WO2024125529A1 - 基于中心化计算的全域自动驾驶控制系统和方法 - Google Patents
基于中心化计算的全域自动驾驶控制系统和方法 Download PDFInfo
- Publication number
- WO2024125529A1 WO2024125529A1 PCT/CN2023/138258 CN2023138258W WO2024125529A1 WO 2024125529 A1 WO2024125529 A1 WO 2024125529A1 CN 2023138258 W CN2023138258 W CN 2023138258W WO 2024125529 A1 WO2024125529 A1 WO 2024125529A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- vehicle terminal
- computing center
- current
- grid
- computing
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 42
- 238000013439 planning Methods 0.000 claims description 38
- 238000004364 calculation method Methods 0.000 claims description 37
- 238000004891 communication Methods 0.000 claims description 28
- 230000007704 transition Effects 0.000 claims description 28
- 238000012546 transfer Methods 0.000 claims description 14
- 238000005265 energy consumption Methods 0.000 claims description 12
- 230000008569 process Effects 0.000 claims description 12
- 230000001934 delay Effects 0.000 claims description 8
- 238000005516 engineering process Methods 0.000 description 6
- 238000012986 modification Methods 0.000 description 4
- 230000004048 modification Effects 0.000 description 4
- 238000005457 optimization Methods 0.000 description 4
- 101001121408 Homo sapiens L-amino-acid oxidase Proteins 0.000 description 3
- 102100026388 L-amino-acid oxidase Human genes 0.000 description 3
- 238000013475 authorization Methods 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000007689 inspection Methods 0.000 description 3
- 101100233916 Saccharomyces cerevisiae (strain ATCC 204508 / S288c) KAR5 gene Proteins 0.000 description 2
- 238000003745 diagnosis Methods 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 101100012902 Saccharomyces cerevisiae (strain ATCC 204508 / S288c) FIG2 gene Proteins 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000004888 barrier function Effects 0.000 description 1
- 238000011217 control strategy Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013523 data management Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000004806 packaging method and process Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 239000000523 sample Substances 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/09—Arrangements for giving variable traffic instructions
- G08G1/0962—Arrangements for giving variable traffic instructions having an indicator mounted inside the vehicle, e.g. giving voice messages
- G08G1/0967—Systems involving transmission of highway information, e.g. weather, speed limits
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/09—Arrangements for giving variable traffic instructions
- G08G1/0962—Arrangements for giving variable traffic instructions having an indicator mounted inside the vehicle, e.g. giving voice messages
- G08G1/0968—Systems involving transmission of navigation instructions to the vehicle
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/12—Protocols specially adapted for proprietary or special-purpose networking environments, e.g. medical networks, sensor networks, networks in vehicles or remote metering networks
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/50—Network services
- H04L67/52—Network services specially adapted for the location of the user terminal
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
Definitions
- Embodiments of the present application relate to a global autonomous driving control system and method based on centralized computing.
- the embodiments of the present application provide a global autonomous driving control system and method based on centralized computing.
- One or more embodiments of the present application provide the following technical solutions:
- a global autonomous driving control system based on centralized computing wherein the area covered by the system is divided into multiple grids, and the system includes a root server and multiple computing centers; each of the grids includes at least one computing center; wherein:
- the root server is configured to receive the travel demand sent by the vehicle terminal, plan a global path according to the travel demand; divide the global path into multiple subtasks according to the grids covered by the global path, and send them to the computing center of each grid respectively;
- the computing center is configured to plan a local path when the vehicle terminal is traveling in the grid where the computing center is located, and control the vehicle terminal to travel along the local path to complete the assigned subtask.
- the system further includes a plurality of computing center intermediate layers corresponding one to one with the plurality of grids; each computing center intermediate layer is connected to one or more computing centers in the grid, and each computing center intermediate layer is connected to the root server.
- the system further includes a plurality of wireless communication base stations, each grid includes at least one of the wireless communication base stations, and each of the wireless communication base stations is connected to the middle layer of the computing center in the grid.
