Love is parallel to the world 2021-08-08 14:58:48 阅读数:967
Internet of vehicles white paper （ Netlink autopilot fascicle ）
source ： China academy of information and communications 2020 year 12 monthOne 、 The connotation of Internet connected automatic driving
Automatic driving is an intelligent vehicle as a vehicle 、 Core application of networking development , It's also the Internet of vehicles 、 The core application service of intelligent transportation industry development .
On a technical level , Currently there are There are two different implementation paths: single vehicle intelligent automatic driving and network connected automatic driving .
Single vehicle intelligent automatic driving mainly depends on the vehicle's own vision 、 Millimeter wave radar 、 Sensors such as lidar are used to sense the environment 、 Computing decisions and controlling execution .
Environmental perception completes the detection and positioning of the surrounding environment through on-board sensors . On the one hand, the sensor data are analyzed and processed , Realize target recognition ; On the other hand, conduct behavior prediction and global path planning 、 Local path planning and real-time action planning , Determine the current and future trajectory of the vehicle .
Control execution mainly includes vehicle motion control and human-computer interaction , Determine each actuator such as motor 、 throttle 、 Braking and other control signals .
At present, single vehicle intelligent automatic driving is in environmental perception 、 There are different degrees of technical bottlenecks in many links of computing decision-making and control execution , In the application process, there are also various failure problems , Therefore, on the one hand, it is necessary to continuously strengthen the perception of bicycle intelligence 、 Decision making and control capability ; On the other hand, we also hope to introduce different technical means to make up for it .
Internet connected automatic driving is based on the existing single vehicle intelligent automatic driving , The aim is to integrate “ people - vehicle - road - cloud ” The elements of traffic participation are organically linked , Expand and assist the application of single vehicle intelligent automatic driving in environmental perception 、 Ability upgrade in computing decision making and control execution , Accelerate the maturity of automatic driving application .
In the context of environmental awareness , Support the vehicle to obtain more information than the single vehicle intelligent perception , For example, non line of sight perception or solving problems that are easy to be affected by harsh environment ;
Collaboration in computing and decision making , Add cars and cars 、 Systematic decision making between car and road , For example, solve vehicle priority management 、 Traffic intersection optimization control and so on ;
Cooperate in the control and execution process , Intervene in vehicle driving behavior , For example, remote control vehicle rescue, etc .
meanwhile , The development of Internet connected automatic driving will also drive “ people - vehicle - road - cloud ” Construction and improvement of new infrastructure system for collaborative vehicle networking , help 5G、 Application and promotion of information and communication technologies such as artificial intelligence in vertical industries , Promote the development of new models and new business forms of automobile and transportation services .Two 、 Requirements and typical applications of networked automatic driving
According to the American Society of Automotive Engineers （SAE） The division of L0-L5 In terms of automatic driving level , At present, the intelligent automatic driving of single vehicle is in its infancy L2-L3 Landing development stage of grade , Market penetration and application scale are still small , Advanced auxiliary driving system functions （Advanced Driver Assistance Systems,ADAS） Still the main force . Gaogong Intelligent Automobile Research Institute 《2020 year 1-2 Monthly new passenger car insurance volume ADAS Market data report 》 Put forward , Domestic independent and joint venture brands launch new cars ADAS The carrying rate is 28.15%.《 Intelligent network connected vehicle technology roadmap 2.0》 Put forward , To 2025 year ,L2 and L3 Considerable part of automatic driving （PA ） And conditional autopilot （CA） The proportion of car sales will exceed 50%, Highly autonomous （HA, amount to L4） Start entering the market . From the R & D Progress of various automobile enterprises ,2018 BMW in 、 Volvo 、 audi 、 General motors and others have been launched one after another L2 Class a autonomous driving application ;2020-2022 The year is planned to be launched by various car enterprises L3 Time of automatic driving vehicle , But there are only a few L2.5、L2.99 Conceptual application of automatic driving , On the one hand, limited by L3 Improvement of laws and regulations for automatic driving vehicles , On the other hand, it is limited by technical implementation . Comprehensive analysis of , From the perspective of Technological Development , At present, the technical realization ability of single vehicle intelligent automatic driving is L2-L3, But it is limited by the scene , For example, it needs to be on the highway 、 Urban trunk roads with standardized and clear road signs, etc ; From the perspective of industrial application ,L2 And the following ADAS The application still maintains a low penetration level ,L3 And above autopilot applications are still mainly experimental and regional demonstration ,L4 And higher level automatic driving applications take longer .
At present, there are a lot of commercial mass production ADAS The function still has the risk of insufficient coping ability and failure under specific scenarios . Take automatic emergency braking as an example ,2019 American Automobile Association （AAA） To Chevrolet Malibu XL、 Honda accord 、 tesla Model 3 And Toyota Camry , On the one hand, it is found that the vehicle's coping ability is insufficient in scenes such as night or children's shuttle , Most of the models had collision ; On the other hand, it's an umbrella on a rainy day 、 A raincoat 、 Tunnel and other scenarios are easy to cause failure . Existing commercial ADAS The function is not yet able to cope with complex traffic conditions or severe weather conditions , The lack of perception is still the main reason . But under the condition that the driver is responsible for the whole driving process ,L2 And the following ADAS The application function is ready for commercial landing .
In terms of high-level automatic driving road test , The reliability of automatic driving and the ability to cope with challenging traffic scenes still need to be improved . from 2019 Annual California autopilot disengagement Report 《Autonomous Vehicle Disengagement Reports》 Look at ,36 Enterprises enter
Did the autopilot test , Google Waymo With 234 10000 km test mileage is far ahead , Its every 21273 There was a breakaway takeover , But by comparison, apples are every 189 There's going to be a breakaway takeover . From the reasons for leaving the takeover , The reliability of software and hardware system performance is a common problem faced by almost every testing enterprise ; Besides , Insufficient perception of sudden targets 、 The prediction ability of target motion behavior is insufficient 、 Decision time timeout and wrong trajectory generation 、 The error of traffic information identification is also the main problem .
Besides , Some commercial autonomous vehicles also have various types of accidents , There is a perceived failure 、 Failure of prediction and decision-making .2018 year 1 month , One in Los Angeles, USA Model S The fire truck that fails to detect the stop ahead in time due to the sudden lane change of the following vehicle , Failure to brake in time / Accident caused by deceleration , Lead to the discussion of its perceived failure .2016 year 2 month , A Lexus vehicle in California sensed that a bus was ready to pass behind , However, it is judged that there is not enough space on the road for buses to pass , It is assumed that the bus driver will slow down , The bus driver judged that Lexus would yield , Therefore, collision occurs , It shows the intelligent automatic driving of a single car in this kind of “ game ” It is difficult to accurately judge the intention of the surrounding traffic participants under the driving conditions , The challenge of collaborative decision-making is put forward .
at present , Technical solution of single vehicle intelligent automatic driving , Visual sensors 、 Laser radar 、 Millimeter wave radar and infrared night vision 、 Mature sensors such as ultrasonic are the main product components .
