisions . From obtaining environmental information to driving decision It can be divided into three parts 3 Stages , They are environmental information acquisition 、 Information fusion processing and driving behavior decision making .
1) Environmental information acquisition . Autopilot goes through multiple vehicles. Sensors get information about their environment , Such as positioning system 、 inertia The measurement unit obtains the position information , Drawing the surrounding environment with lidar Point cloud of , Using camera to get image data of environment , Using radar and Sonar detects the object closest to the vehicle .
2) Information fusion processing . The task of this phase is to get the ring Environmental information fusion processing , Make the autonomous driving vehicle understand the surroundings. Environmental Science . There are three stages in information fusion processing 3 There are three main tasks , , respectively, It's self positioning 、 Target recognition and tracking . this 3 It's all a mission It needs a lot of computing resources to complete .
3) Driving behavior decision making . The self driving car understands itself. After the environment , Start to predict other obstacles such as vehicles or pedestrians Moving path of obstacles , Then make your own path according to the prediction results Path planning and obstacle avoidance decision making .
High level self driving cars need no interference. Complete the above tasks under the same conditions 3 A step , But only rely on a single autopilot The ability of a car is very difficult to achieve , Here's why .1) In the aspect of environmental information acquisition , Single autopilot In the blind area of sensor field of vision , Unable to obtain complete information about the environment Rest , This may result in autonomous vehicles not being able to detect the coming traffic dangerous .2) In terms of data processing , It is difficult for on-board computing system to complete the calibration Real time processing of massive heterogeneous sensor data , Only each camera In seconds 1.8 GB The data of . If you are driving on an autopilot car High performance computing system , It will greatly increase the number of self driving cars. Cost of . On the whole , Environmental information acquisition and data processing The ability to rely solely on a single autopilot can not be achieved. Level autopilot , Therefore, the assistance of other intelligent nodes is needed . In order to solve the problem of data processing , Some researchers suggest that A scheme of combining automatic driving with cloud computing is presented. , Put the data Upload to the cloud for processing .
Although there are a lot of computing resources in the cloud , It can complete the data processing in a very short time , But only relying on Cloud services for autopilot cars are not in many cases. feasible . Because autopilot can generate great driving. The amount of data that needs to be processed in real time , If these data are transmitted to the remote cloud through the core network for processing , Then only the data transmission will be completed It causes a lot of delay , It can not meet the real-time requirements of data processing seek . The bandwidth of the core network is also difficult to support a large number of self driving cars. Send a lot of data to the cloud at the same time , And once the core network goes out The current congestion leads to the instability of data transmission , Driving of a self driving car Security is not guaranteed . Applying edge computing to the field of autopilot will help to solve the problem Self driving vehicles are faced with environmental data acquisition and processing. The problem of . Edge computing is a process of computing at the edge of a network Three kinds of computing models , Its operation objects come from the downstream data and data of cloud services Uplink data of Internet of things service , But in edge computing “ edge ” It refers to any computing path from data source to cloud computing center And network resources . In short , Edge computing deploys servers to Edge nodes near users , On the edge of the Internet （ Such as wireless access point ） Provide services to users , Long distance data transmission is avoided , To the user Provide faster response . Edge of computing 、 Mobile edge computing 、 The application of fog calculation in the field of automatic driving is similar , In this paper, they are discussed It is called edge computing . Collaborative sensing and task offloading are the key technologies of edge computing in autopilot Main applications in the field , this 2 Three technologies to realize high level automatic driving Driving is possible . among , Collaborative sensing technology enables cars to acquire Sensor information of other edge nodes , Extended autopilot Range of perception , The integrity of environmental data is increased ; Task unload Technology will unload the computing tasks of auto driving to other edges. Node execution , Solved the problem of insufficient computing resources for autopilot cars. topic . Self driving vehicle for edge calculation cannot do without driving wireless Communications technology （V2X, vehicle-to-everything） Support for , it Other elements in self driving cars and intelligent transportation systems are provided. The means of communication , It is the base of the cooperation between the self driving vehicle and the edge node. Foundation . at present ,V2X Mainly based on dedicated short-range communication （DSRC, dedicated short range communication） And cellular networks . among DSRC It's a kind of vehicle that is specially used for vehicle and vehicle （V2V, vehicle-to-vehicle） And vehicle and road infrastructure （V2I, vehicle-to-infrastructure） Communication standards between , Have number High data transmission rate 、 Low latency 、 Support point-to-point or point to multipoint communication Credit and other advantages . With 5G The cellular network represented by has network capacity Large amount of 、 Wide coverage and other advantages , Apply to V2I Communication and edge Communication between servers .
This paper first introduces the relationship between the self driving vehicle and the edge nodes. Collaborative awareness and task offloading are introduced , And the problems they face are discussed The challenge of ; Then, the cooperative sensing technology and task offloading are summarized respectively The current situation of the research of the technology ; Finally, the problems to be further studied in this field are pointed out .
本文为[Sharer: Xiao Ke]所创，转载请带上原文链接，感谢
- C++ 数字、string和char*的转换
- Won the CKA + CKS certificate with the highest gold content in kubernetes in 31 days!
- C + + number, string and char * conversion
- C + + Learning -- capacity() and resize() in C + +
- C + + Learning -- about code performance optimization
C + + programming experience (6): using C + + style type conversion
Latest party and government work report ppt - Park ppt
Online ID number extraction birthday tool
Field pointer? Dangling pointer? This article will help you understand!
GVRP of hcna Routing & Switching
- LeetCode 91. 解码方法
- Seq2seq implements chat robot
- [chat robot] principle of seq2seq model
- Leetcode 91. Decoding method
- HCNA Routing＆Switching之GVRP
- GVRP of hcna Routing & Switching
- HDU7016 Random Walk 2
- [Code+＃1]Yazid 的新生舞会
- CF1548C The Three Little Pigs
- HDU7033 Typing Contest
- HDU7016 Random Walk 2
- [code + 1] Yazid's freshman ball
- CF1548C The Three Little Pigs
- HDU7033 Typing Contest
- Qt Creator 自动补齐变慢的解决
- HALCON 20.11：如何处理标定助手品质问题
- HALCON 20.11：标定助手使用注意事项
- Solution of QT creator's automatic replenishment slowing down
- Halcon 20.11: how to deal with the quality problem of calibration assistant
- Halcon 20.11: precautions for use of calibration assistant
- "Top ten scientific and technological issues" announced| Young scientists 50 ² forum
- Reverse linked list
- JS data type
- Remember the bug encountered in reading and writing a file
- Singleton mode
- 在这个 N 多编程语言争霸的世界，C++ 究竟还有没有未来？
- In this world of N programming languages, is there a future for C + +?
- js Promise
- js 数组方法 回顾
- ES6 template characters
- js Promise
- JS array method review
- 【Golang】️走进 Go 语言️ 第一课 Hello World
- [golang] go into go language lesson 1 Hello World