- the travel demand includes destination information; and the root server planning a global path according to the travel demand includes:
- the computing center planning a local path includes:
- the computing center transfers the vehicle control to the computing center of the grid where the next subtask is located.
- the transfer process is:
- the current computing center and the next computing center both plan a local path for the vehicle terminal and send actuator input parameters to the vehicle terminal;
- the vehicle terminal when just entering the transition area, still keeps executing the actuator input parameters issued by the computing center of the current grid, and after driving the set distance, preferentially executes the actuator input parameters that arrive first;
- the vehicle terminal monitors the message delays sent by the two computing centers. When the packet loss rate and network delay of the message from the next computing center are less than or equal to the message from the current computing center and remain stable for a period of time, feedback information is sent to the current computing center, and the current computing center no longer controls the vehicle terminal.
- One or more embodiments of the present application provide a root server for global autonomous driving control, where a control area is divided into a plurality of grids, each of which includes at least one computing center; the root server maintains a communication connection with each computing center through a computing center intermediate layer, and the root server is configured to perform the following steps:
- the global path is divided into multiple subtasks, which are sent to the computing center of each grid respectively.
- the travel demand includes destination information; and the root server planning a global path according to the travel demand includes:
- the root server also obtains the data request sent by the computing center when executing a subtask, and uses the global path, current position and current driving status of all vehicle terminals in the grid where the computing center is located to perform local path planning for the vehicle terminal corresponding to the current subtask.
- One or more embodiments provide a computing center that is communicatively connected to the root server, and the computing center is configured to perform the following steps:
- a local path is planned, and the vehicle terminal is controlled to travel along the local path to complete the assigned subtasks.
- planning a local path includes:
- the vehicle control is transferred to the computing center of the grid where the next subtask is located.
- the transfer process is:
- the current computing center and the next computing center both plan a local path for the vehicle terminal and send actuator input parameters to the vehicle terminal;
- the vehicle terminal when just entering the transition area, still keeps executing the actuator input parameters issued by the computing center of the current grid, and after driving the set distance, preferentially executes the actuator input parameters that arrive first;
- the vehicle terminal monitors the message delays sent by the two computing centers. When the packet loss rate and network delay of the message from the next computing center are less than or equal to the message from the current computing center and remain stable for a period of time, feedback information is sent to the current computing center, and the current computing center no longer controls the vehicle terminal.
- One or more embodiments of the present application provide a global autonomous driving control method based on centralized computing, which is applied to a global autonomous driving control system based on centralized computing, wherein the system includes a root server and multiple computing centers; the area covered by the system is divided into multiple grids, each of which includes at least one computing center; wherein,
- the root server receives the travel demand sent by the vehicle terminal, and plans a global path according to the travel demand; divides the global path into multiple subtasks according to the grids covered by the global path, and sends the multiple subtasks to the computing center of each grid respectively;
- the computing center plans a local path and controls the vehicle terminal to travel along the local path to complete the assigned subtask.
- sending the multiple subtasks to the computing centers of the respective grids includes:
- the root server sends the multiple subtasks to the computing center middle layer of the corresponding grid respectively, each of the multiple subtasks corresponds to a grid, and a computing center middle layer is set in each grid;
- the computing center intermediate layer After receiving the subtask, the computing center intermediate layer sends the subtask to at least one computing center connected to the computing center intermediate layer.
- the computing center plans a local path and controls the vehicle terminal to travel along the local path to complete the assigned subtasks, including:
- the computing center determines all current vehicle terminals within the grid where the computing center is located;
- the computing center For each of all current vehicle terminals, the computing center obtains the subtask, current position and current driving status of the vehicle terminal in the grid where the computing center is located; based on the preset intersection position, the position where the vehicle enters and exits the grid as nodes, the subtask corresponding to the vehicle terminal is segmented to obtain the nodes corresponding to the subtask; based on the nodes corresponding to the subtask, the current position and current driving status, a local path is planned for the vehicle terminal, and the vehicle terminal is controlled to travel along the local path to exit the position of the current grid, thereby completing the assigned subtask corresponding to the current grid.