The nominal technical indicators of various types of sensors continue to develop steadily , Continuously meet the needs of automatic driving , Gradually approaching the perception of human drivers , It has even surpassed some technical capabilities , For example, detection distance 200 Meters above 、 The comprehensive accuracy can reach centimeter level or even millimeter level . However, the reliability of various sensors , And the ability to respond to emergencies is still insufficient . One side , Vulnerable to occlusion 、 Adverse weather and other environmental conditions , Such as intersection 、 Tunnel entrance and exit, etc .
Networking through vehicle road collaboration 、 Car car collaboration , It can greatly expand the perception range of bicycles , And it's not restricted by occlusion , It allows the bike to discover the unknown ahead of time , Be able to deal with situations that are difficult to deal with in autopilot tests and accidents, such as the sudden entry of the target . Besides , Application of single vehicle intelligent automatic driving in target prediction 、 Driving intention “ game ” And so on .
Networking can directly give the state information of key results , For example, signal lamp status 、 Next action intention of surrounding vehicles 、 The best driving route under the current road conditions, etc , It reduces the complex calculation and processing process based on sensor information , And can accurately understand the intention of the surrounding traffic participants .
The main functions of computational decision can be divided into two categories , One is target recognition based on environment perception data , Deep neural network is the most widely used way in perception , It is also the computing task with the greatest demand for computing power consumption ; The second is the result of perception and the driving task of the vehicle , Give the route 、 Decision planning of vehicle movements .
On hardware , Computational decision-making is mainly based on CPU、GPU、DSP、AI chip 、MCU On the computing and processing platform with multi-core and heterogeneous distribution .
The contradiction between computing power and power consumption is an important bottleneck encountered by the single vehicle intelligent automatic driving computing and processing platform .
meanwhile , Because the traffic behavior is more between many participants “ game ”, In the decision-making and planning of path action , It is difficult to give the best solution for single vehicle intelligent automatic driving .
Network connection is expected to share the computing power consumption of bicycles , The global optimal driving strategy is given based on the cloud control platform .
In computing power , One is the networking as “ Super sensors ” Can directly give the perceived target results , The complex calculation and analysis process of sensor signal is omitted , Such as the judgment of traffic lights , This greatly reduces the computing power demand of the bicycle ; Second, we can use cloud computing 、 Edge computing, etc , It is expected to introduce the roadside calculation force into the road , For example, install a vision sensor on the roadside 、 Lidar and other sensors , Distribute the roadside perception results, etc .
In terms of driving strategy , In certain situations , Network connection can collect traffic participants within its scope , According to the purpose and state of all subjects , Give a global optimal solution , There is no need to pass “ Probe ” and “ game ” Give a decision plan , In the mine 、 port 、 Logistics and other non-public open roads have been verified and applied in specific scenarios .
The control execution of single vehicle intelligent automatic driving is mainly based on the action command given by the calculation decision , Through the vehicle dynamics model and human-computer interface , To the motor 、 throttle 、 Brake and other actuators .
In terms of control execution , The coordination between automatic driving system and human driving is considered, and the reliability of vehicle control is also considered. 、 Security , Redundant control systems 、 High real-time response is the main technical requirement .
In terms of control execution, the network can provide remote control driving 、 The application mode of cooperative driving . For example, in some dangerous or unsuitable occasions for human access , Need to pass through 5G Remote control driving to operate the remote vehicle for operation , At present, it has been applied in unmanned mines and other occasions .
In terms of vehicle formation driving, etc , Information interaction is performed by means of the control of the front and rear vehicles , The rear vehicles can drive according to the unified command of the head vehicle , Reduce the perceived computing task load of rear vehicles .
Networking can separate vehicle control and execution from single vehicle , Help create a number of innovative application models .
On the basis of single vehicle intelligent automatic driving , Introduction of networking technology , So that more cooperative automatic driving applications can be realized .
It is considered from the two dimensions of typical working conditions and collaborative links , The application scenario of networked automatic driving can be divided into a matrix .
Typical working conditions include various intersections 、 Expressway 、 Tunnel 、 Parking spaces 、 mine / Port and other relatively structured road environment , And general working condition environment . Collaborative link includes collaborative perception 、 Collaborative decision-making and collaborative control .
The dependence of different application scenarios on Internet connection can be divided into two categories , One is the application scenario that requires network connection to be realized in the full working condition environment . For example, signal lamp 、 Information acquisition such as identification plate , In rainy, snowy and foggy weather with low visibility , Single vehicle sensing devices will be difficult to accurately identify , The above information can be obtained through the networked communication between the vehicle and the corresponding traffic infrastructure on the roadside ; More Than This , The digitization of signal lights and other information increases the possibility of dynamic adjustment , Make tidal Lane 、 Dynamic speed limit and other scenarios are easier to implement . secondly , Blind zone perception , The camera and lidar that can be carried by a single car have physical limits , The deployment of some roadside sensing devices can easily solve the visual blind spot of a single car , And the remote dynamic traffic event information can be 、 Send road information to vehicles . also , Group intelligence , Global path planning based on traffic information needs a powerful decision-making platform based on cloud , And distribute the planning results to the vehicles in the area , This can not be achieved by single vehicle intelligence , And the remote monitoring and takeover intervention of abnormal vehicles can be realized through network connection . The second is to realize more economical application scenarios through network connection . Such scenarios are also technically feasible , However, the implementation cost is too cumbersome and expensive , It is not conducive to the rapid popularization of applications . For example, typical urban working conditions such as intersections , In the process of identifying mixed traffic subjects , Bicycles need to pay a great deal of perceptual and computational overhead , And through roadside perception 、 Calculation 、 Deployment of communication and other equipment , It can achieve the large-scale effect of resource reuse . Another example is the tunnel 、 Location of closed spaces such as parking lots , With roadside communication equipment 、 Cellular communication base station 、 Edge computing server, etc , With the same accuracy , It can greatly reduce the single vehicle inertial navigation 、 The overhead of a series of complex fusion sensing algorithms such as radar .
at present , Relevant application scenarios of Internet connected automatic driving have also been carried out at home and abroad .
port 、 mine 、 Logistics parks and other closed scenes have become the first places to deploy commercial vehicles L4 Demonstration area for automatic driving .
2019 year 11 month , By SAIC 、 Shanggang group 、 Shanghai Yangshan Port smart heavy truck demonstration operation project jointly built by China Mobile , In Yangshan Port Logistics Park 、 Donghai Bridge 、 In Yangshan phase I Wharf , Realize intelligent container transfer , This is the first time in the world 5G+ Commercial landing of automatic driving heavy truck .
5G Combination of remote control driving and single vehicle intelligent automatic driving , It can solve the manual takeover caused by the failure of automatic driving algorithm and other reasons . Bosch 、 Mercedes Benz 、 Nokia and other partners , Demonstration of non-stop import based on smart infrastructure and edge computing in Ulm, Germany , Here's the picture 1 Shown . The project construction party shall install cameras on road lamp posts 、 Laser radar 、 Roadside communication equipment and MEC Edge computing server , Expand vehicle perception , Solve the visual recognition blind area of the vehicle's own perception equipment , Such as pedestrians blocked by trucks 、 Vehicles coming from the blind area 、 A bicycle approaching and changing lanes from the rear .
The system can collect the image data collected by the road end sensor , Combined with data collected by vehicle sensors , With the support of edge computing force , Combined with high-precision map , It can generate a vehicle surrounding environment model containing all the information of the current road conditions , And transmitted to the vehicle through network technology . After the implementation of the project , Autonomous vehicles can accurately find the traffic flow gap on the main road , It can seamlessly merge into the traffic flow of the main road without stopping .3、 ... and 、 The technical architecture of networked automatic driving
The architecture of networked automatic driving can be “ terminal ”“ Connect ”“ Computing and services ” Three dimensions to deconstruct , Here's the picture 2 Shown . Development of the network automatic driving technology system “ terminal ” It includes two generalized terminals: intelligent networked vehicle and roadside intelligent system .“ Connect ” It's the realization of “ people - vehicle - road - cloud ” Interconnected communication technologies , According to the network type, it can be divided into public telecommunication networks 、 Public security network 、 Traffic network, etc , According to network technology, it can be divided into mobile communication networks 、 Optical fiber access network, etc .“ Computing and services ” It carries the data support and application implementation capability of various services of Internet connected automatic driving , Physical entities include edge computing platforms and cloud platforms ; Include data base from business logic 、 Open interface 、 Application services, etc. ; In terms of service functions, it includes push with traffic light information 、 Blind spot perception 、 Collaborative sensing applications represented by tunnel high-precision positioning , Global path planning 、 Cooperative decision-making applications such as vehicle formation driving ,
5G Cooperative control applications such as remote driving , And high-precision map download 、OTA Upgrade and other data support services .
Intelligent networked vehicle and roadside intelligent system can be regarded as two system levels “ terminal , Can be further deconstructed into “ terminal ”“ Connect ”“ Computing and services ” And other subclasses .
Internal of intelligent networked vehicle “ terminal ” Including on-board communication gateway 、 Vehicle sensing equipment , And smart cockpit 、 Domain controller and other automotive electronic systems ;“ Connect ” It is mainly the integration of in vehicle high-speed bus and multi protocol in vehicle communication technology ;“ Computing and services ” Involving heterogeneous computing platforms 、 On board operating system 、 Perceptual recognition and behavioral decision algorithm, etc .
Design of roadside intelligent system “ terminal ” There are mainly roadside communication terminals 、 Roadside sensing devices 、 Road traffic intelligent equipment, etc ;“ Connect ” Mainly by traffic 、 Public security network or local LAN ;“ Computing and services ” It mainly includes roadside edge calculation system 、 Roadside perception and service capability 、 road - Cloud Association equivalence .
In the idea of decomposing the overall structure of this system ,“ terminal ” It's the carrier ,“ Connect ” It's the way ,“ Calculation ” It's the means ,“ service ” It is purpose ,“ people - vehicle - road - cloud ” In depth collaboration of key technologies in all links , A unified and integrated network automatic driving technology architecture has been formed .
From the perspective of intelligent networked vehicles , Deconstruction of technology system from a global perspective , Internet connected automatic driving can be divided into perception according to technical links 、 Connect 、 Computing and services , The networked system is connected with radar 、 camera 、 A perceptual technology of parallel positioning , Interior electronic and electrical structure （E/E）、 The on-board computing platform belongs to the category of connection and computing , Services mainly include prediction and decision-making related to automatic driving , Here's the picture 3 Shown . From the perspective of intelligent networked vehicles , Vehicle safety is a very important technical link , The corresponding functional safety system shall be followed （ISO 26262）, And expected functional safety system （SOTIF）.
From the perspective of intelligent networked vehicles , Autopilot to C-V2X It puts forward many requirements for the representative network system , Including application delay 、 Availability of message sets 、 Reliability of data in message 、 Mutual backup between messages and in vehicle perception, etc .
The characteristics of this perspective can be summarized as ：
（1） The network connection system will be a part of the vehicle perception system , It can provide vehicle blind area information 、 Over the horizon information 、 Traffic control information and other information that can not be directly provided by traditional on-board sensors ;
（2） Automatic driving is safe for functions related to network connection system 、 The expected functional safety system is of great concern , It needs further research and cross industry consensus ;
（3） Automatic driving has a strong demand for message set standardization in typical application scenarios , It is necessary to promote the formation of message set continuous evolution mechanism , Support the extensive deployment of networked automatic driving infrastructure .
From the perspective of information communication , Netlink autopilot “ End - tube - cloud - Business ” Typical information communication architecture , One side , Whether on board 、 Various terminals of roadside system , They are all end users of the Internet of vehicles in the network ; On the other hand , Various terminals pass C-V2X/5G And other mobile communication networks or optical fiber access networks are connected to edge computing platforms or cloud platforms , Various applications on the platform provide general or customized services for end users , Here's the picture 4 Shown .
From the perspective of information communication , On the one hand, the terminal is the main body of network services , It is necessary to adjust the network capability or service characteristics for the needs of terminal services ; On the other hand, it puts forward clear requirements for the matching of the terminal to the network or service , For example, operating frequency requirements 、 RF consistency 、 Communication protocol consistency 、 Data set consistency, etc .
The characteristics of this perspective can be summarized as ：
（1） Network automatic driving system follows “ End - tube - cloud - Business ” framework , The subject of each link needs to be subject to frequency band license and telecom business qualification ICT Supervision of policies and regulations in the field ;
（2）ICT Enterprises want to use autopilot 、 The Internet 、 Service, etc , On this basis, carry out management 、 cloud 、 System development and deployment in service ;
（3）ICT Enterprises can provide for autonomous driving applications “ standard + customized ” service , However, how to clarify and provide the reliability or confidence of Internet connection data still needs further research .
From the perspective of traffic and traffic management , vehicle Networking and intelligent transportation systems are vertically classified according to business types , Netlink autopilot can be regarded as an emerging business system involving the enhancement and integration of multiple traditional business types , For example, data opening and broadcasting are added to the traditional traffic signal intelligent control system , Marking at traffic signs 、 Vehicle overspeed warning 、 On the basis of typical illegal early warning system, information fusion and real-time broadcasting functions are added , Here's the picture 5 Shown .
From the perspective of traffic and traffic management , Various business applications need the support of road side intelligent equipment and basic data platform .
For netlink autopilot , New entities and functions need to be added in roadside intelligent equipment and basic data platform , Including new C-V2X Roadside communication unit 、 New roadside perception computing fusion system 、 Deploy V2X Information collection and release function, etc , Realize the integration and evolution of networked system and traditional intelligent transportation system .
The characteristics of the perspective of traffic and traffic management can be summarized as follows: ：
（1） Traffic and traffic management system “ perception - transmission - Calculation - application ” Design at different levels , With C-V2X The integrated development of network connection system construction and new transportation infrastructure represented by has formed an industry consensus ;
（2）C-V2X As a new way of perceptual publishing, etc , It can enable the traffic and traffic management system to enhance the interaction with intelligent networked vehicles , Promote the extension of traffic and traffic management business from road end to vehicle end ;
（3） The transportation and traffic management industry needs to refine the services required for Internet connected automatic driving based on the existing intelligent transportation services , Integration and evolution with existing systems 、 And pass C-V2X The isoconnect system is coupled with the isoconnect autopilot .