- planning a local path for the vehicle terminal according to the node corresponding to the subtask, the node corresponding to the subtask, the current position and the current driving state includes:
- the theoretical arrival time of each vehicle terminal to the next node is calculated, the order of each vehicle terminal is arranged according to the theoretical arrival time, and the theoretical arrival time of each vehicle terminal is updated;
- the position of the vehicle terminal is obtained in real time, the actuator input parameters of the vehicle terminal are calculated according to the theoretical arrival time, the vehicle terminal is controlled to arrive at the next node within the theoretical time according to the actuator input parameters, and the current position of the vehicle terminal is updated;
- the method further comprises:
- the computing center transfers the vehicle control to the computing center of the grid where the next subtask is located until all subtasks corresponding to the global path are completed.
- the computing center transfers the vehicle control right to the computing center of the grid where the next subtask is located, including:
- the current computing center and the next computing center both plan a local path for the vehicle terminal and send actuator input parameters to the vehicle terminal;
- the vehicle receives actuator input parameters sent by the current computing center and the next computing center respectively, and the vehicle terminal determines the position of the vehicle terminal in the transition area, and selects the actuator input parameters received from the received actuator input parameters according to the position.
- Target actuator parameters so as to control the target vehicle terminal through the target actuator parameters.
- selecting target actuator parameters from received actuator input parameters according to the position includes:
- the vehicle terminal uses the actuator input parameters sent by the computing center of the current grid as target actuator parameters;
- the vehicle terminal uses the actuator input parameters that first arrive at the vehicle terminal as the target actuator parameters.
- the method further comprises:
- the vehicle terminal monitors the message delays sent by the two computing centers.
- the packet loss rate and network delay from the next computing center are less than or equal to the packet loss rate and network delay of the current computing center, and remain stable for a period of time, feedback information is sent to the current computing center so that the current computing center no longer controls the vehicle terminal.
- FIG1 is an architecture diagram of a global autonomous driving control system based on centralized computing provided in an embodiment of the present application
- FIG2 is a functional framework diagram of a root server, a computing center, and a vehicle terminal provided in an embodiment of the present application;
- FIG3 is a flow chart of a global autonomous driving control method based on centralized computing provided in an embodiment of the present application
- Figure 4 is a flowchart of a global autonomous driving control method based on centralized computing provided in an embodiment of the present application.
- the embodiment of the present application provides a global autonomous driving control system based on centralized computing, which is used to control autonomous driving cars in closed roads within the jurisdiction of the system.
- Conventional road access is not allowed at any node of the closed road in the embodiment of the present application, and physical barriers are set at the edge of the closed road to prevent any vehicles out of the control of the system from entering the closed road, and pedestrians are required to move within the designated area divided by road marking lines.
- the closed roads within the jurisdiction of the system are the area covered by the system, which will be pre-divided into multiple grid pairs of areas, and the area corresponding to each grid will be described as a grid in the following.
- the self-driving car in the embodiment of the present application is a car controlled by the self-driving system network without human takeover.
- the car completes the driving operation performed by the controller by uploading the vehicle status (vehicle speed, yaw angle, steering angle, acceleration, centimeter-level positioning coordinates, etc.) in real time and receiving the takeover instruction from the computing center.
- the vehicle needs to include a controller and sensors to collect environmental, vehicle and pedestrian information, such as ultrasonic probes, cameras, millimeter-wave radars, etc. On the one hand, it provides perception data of the surrounding environment to optimize system decisions in scenes such as intersections, narrow, parking, and poor signals; on the other hand, it can be used as a safety redundant backup to enable the vehicle to achieve local self-driving functions.
- a global autonomous driving control system based on centralized computing includes a root server and multiple computing centers (9 computing centers are taken as an example in Figure 1).