In order to better describe the network automatic driving technology system , This white paper attempts to summarize the network automatic driving technology system with a three-way view , Here's the picture 6 Shown .
The forward view shows the system architecture and information flow , According to different business types , Information flow can be divided into three types .
First, the traffic information flow , Roadside intelligent system provides over the horizon perception information for intelligent networked vehicles 、 Road information 、 Geographic information, etc ; The intelligent networked vehicle provides its own status information to the roadside intelligent system , It is used for the roadside to get a more comprehensive understanding of the state of road vehicles , Form a closed loop of vehicle road collaborative flow .
The second is the vehicle road cloud information flow for traffic management , The traffic and traffic management platform obtains vehicle driving information from roadside intelligent system and intelligent network connected vehicle , Used to analyze and form macro or micro traffic conditions , And control the traffic 、 management 、 Information such as guidance shall be sent to intelligent networked vehicles , Promote the improvement of traffic environment , Form a closed loop of intelligent traffic information flow .
Third, the car road cloud information flow for Internet connected automatic driving , The intelligent networked vehicle and roadside intelligent system upload the perception information and road status information to the cloud control platform , The cloud control platform sends the fusion perception and decision information applicable to automatic driving to the intelligent networked vehicle , Form a closed loop of network connected automatic driving information flow .
The top view shows the logical architecture of the key technologies , Information communication 、 traffic 、 Automobile and other industries “ terminal ”“ Calculation ”“ Connections and services ” We basically reached a consensus on the division of technical logic , However, different industries pay different attention to this logical division .
The automotive industry pays more attention to the internal system of intelligent networked vehicles , At the same time, pay attention to the implementation of automatic driving application service on cloud control platform . The communication industry pays more attention to the connection and services between vehicle and road clouds , And the side - In the clouds MEC、 Data Center 、 Cloud computing 、 Artificial intelligence and other key technologies .
Roadside intelligent system needs information communication 、 traffic 、 Cross industry collaboration such as traffic management , Among them, the responsibility and authority interface of the traffic management industry is relatively clear , The cooperation mode between information and communication industry and transportation industry in roadside system construction and operation needs to be further explored ; Application platform , Information communication 、 traffic 、 Traffic management 、 There are still inconsistent data sets among platforms in various industries such as automobile 、 Interconnection difficulties and problems waiting to be solved .
The side view shows the system safety view , The security system includes two parts: network automatic driving function security and information security .
Functional safety includes the traditional functional safety system of automobile , It also includes the expected functional safety system for autopilot cars. , And future roadside equipment 、 Expected functional security of cloud services .
Information security includes network security 、 Identity authentication and other key technical systems , On the one hand, the identity authentication system needs to clarify the management mechanism , Consider establishing cross industry 、 Cross regional collaborative management mode , Jointly maintain the digital identity trust relationship of collaborative mutual recognition .
In recent years, vehicle vision camera has become the focus of automobile industry ADAS Sensors with the fastest growing market demand . According to the difference of camera installation position and function , Can be divided into forward-looking 、 Look around 、 Look sideways 、 Rear view and internal view, etc .
The front view camera is mainly used for Vehicle and pedestrian detection 、 Traffic sign recognition 、 Lane Departure Warning 、 Distance monitoring and adaptive cruise control , It usually needs to be matched with complex algorithm chips .
The look around camera is mainly used for Panoramic parking and lane departure warning , The side view camera can be used for blind spot detection , Rear view camera for parkassist , Internal view camera is used for fatigue driving warning and emotion recognition .
Currently, on-board cameras are mainly 720P、1080P Resolution based , The spatial resolution is close to that of human eyes , The perceived distance is usually 200m, Distance from human naked eye （500+m） There is still a gap . Backlight 、 Image dynamic range is the main challenge affecting the reliability of vision sensor .
Visual perception technology transforms the input data of sensors into scene semantic expression and object structured expression that can be understood by computer , Including object detection 、 Identify and track 、3D Environment modeling 、 Motion estimation of objects, etc .
When the vehicle is running , Continuous acquisition through HD camera , Realize real-time perception of environmental information , With the improvement of automatic driving level , It will inevitably lead to the balance between pixel demand and chip computing power .
Vehicle mounted lidar is mainly used for obstacle avoidance , Will go 3D Point cloud recognition and location .
Vehicle lidar is the most accurate way of vehicle environment perception at present , The detection distance can reach 300m, The accuracy can be controlled at the centimeter level . Mechanical lidar has strong anti-jamming ability 、 The advantages of high signal-to-noise ratio have become the main form of lidar in the early stage , In the long run MEMS、3D Flash And other solid-state lidar is expected to become the focus , Phased array OPA Technology solutions still need a long technology R & D cycle .
At present, the main restrictive factors restricting the mass production and commercial use of lidar are reliability and cost .
The technology of vehicle mounted millimeter wave radar is the most mature 、 Highest robustness , It can detect the distance between the vehicle and the target object , It is mainly used for Collision warning 、 cruise control 、 Functions such as brake assist and park assist .
at present , The frequency of vehicle mounted millimeter wave radar is mostly used 24GHz Frequency band and 77GHz Frequency band .24GHz The technical difficulty and cost are low , It is suitable for measuring medium and short distance objects , It occupies the main application market of millimeter wave radar .
77GHz The millimeter wave radar has smaller volume 、 Longer measurement distance 、 Higher measurement accuracy , It is suitable for measuring long-distance objects ,77GHz And other high-frequency millimeter wave radar products are mainly foreign products .
As the technology matures ,3D、 High spatial resolution millimeter wave radar has become the future trend , Theoretically, it is possible to replace lidar .
Multi-sensor fusion has become the main means to improve the reliability of perception . Different sensors in sensing accuracy 、 Robustness 、 Reliability varies , Therefore, it is suitable for different environments 、 Perceptual measurement of different objects .
Combined with lidar 、 Sensor fusion technology of millimeter wave radar and vision , Higher accuracy can be obtained 3D Information , And through the complementarity of different information 、 Cross validation , Improve the accuracy of semantic perception by several orders of magnitude .
at present , There are two different perceptual fusion schemes: vision dominated and laser dominated , The adopters of visual leading scheme are represented by Tesla , The adopters of laser led solutions are represented by Google . Millimeter wave radar has excellent anti-jamming performance , It has become an important supplement to cameras and lidar . On the technical route , Multisensor fusion sensing mainly includes data level forward fusion and feature level backward fusion . Data level forward fusion means simultaneous interpreting the information of different sensors at the original data level. , The program is mostly used for academic research , Contrapuntal force 、 The sensor requires high reliability , Poor robustness .
The backward fusion scheme based on feature level is to extract the features of a single sensor , Then the limited feature information is fused and optimized , This scheme is the mainstream of current engineering implementation , Simultaneous interpreting of confidence levels for different sensors is needed. 、 Sensor noise and other problems .
International standards organization 3GPP The definition is based on LTE Mobile communication technology evolved into LTE-V2X、5G And 5G V2X Standardization Technology .LTE-V2X On 2017 year 3 Standardization completed in June , Introduced work in 5.9GHz Direct link of frequency band （PC5 Interface ） communication mode . 2020 year 7 month ,3GPP announce R16 NR-V2X Version freeze , Unicast and multicast modes are introduced 、HARQ feedback 、CSI Measurement report 、NR/LTE Base station scheduling LTE-V2X/NR-V2X resources 、NR-V2X And LTE-V2X Coexistence and other new technical features , It supports high-order modulation and spatial multiplexing, and optimizes the resource selection mechanism .3GPP On 2020 In the first 3 Quarterly launch R17 Relevant standardization work . Under the guidance of the vehicle Networking Industry Development Committee of the national manufacturing power construction leading group , focusing C-V2X field , Automobile Standards Committee 、ITS Standardization Committee 、 General standard committee 、 Bid submission committee to speed up the implementation 、 Important standard setting . The general standard committee has basically completed LTE-V2X Overall framework 、 Air interface 、 The network layer 、 Message layer 、 Communication security and other basic support as well as the formulation of technical standards and test specifications related to interconnection . Automobile Standards Committee 、ITS The Standardization Committee and the bid submission committee are working out respectively LTE-V2X Relevant application standards , promote LTE-V2X Technology in driving service 、 Practical applications in transportation infrastructure and traffic management .
Relying on the domestic good industrial environment , be based on LTE-V2X Chip module of 、OBU、RSU And other core equipment have practical commercial capabilities , And the supporting end-to-end industrial chain has been established .
As an important national development strategy , At home 5G NR（Uu） Infrastructure construction has also begun to take shape .
From the current industrial R & D focus and subsequent product planning ,LTE-V2X And 5G NR（Uu） Multimode terminal equipment is the focus of future R & D and mass production ,NR-V2X The industrialization of direct link communication technology will take some time . In an urban environment , The infrastructure construction of Internet of vehicles is in key areas, from test demonstration to pilot application 、 A critical period of widespread deployment across the country . In the highway environment , All parties also actively promote the Internet of vehicles 、 Intelligent transportation system and other related infrastructure construction , Build a vehicle road collaborative service and management system , Relevant vehicle road collaborative Expressway demonstration projects have partially constructed or planned vehicle networking infrastructure .
Multi access edge computing （MEC） And with C-V2X Deep integration of networking technologies represented by , It can support multiple application scenarios to realize online automatic driving .MEC And C-V2X The fusion system can adopt multi-layer system architecture , Usually including roadside MEC Equipment and area MEC platform , Here's the picture 7 Shown . Two types of MEC Relatively independent , It can be flexibly combined according to the different requirements of application scenarios for edge computing “ standard + Customized ” Solutions for .
roadside MEC The equipment has the characteristics of flexible deployment , It can be combined with the direct communication mode to form a service closed loop in the local area , Suitable for high-density deployment in key locations , Realize more comprehensive fusion perception ability and certain fusion decision-making ability .
Area MEC The platform can provide powerful computing for a wider range of users 、 Storage capacity , With routing 、 Shunt and other functions , It can communicate with the user through cellular communication mode 5G Close combination of network , Support higher wireless transmission rate , Achieve more accurate fusion decision-making and fusion control capabilities .
China Communications Standardization Association has set up a project for C-V2X Business MEC Series standard , From requirements and architecture 、 Service capability and open interface 、 Carry out standardization work in cross domain collaboration , Strive to form a standard MEC Cross domain collaboration mechanism for platform north-south data set and user mobile handover , Push MEC Support the maturity of Internet connected automatic driving application scheme .
Next stage ,MEC Dynamic balance technology between computing network and storage network 、 roadside MEC Equipment and area MEC The evaluation methods of the platform will become the focus of research in the industry . 2019 year 9 month ,IMT-2020（5G） The promotion group supported the creation of the first batch 10 individual MEC And C-V2X Fusion test bed . China Telecom 、 China Mobile and other telecom operators , Datang Mobile 、 ZTE and other equipment manufacturers , Chongqing Vehicle Inspection Institute 、 Shanghai Songhong 、 Xiangjiang intelligent and other detection and operators , Chinese express 、 sound of dripping water 、 Autopilot providers such as tus cloud control took the lead in participating in the construction of the test bed .
From the test bed MEC By type ,8 A test bed was built in the area MEC platform , all 10 Each test bed project is planned to build a roadside MEC equipment .
From the function of the test bed , All test beds are planned with fusion sensing function for networked automatic driving ,5 A test-bed project designs the decision-making or control function for automatic driving . At present , Suzhou 、 Beijing 、 Chongqing 、 Many test bed projects in Shanghai and other places have made remarkable progress , be based on MEC Roadside data fusion processing based on 、 Remote driving and other functions have been fully verified .Four 、 The challenge of networked automatic driving
Internet connected automatic driving has become a clear technical main line for the development of high-level automatic driving in China ,“ Internet connection ” It is an important way to strengthen the roadside infrastructure to enable the vehicle end “ The Conduit ”, It's also driving “ Smart cars ” And “ The way of wisdom ” Deep integration of supporting technology . But from the maturity of technology and industrial development , In depth support of Internet connection, automatic driving still faces technical integration 、 Infrastructure construction 、 And the difficulties and challenges in business operation mode .
The technical system of deep collaboration of Netcom still needs to be improved , The roadside infrastructure has not yet established the concept of functional safety level .
First, it is difficult to establish roadside message acceptance mechanism . At present , Vehicle factory 、 Component manufacturers and Internet companies generally believe that roadside sources can only be treated the same as on-board sensors , That is, autonomous vehicles cannot directly accept external information , The on-board sensor is still required to be combined with roadside information input , Fusion perception judgment , Roadside sensor as redundant source , To some extent, it can solve the problem of common cause failure . In the future, reliability level can be added to roadside messages to ensure the reliability of messages , for example , Daimler proposes to add vehicle safety integrity level to roadside transmission information or certification information （ASIL） Or safety integrity level （SIL） Information to ensure the credibility of the message .
Second, the reliability of transmission channel is difficult to guarantee . Need for network automatic driving 5G The network provides a lot of bandwidth 、 Ultra high reliability and low delay 、 Wide connection communication environment , But the quality of wireless channel is often blocked 、 scattering 、 Multipath fading and other factors have a greater impact , Lead to delay 、 It is difficult to guarantee the reliability of roadside message transmission .
Third, cross industry 、 Cross regional digital identity authentication has not been coordinated and unified . Cars and cars 、 Trust between car and road , It is a necessary condition for fully automatic driving relying on network technology , At present, the industry adopts PKI （PKI） The digital identity authentication mechanism of , For vehicular communication equipment 、 The roadside communication equipment issues legal digital certificates , Realize the identity authentication of communication process . From the perspective of industry and local management , Different industry authorities are needed 、 Establish a collaborative and unified digital identity authentication mechanism in different regions , Maintain the digital certificate trust relationship of collaborative mutual recognition .