- each grid includes at least one computing center (3 computing centers are set in each grid in Figure 1), and all computing centers in the grid are connected to the root server.
- the global autonomous driving control system further includes at least one computing center intermediate layer (three computing center intermediate layers are taken as an example in FIG1).
- each grid includes a computing center intermediate layer, each computing center intermediate layer is connected to the root server, and each computing center intermediate layer is connected to at least one computing center in the grid (three computing centers in each grid are taken as an example in FIG1), and the root server and multiple computing centers communicate via the computing center intermediate layer.
- the root server and the computing center middle layer communicate through wired or wireless means.
- the root server, computing center and computing center middle layer are all computer devices with processors, storage and communication interfaces.
- the global autonomous driving control system also includes multiple wireless communication base stations, each grid includes at least one wireless communication base station, and each wireless communication base station is connected to the middle layer of the computing center in the grid.
- a wireless communication base station is deployed in the middle of the road (such as a street lamp), and the wireless communication base station is not only used to connect the wireless communication device to the operator network, but also connected to the middle layer of the computing center in the grid of the geographical location to reduce the communication delay between the autonomous driving vehicle and the computing center.
- the wireless communication base station is used to establish communication between the vehicle terminal and the root server and the computing center.
- the deployment density of the communication base station is determined by the communication standard and medium, and is not limited to 5G, 6G or quantum communication. The communication speed and stability directly determine the overall performance upper limit of the system.
- the root server is composed of a computer group with super-large computing power, which is responsible for vehicle path planning, road data analysis, regional traffic control, and abnormal event processing for the entire closed road corresponding to the area covered by the root server, so as to coordinate and dispatch traffic resources, transfer and process traffic flow information across grid computing centers within the coverage of the root server.
- a high-precision map database, vehicle database, and road infrastructure database are also established on this root server.
- the computing center is, for example, a cloud server unit.
- the autonomous driving system operated by the computing center models and analyzes the vehicles in the area under its jurisdiction, calculates the driving trajectory of the vehicles in the area under its jurisdiction and the input parameters of the vehicle chassis actuators in each time period to reach the target state; on the other hand, the computing center's calculation results are synchronized to the autonomous driving network root server in real time through the computing center's middle layer, and the traffic flow control strategy of the root server is received and executed.
- the computing center also monitors and reports the operating status of base stations and road-side infrastructure.
- the computing center is a computing cluster isolated from the outside world.
- the computing center establishes communication with other devices through the computing center middle layer.
- the computing center middle layer is responsible for monitoring the working status of the computing center and scheduling the computing resources of each computing center in the grid. At the same time, it serves as the external interface of the grid computing center and communicates with the root server and the computing centers of other grids.
- FIG. 2 is a functional framework diagram of a root server, a computing center, and a vehicle terminal provided in an embodiment of the present application.
- the root server includes:
- Basic data management module used to manage map data and road infrastructure in the area
- the vehicle management module is used to manage the terminal information of the autonomous driving vehicles and the corresponding passenger identity information in the area, and is used for identity authentication when driving;
- the driving authorization module is used to receive the identity authentication request sent by the passenger when starting the vehicle terminal, and perform the authentication. If the authentication is passed, the global path planning module is entered;
- the global path planning module is used to receive the travel demand sent by the vehicle terminal and plan the global path according to the travel demand.
- the travel demand includes the destination information; planning the global path according to the travel demand includes:
- the driving path selected by the occupant is used as the global path of path planning, and the global path is divided into multiple subtasks.
- Each subtask generally refers to the task of the vehicle driving within a grid.
- the subtask allocation module is used to divide the global path into multiple subtasks according to the grids covered by the global path, and send them to the computing center middle layer of each grid respectively, and then the computing center middle layer allocates them to the computing center in the grid.
- the subtask tracking module is used to track the actual completion status of all subtasks and evaluate the impact on the global planning based on the results and deviations of task completion to request subtask reallocation.