Fourth, the definition of roadside function safety consistent with the vehicle end is not clear . Traditional automobile enterprises still have doubts about the willingness to deeply integrate the networking technology into the vehicle R & D iteration , The key is that the roadside infrastructure lacks the functional safety and expected functional safety system that matches the vehicle end , It is difficult to establish an accident responsibility identification mechanism for ICV . Based on functional safety and SOTIF The safety evaluation method of this project , Actively explore the functional level requirements and safety definition standards of roadside infrastructure , Or it will promote the deep integration of vehicle and road cooperation .
Infrastructure construction planning is not clear , Barriers to industrial data interoperability .
First, the participants are diversified , The construction and operation mode is not clear . Vehicle networking cross industry 、 The cross domain attribute determines the diversification of industrial participants , It directly leads to the fragmentation of construction and operation . At present, the government is wholly-owned / joint venture 、 Highway owners 、 Operators and other participants have their own advantages and disadvantages in construction and operation , However, they all face the problem of unclear operation mode , In the future, the business form and business model of the Internet of vehicles still need to be explored by the government and the industry .
Second, the scale of infrastructure investment is large , The construction planning path is not clear . Internet connected automatic driving relies on the leapfrog improvement of roadside infrastructure coverage and vehicle terminal penetration , But roadside infrastructure involves many kinds 、 The industry is widely distributed 、 Large scale of investment , There is uncertainty about the return on investment 、 Safety responsibility risk and other issues . At present, some cities or high-speed road sections are undergoing intelligent networking transformation , There is also a lack of unified engineering construction scheme and consideration of the overall layout of transportation . meanwhile , Vehicle terminal penetration 、 The density of roadside facilities is low , Can't support the whole time 、 All elements of road traffic information perception , It is difficult to support the landing of all kinds of autopilot applications .
Third, it is difficult to establish industrial ecology , We need to break down barriers to data connectivity . All kinds of infrastructure belong to different construction subjects , The data collected belong to different enterprises 、 Different authorities , There is bound to be information islands . Realize the interconnection of devices 、 Platform data exchange , On the one hand, it needs cross industry 、 Cross departmental coordination , Break the barriers of industry platform management ; On the other hand , It is urgent to improve the equipment communication interface 、 System platform interface 、 Standardization system for message consistency, etc .
The commercial operation mode of Internet connected automatic driving is still in the design and exploration stage , Supporting policies and regulations need to be improved . One is the lack of public service “ Killer level ” application . Networking applications are gradually oriented to traffic safety and efficiency scenarios , Prompt through early warning information 、 The speed of the car / Improve traffic efficiency by means of path guidance , However, passenger car users do not respond strongly to such businesses , Rigid demand unknown . Even for some relatively high-value Internet applications , Such as signal light information push 、 Intersection collision warning, etc , As the infrastructure deployment has not yet reached the urban scale coverage ,
Low penetration of vehicle terminal , Resulting in reduced user experience . Second, supporting policies and regulations need to be improved , Restrict the transformation of demonstration application to commercial operation . Networked automatic driving depends on the support of high-precision map and positioning , Both of them will be subject to the management and constraints of Surveying and mapping related laws and regulations , The map still needs to be further refined 、 Crowdsourcing mapping 、 original GPS Requirements of acquisition and other links . Besides , For autonomous taxis 、 Independent valet parking （AVP） And other scenarios with relatively clear business models , Supporting safety 、 Road traffic regulations still need to be improved . At present, Beijing 、 Shanghai 、 Guangzhou and other cities have opened trial operation areas for autonomous taxis , However, there is still a gap with the real commercial operation .5、 ... and 、 Current situation and Prospect of collaborative development policy of Internet connected automatic driving
2020 year 3 month , The U.S. Department of transportation issued 《ITS strategic planning （2020-2025）》, Pay more attention to the research and development of automatic driving and network security . Besides , The Ministry of communications takes policy means as the development direction of automatic driving 、 Resource integration 、 Provide government communication channels in terms of supervision mode ; On the other hand, with “ Immunity ” For the main means , Provide policy protection for exploring automatic driving operation on the road ,Nuro During the two-year test period, up to 5000 Exemption for vehicles . In terms of network collaboration , The Ministry of communications has always encouraged enterprises to use 5.9GHz To improve the ability of automatic driving , However, from the perspective of the technical route for enterprises to develop automatic driving , At present, it still focuses on single vehicle intelligence . It can also be seen from this , Does autopilot develop and infrastructure capabilities through Internet coupling 、 Enterprise capability is strongly related to industrial environment . The reasons why American industry focuses on single vehicle intelligence for automatic driving research, development and deployment ： First, the deployment of networking facilities is insufficient , Affect the effect of automatic driving ; Second, the interstate inconsistency of road traffic facilities , Impact on cross state implementation and interoperability of Technology .
The EU is guided by the strategic framework , Promote the development of automatic driving industry , Attach great importance to the coordinated development of vehicle intelligence and networking . Laws and regulations , On the one hand, solve the obstacles to industrial development , On the other hand, it takes automobile enterprises as the core force , Steadily promote the application of autopilot landing . The EU continues to release the strategic plan of online automatic driving , Road map for sound development , The goal is ：2020 Through cloud computing in 、IoT、 Big data and V2X Promote the development of Internet connected automatic driving ;2022 In, netlink autopilot realized open data interaction with big data trusted platform ;2025 Next generation V2X promote L4 Self driving ability . Germany launches autopilot in the capital Berlin , Installed along the line 100 More than one sensor , Real time collection of road traffic information , And send it to the cloud for artificial intelligence integration at the same time . With the cooperation of the system , Autonomous vehicles can sense 400 Prepare vehicles to leave within meters and respond in time .
The Japanese government has paved the way for the commercial use of autonomous driving from legislation to policy .2019 year 5 month , Japan passed 《 Road transport vehicle law 》 Amendment , And in 2020 year 4 Month of formal implementation , The goal is to promote the commercialization and popularization of automatic driving technology . The Japanese government originally planned to 2020 The Olympic Games are regarded as an important opportunity to show Japan's leading technology . The Korean government paid attention to bicycle intelligence in the early stage , On 2019 In, we began to actively promote the construction of automatic driving infrastructure , Closed test site Kcity Realized 5G Full network coverage , And plans to establish the world's first 5G Automatic driving test field based on Network . Overall objectives , The Korean government hopes 2027 Relevant communication facilities will be built in 、 Precision Map 、 Traffic control 、 Roads and other infrastructure , It is necessary to make autopilot commercially run on major roads across the country. .