- the vehicle status management module is used to obtain information such as the current position, current driving status, grid location, and computing center currently controlling the vehicle of all vehicle terminals; and, to receive data requests sent by the computing center when executing a subtask, and use the global path, current position, and current driving status of all current vehicle terminals in the grid where the computing center is located to perform local path planning for the vehicle terminal corresponding to the current subtask.
- Computing Center including:
- the local path planning module is used to plan the local path when the vehicle terminal is driving in the grid where the computing center is located; specifically, it includes:
- the global paths of these vehicles are segmented with the intersection positions and the positions where the vehicles enter and exit the grid as nodes.
- the nodes obtained are generally the nodes corresponding to the intersection positions and the positions where the vehicles enter and exit the grid.
- the positions where the vehicles enter and exit the grid are used to divide the global path into local paths corresponding to the grids, that is, the driving path of the vehicle within a grid; the intersection position is used to divide the path in the same grid.
- the nodes in the same grid the nodes will be arranged according to the order of the vehicle's travel in order to determine the next node at the current location.
- the vehicle driving control module is used to control the vehicle terminal to travel along the local path and complete the assigned subtasks.
- the vehicle control transfer module is used to transfer the vehicle control to the computing center of the grid where the next subtask is located after the computing center completes the assigned subtask.
- the transfer process is as follows:
- both the current computing center and the next computing center plan a local path for the vehicle terminal and send actuator input parameters to the vehicle terminal;
- the vehicle terminal When the vehicle terminal just enters the transition area, it still executes the actuator input parameters issued by the computing center of the current grid. After driving the set distance, it preferentially executes the actuator input parameters that arrive first.
- the vehicle terminal monitors the message delays sent by the two computing centers. When the packet loss rate and network delay of the message from the computing center of the next grid are less than or equal to the message from the computing center of the current grid, and remain stable for a period of time, feedback information is sent to the current computing center, and the current computing center no longer controls the vehicle terminal.
- the packet loss rate and network delay of the computing center remain stable over a period of time, which means that the packet loss rate and network delay are controlled within a certain range respectively to ensure normal communication between the computing center and the vehicle terminal.
- Vehicle terminal including:
- the current status acquisition module is used to obtain the current position and driving status of the vehicle in real time and send them to the root server;
- the control system login module is used to obtain the identity information of the passengers and send it to the root server;
- the execution control module is used to obtain the actuator input parameters sent by the computing center and control the vehicle terminal to travel.
- the system also has emergency handling strategies:
- the system will downgrade from autonomous driving to limited autonomous driving.
- the vehicle's external sensors will perceive the environment, calculate the driving path locally, and automatically drive into a temporary lane to wait for rescue.
- the middle layer allocates resources from other computing centers in the region
- the base station cannot establish communication with the middle layer of the grid, and the root server takes over the entire grid.
- avoidance strategies are executed first, including but not limited to: automatic emergency braking, automatic emergency steering, emergency lane keeping, automatic lane changing, and driving into the emergency lane to wait for rescue.
- the computing center replans node targets and arrival times and updates the order
- the vehicle resumes remote takeover of the autonomous driving network.
- the control method performed by the above-mentioned automatic driving control system includes the following steps:
- Steps for logging into an autonomous driving vehicle When the autonomous driving vehicle is started at a parking space, it obtains the passenger's identity information and sends it to the root server. The root server authenticates the identity information. After the authentication is passed, the autonomous driving vehicle logs into the autonomous driving control system.
- the automatic driving control system includes:
- Biometric authentication of the vehicle owner that is, using biometric recognition technology, such as fingerprint, face, iris, etc., to authenticate the vehicle owner.
- the root server performs a fault check on the data uploaded by the vehicle. If there is no problem, login is allowed. If there is a low-risk fault, conditional login is allowed. If there is a serious fault, login is not allowed.