2020 year , All government parties actively strengthen top-level normative coordination , Create a good environment for industrial development .2020 year 2 month , The national development and Reform Commission and other 11 ministries and commissions jointly released 《 Smart car innovation and development strategy 》, Around the development of intelligent vehicles, it is clearly proposed to build an advanced and complete intelligent vehicle infrastructure system , Aim to 2025 In, the construction of intelligent transportation system and smart city related facilities made positive progress , Vehicle wireless communication network （LTE-V2X etc. ） Achieve area coverage , New generation vehicle wireless communication network （5G-V2X） In some cities 、 The expressway is gradually applied , Full coverage of high-precision spatio-temporal benchmark service network .2020 year 3 month , Issued by the Ministry of industry and information technology 《 About pushing 5G Notice to accelerate development 》, Propose to promote “5G+ Car networking ” Coordinated development . Promote the integration of the Internet of vehicles into the national new information infrastructure construction project , promote LTE-V2X Scale deployment . Build a national pilot area for vehicle networking , Enrich application scenarios , Explore and improve the business model . combination 5G Commercial deployment , Guide key areas to plan ahead , Strengthen cross departmental collaboration , Push 5G、LTE-V2X Included in smart city 、 Important communication standards and protocols for intelligent transportation construction . Develop 5G-V2X Standard development and R & D verification .2020 year 4 month , The national development and Reform Commission defines the scope of new infrastructure , Including integrated infrastructure such as intelligent transportation infrastructure , Among them, vehicle networking is an important guarantee for the intellectualization and networking of traditional transportation infrastructure .
2020 year 8 month , Issued by the Ministry of transport 《 Guidance on promoting new infrastructure in Jiatong transportation field 》, It is proposed to build an integrated and efficient intelligent transportation infrastructure , Improve industry innovation infrastructure , The emphasis is on helping 5G And other information infrastructure construction .2020 year 10 month , The general office of the State Council officially issued 《 New energy automobile industry development plan （2021－2035 year ）》, Clear will promote new energy vehicles and energy 、 traffic 、 Comprehensive and deep integration of information and communication . Accelerate the construction of front-end information collection 、 Edge distributed computing 、 A new intelligent transportation management and control system with centralized management and control in the cloud . Give full play to the basic advantages of cellular communication network , Wireless communication 、 Supported by information and communication technologies such as positioning and navigation , Promote vehicle and road traffic 、 Extensive interconnection and data interaction of information and communication infrastructure , Provide guarantee for multi-level linkage automatic driving control decision-making and application services . Coordinate and promote the construction of intelligent road network facilities . Build a wireless communication network supporting vehicle road cooperation . Accelerate the upgrading of vehicle wireless communication technology , Continuously meet high-level automatic driving intelligent networked vehicle applications . Besides , Relevant departments for mapping 、 Intelligent network connected vehicle test and so on formulated the corresponding specification successively , Include 《 Management measures for surveying and mapping qualification （ Solicitation draft ）》《 Intelligent network connected vehicle road test and demonstration application management specification 》（ Solicitation draft ）》 etc. .
Under the guidance of national policies , All localities have issued guidance on actively promoting industrial development in combination with their own development needs and basic advantages . Establishment and formation of Jiangsu Province 《 Breakdown of key tasks for the development of Internet of vehicles industry in Jiangsu Province （2020 year -2021 year ）》, Determine total 7 class 65 Tasks , It defines the action guide for promoting the development of the Internet of vehicles industry . Issued by tianjin 《 Tianjin Internet of vehicles （ Intelligent Internet car ） Industrial Development Action Plan 》, Put forward to speed up LTE Network upgrading and transformation 5G Large scale deployment , promote LTE-V2X Network coverage level , Construction based on LTE-V2X Internet of vehicles service platform for key technologies of wireless communication . Changsha releases intelligent automobile industry “ Torch Plan ” and “ Sheep project ”,“ Torch Plan ” Focus on supporting relevant enterprises of intelligent network vehicle road cooperation , And give financial incentives and policy support ,“ Sheep project ” By adding on-board intelligent terminal products to key vehicles 、 Intelligent transformation of urban roads , Construction of intelligent networking cloud platform 、 Demonstration application of intelligent networking in specific scenarios, etc , Strive to 2022 year , Build a platform for the integrated development of intelligent vehicles and intelligent transportation “ Changsha model ”. Issued by Guangzhou 《 Guangzhou's three-year action plan to accelerate the development of new digital infrastructure (2020—2022 year )》, Carry out trial operation of direct communication frequency band of Internet of vehicles , combination 5G commercial , promote C-V2X Scale deployment , Launch mass production intelligent networked vehicles , Promote the establishment of inter city cooperation in Guangdong, Hong Kong and Macao Dawan district 、 Cross border testing and application collaboration mechanism , Construction of vehicle road collaborative test network around Dawan District . Issued by Beijing 《 Action plan for innovation and development of Beijing intelligent Internet vehicle (2019 year -2022 year )》, It is proposed to deploy the pilot reconstruction project of intelligent road network , Planning and construction of satellite ground augmentation station 、LTE-V2X、NR-V2X Roadside units . Issued by Sichuan Province 《 Notice on promoting the development of intelligent networked automobile industry 》, Put forward to promote LTE Network transformation and upgrading , Meet the large-scale application of Internet of vehicles . hebei 、 Shanghai 、 Deqing, Zhejiang 、 Shenzhen and other provinces and cities have also issued relevant promotion policies .
Ministry of industry and information technology 、 The Ministry of Public Security 、 The Ministry of transport and other departments should work together to promote cross sectoral cooperation and ministerial provincial cooperation , Support the Internet of vehicles （ Intelligent Internet car ） Demonstration area 、 Pilot zone construction .2019 year 5 The Ministry of industry and information technology approved the establishment of Jiangsu （ wuxi ） The pilot area of the Internet of vehicles . Achieve scale deployment C-V2X The Internet 、 Roadside units , Assemble a certain scale of on-board terminal , Complete the functional transformation of transportation facilities in key areas and the improvement of core system capacity , Enrich Internet of vehicles application scenarios .2019 year 12 Monthly approval from the Ministry of industry and information technology supports the establishment of Tianjin （ Xiqing ） National vehicle networking pilot area . Play in standards bodies 、 Advantages of test environment , Actively explore a new mode of cross industry standardization , Scale deployment of cellular Internet of vehicles C-V2X The Internet , Define the installation scheme of the Internet of vehicles communication terminal , Clarify the main body and responsibilities of vehicle networking operation , Building open integration 、 Industrial ecology of innovation and development .2020 year 9 The Ministry of industry and information technology approved the establishment of Hunan （ Changsha ） The pilot area of the Internet of vehicles , On key highways 、 Urban road scale deployment of cellular Internet of vehicles C-V2X The Internet , combination 5G And smart city construction , Complete the functional transformation of transportation facilities in key areas and the improvement of core system capacity , Drive the scale deployment of the whole road network .2020 year 9 month , Beijing Yizhuang released the world's first high-level automatic driving demonstration zone of Netcom cloud control , To support the L4 Large scale operation of high-level autonomous vehicles at and above . Plan to 2022 year , Will complete “ The way of wisdom 、 Smart cars 、 Real time cloud 、 Reliable nets and accurate graphs ” Five systems construction , Get through the key links of technology and management of network connected cloud controlled automatic driving , Form a city level engineering test platform , Finally, realize the unmanned logistics of expressway 、L4 Class a self driving taxi 、 Intelligent Internet bus 、 Independent valet parking and other high-level application scenarios .