- Travel request acquisition step After logging into the autonomous driving network, the autonomous driving vehicle obtains the travel request input by the passenger and sends it to the root server.
- the travel request includes the destination information and, more specifically, the final parking area.
- Global path planning step The root server performs global path planning based on the current location and destination information, and decomposes the global path into multiple subtasks according to the grid covered by the global path, and sends the multiple subtasks to the middle layer of the computing center of the corresponding grid.
- the root server performs global path planning including:
- the root server receives a travel request from a vehicle terminal
- the computing center middle layer After receiving the subtask, the computing center middle layer sends it to one of the computing centers connected to it.
- the root server After receiving the travel request, the root server also sends an inquiry message to the vehicle terminal, including whether there is a need for intermediate parking. If the intermediate parking location is obtained, global path planning is performed based on the current location, intermediate parking location and destination information.
- the root server assigns the vehicle control right to the computing center of the grid where the target vehicle terminal is currently located.
- the computing center brings the vehicle driving data into the entire traffic flow model, performs local path planning for the target vehicle terminal in the current grid, and calculates the actuator input parameters to achieve the vehicle's target driving state in real time, controls the target vehicle terminal to complete the subtask in the current grid; and assigns the vehicle control right to the computing center that executes the next subtask.
- the computing center is configured to perform the following steps:
- the terminal position of the target vehicle is acquired in real time, and the actuator input parameters of the vehicle terminal are calculated according to the theoretical arrival time to ensure that the vehicle reaches the next node within the theoretical time.
- the actuator input parameters of the target vehicle terminal are fed back to the computing center, and the computing center substitutes the vehicle data into the model for recalculation based on the offset, and continuously corrects the actuator parameters until the next node is reached.
- the vehicle terminal combines GNSS (Global Navigation Satellite System), IMU (inertia measurement unit) track calculation, wheel encoder, etc. to obtain its own real-time position and feed it back to the computing center.
- GNSS Global Navigation Satellite System
- IMU inertia measurement unit
- the computing center assigns the vehicle control rights to the computing center of the grid where the next subtask is located, and repeats steps (1)-(6) until all subtasks are completed.
- the current computing center and the next computing center both plan a local path for the vehicle terminal and send actuator input parameters to the vehicle terminal.
- the two adjacent base stations receive the data packets transmitted by the vehicle and send them to the middle layer of their respective grid computing centers, which are then sent to the computing center; both computing centers calculate the control parameters based on the vehicle flow in the next grid and the real-time information of the target vehicle terminal, and send the control parameters and the request to take over command to the vehicle terminal.
- Vehicle information is received through the base station, ensuring the real-time data transmission.
- the vehicle terminal receives the calculation results and takeover instructions sent by the two calculation centers. When it just enters the transition area, it still executes the calculation results and takeover instructions sent by the calculation center of the current grid. After traveling a certain distance, it takes precedence over the calculation results and takeover instructions sent by the calculation center of the current grid.
- the executor input parameters that arrive first.
- the vehicle terminal monitors the message delays sent by the two computing centers. When the packet loss rate and network delay of the message from the computing center of the next grid are less than or equal to the message from the computing center of the current grid, and remain stable for a period of time, feedback information is sent to the current computing center. The current computing center no longer controls the vehicle terminal, and the handover is completed.
- the vehicle is taken over by a computing center in the new grid, which executes the subtasks assigned by the root server and repeats the above process.
- the vehicle reaches the destination, enters the parking space, and the autonomous driving process ends.
- an inquiry message is sent to the vehicle terminal after the travel request is obtained, and the inquiry message includes whether there is a parking location preference.
- multiple recommended parking spaces are calculated based on the expected arrival time and parking location preference.
- the time consumption, energy consumption and recommended parking space of each driving path are sent to the vehicle terminal.
- the target parking space is gradually locked from multiple recommended parking spaces by continuously correcting the expected arrival time, and finally the vehicle terminal is controlled to enter the target parking space. It should be noted here that if there are multiple vehicle terminals locking the same target parking space, the root server estimates the arrival time of multiple vehicle terminals and sorts them, giving priority to allocating parking spaces to those who arrive first.