Cross industry organizations carry out end-to-end cooperation 、 Large scale test verification , Solve cross industry problems 、 Key issues of cross regional interconnection .2020 year 10 month ,IMT-2020（5G） Propulsion group C-V2X Working group 、 China intelligent network automotive industry innovation alliance and other parties jointly organized C-V2X“ The new four spans ” Large scale pilot application demonstration activities in Jinan , Promote large-scale testing 、 Compliance test of high-precision map and high-precision positioning , attracted 40 More than domestic and foreign vehicle enterprises 、40 More than one terminal enterprise 、10 More than one chip module company 、20 More than information security enterprises 、5 Home map business and 5 Home location service providers, etc . stay “ On a large scale ” In the test , Set the 180 Real working C-V2X On board communication unit and roadside communication unit , Carry out performance and application function test in large-scale application environment , Verify the Internet of vehicles C-V2X System large-scale operation capability . stay “ Interconnection ” In demonstration , Increase the application of high-precision map and positioning , By deflecting first and then encrypting , Explore C-V2X Technical solutions to map and location regulations .
All parties in the industry should strengthen coordination , From closed to open 、 From unmanned to manned 、 From the city to the highway , Test demonstration continues to advance . By 2020 year 9 month , There are already 26 Provinces and cities have successively issued the detailed rules for the implementation of the road test of the intelligent network connected vehicles and designated the road test sections of the intelligent network connected vehicles , Provinces and cities have distributed about 455 Zhang intelligent network vehicle road test license plate , Covering vehicle manufacturing enterprises 、ICT Enterprises 、 Start-ups 、 Scientific research institutions, etc . Vehicle manufacturing enterprises have 30 Several companies have obtained test licenses , Baidu 、 Huawei 、 tencent 、 Alibaba, etc ICT Enterprises have an absolute advantage in the number of test licenses . for example , Baidu Apollo The test fleet has reached 500 Vehicle class , The total test mileage exceeds 600 Thousands of kilometers , The total number of test licenses obtained exceeds 150 Zhang . Baidu's first batch of self driving taxis 2019 year 9 Trial operation in Changsha in May , Tens of thousands of safe passenger trips have been realized in Changsha .2020 year 1 month , Yan Chong Expressway took the lead in developing Expressway scene 80 Kilometers per hour L4 Level automatic driving and vehicle road collaborative testing based on cellular network technology , And the demonstration of car following in the automatic driving queue .
China should seize the rare opportunity of historical development , Adhere to the automatic driving path of networked collaborative development , The government 、 industry 、 Enterprises cooperate with each other , Actively build the environment for industrial development , Promote the maturity of automatic driving , Drive and influence the formation of a global consensus .
Actively build the environment for industrial development , Collaborative promotion of autopilot maturity . Jointly issue top-level planning 、 Clear path selection . Make good use of the organization and coordination role of the special committee for the development of Internet of vehicles industry of the leading group for the construction of a national manufacturing power , Strengthen the car 、 Information communication 、 traffic 、 Collaboration between electronic and other cross industry fields , Jointly formulate a development road map , Introduction of top-level planning . Building intelligent road infrastructure . Promote the deep integration of information communication and artificial intelligence technology with road traffic infrastructure , Improve the intelligent level of road infrastructure .
Deploy edge computing capabilities in key areas 、 Deploy roadside sensing devices, etc .
Building communication network infrastructure . Collaborative construction is based on LTE-V2X、5G Network infrastructure of wireless communication technology , Improve its coverage on major highways and major urban roads .
Improve the data access specification of roadside communication equipment or base station , Improve it with road infrastructure 、 Integrated access capability of intelligent management and control facilities .
Strengthen the construction of laws, regulations, mechanisms and systems .
Plan ahead of time , Pay close attention to studying and solving the problems of laws and regulations that restrict the development of automatic driving industry , Build a system of policies and regulations for the development of automatic driving industry in line with China's national conditions , Promote timely formulation and revision of policies and regulations restricting industrial development .
Give play to the advantages of information and communication , Promote industrial level coordination . Tackling key core technologies .
For vehicle computing chip 、 Car operating system 、 sensor 、 Key areas such as data processing , Increase investment in innovation resources , Accelerate the construction of enterprises as the main body 、 Market oriented 、 Technology innovation system of deep integration of industry, University and research . Grasp the synergy of common key technologies . For the industry key common technologies that are difficult to be solved by a single enterprise , We should pay attention to the choice of technology path 、 Technology deployment schedule 、 Technology standardization and open interface design , For example, it serves automatic driving “ people - vehicle - road - cloud ” Communication network architecture .
Strengthen the car 、 signal communication 、 traffic 、 Public Security 、 Integration of energy and other standards . Explore a new mode of cross industry standard cooperation , On platform interface 、 Application services, etc. , Jointly build and improve the Internet of vehicles （ Intelligent Internet car ） Industrial standard system . Promote the construction and management of big data and cloud platform .
Promote the interconnection of various platforms , Promote cross platform information interaction and data sharing , Build a market-oriented mechanism for data use and maintenance . Encourage the construction of comprehensive big data and cloud platform , Support the scale development and continuous innovation of Internet of vehicles applications .
Try first , In stages 、 step 、 Deployment and implementation by scenario .
Strengthen ministerial coordination and ministerial and provincial cooperation , Encourage all parties in the industry chain to participate , Building Simulation 、 closed 、 Semi open intelligent networked vehicle 、 Demonstration areas and test bases for Internet of vehicles applications .
Give full play to the role of local vehicle networking demonstration areas , Strengthen cross departmental and provincial coordination , Promote the deepening of cooperation among demonstration areas 、 Data sharing 、 Test mutual recognition , Accelerate the promotion of demonstration applications .
Promote the construction and upgrading of intelligent road infrastructure in stages . From local pilot to regional coverage , Priority should be given to key cities with conditions 、 Reconstruction of road infrastructure on highway sections , And then gradually expand to the regional level 、 City level scope .
From basic information interconnection to perception 、 Computing expansion capability extension , Give priority to the networking of traffic lights and existing road traffic signs , And then gradually promote the edge computing platform 、 radar 、 Visual cameras, etc 、 Calculate roadside facility deployment .
Step by step to promote the test, verification and application demonstration of networked automatic driving . Encourage all parties in the industry chain to participate , Application of automatic driving for Internet connection , Building simulation 、 closed 、 Semi open vehicle networking demonstration area or pilot area , Deepen cooperation 、 Data sharing 、 Test mutual recognition , Accelerate the promotion of demonstration applications .
Step by step From regional to global 、 From sub functions to full-featured big data and cloud platform construction , Promote the interoperability of various platforms 、 Information interconnection and data sharing , Accelerate the construction of application service system based on Internet connected automatic driving .
Carry out the application demonstration of Internet connected automatic driving in different scenarios , Gradually explore the maturity of construction and operation mode . Promote the application of self driving taxis and automatic buses in characteristic towns 、 Demonstration promotion and large-scale application of industrial parks and smart cities , Improve residents' life and travel experience in an all-round way .
In specific scenarios where there is clear operational dominance , Such as coal mines 、 port 、 Docks, etc , Explore the integrated application of global automatic driving technology .
In the condition of the highway to carry out vehicle formation driving and other application pilot , Serving the trunk line logistics .
版权声明：本文为[Love is parallel to the world]所创，转载请带上原文链接，感谢。 https://car.inotgo.com/2021/08/20210808144240349z.html