- An embodiment of the present application provides a global autonomous driving control method based on centralized computing, which is applied to a global autonomous driving control system based on centralized computing.
- the root server, the computing center middle layer and the computing center in the global autonomous driving control system based on centralized computing can all execute one or more steps of the global autonomous driving control method based on centralized computing provided in the embodiment of the present application.
- FIG3 is a global autonomous driving control method based on centralized computing provided in an embodiment of the present application, which is applied to a global autonomous driving control system based on centralized computing.
- the method includes:
- the root server receives the travel demand sent by the vehicle terminal, plans a global path according to the travel demand, and divides the global path into multiple subtasks according to the grids covered by the global path.
- the computing center plans a local path and controls the vehicle terminal to travel along the local path to complete the assigned subtasks.
- the embodiment of the present application realizes a distributed computing power layout by geographically gridding the global autonomous driving roads, setting up one or more computing centers in the grids, and setting up a centralized root server, thereby ensuring the real-time nature of control information issuance and vehicle information feedback, and providing a guarantee for the comprehensive popularization of autonomous driving vehicles.
- local path optimization is performed by computing centers in different grids, which is conducive to achieving optimal scheduling of computing power and improving the efficiency of information transmission; in addition, when performing local path planning in each grid, the path of the vehicle terminal in the grid is segmented based on the nodes, and the vehicle terminals in each segment are scheduled and optimized in turn, thereby achieving more refined path optimization.
- Figure 4 is a global autonomous driving control method based on centralized computing provided in an embodiment of the present application, which is applied to a global autonomous driving control system based on centralized computing. The method includes:
- the vehicle terminal sends travel requirements to the root server.
- travel demand includes the current location and destination information of the vehicle terminal.
- the travel demand may further include a parking area corresponding to the destination information.
- the autonomous driving vehicle login steps are as follows:
- the vehicle terminal When the autonomous driving vehicle is started in the parking space, the vehicle terminal obtains the passenger identity information and sends it to the root server.
- the root server authenticates the passenger identity information.
- the autonomous driving vehicle logs into the autonomous driving control system.
- the logging into the autonomous driving control system specifically includes:
- Biometric authentication of the vehicle owner that is, using biometric recognition technology, such as fingerprint, face, iris, etc., to authenticate the vehicle owner.
- the root server performs a fault check on the data uploaded by the vehicle. If there is no problem, login is allowed. If there is a low-risk fault, conditional login is allowed. If there is a serious fault, login is not allowed.
- the vehicle When the vehicle self-checks successfully, it logs into the autonomous driving network, obtains driving authorization and enters remote takeover mode to control the vehicle through a full-domain autonomous driving control method based on centralized computing.
- the vehicle When the vehicle is controlled by the global autonomous driving control method based on centralized computing, if the vehicle is not started and the attempt to reconnect with the base station is invalid, the vehicle exits the autonomous driving network; if the vehicle is started and the attempt to reconnect with the base station is invalid, the autonomous driving is downgraded to limited autonomous driving, and the vehicle's external sensors perceive the environment, calculate the driving path locally, and automatically enter the temporary lane to wait for rescue.
- the root server receives the travel demand sent by the vehicle terminal, and plans a global path according to the travel demand to obtain multiple driving paths.
- the global path is planned according to the travel demand; when the travel demand does not include the parking area corresponding to the destination information, the parking area corresponding to the destination information is determined according to the predicted travel time required to travel to the parking area and the traveler's historical preference for parking areas, and the global path is planned according to the destination information, the parking area corresponding to the destination information, and the current location.
- the root server after receiving the travel demand, the root server will send an inquiry message to the vehicle terminal, and the inquiry information includes whether there is a need for intermediate parking. If the intermediate parking position corresponding to the intermediate parking demand is obtained, global path planning is performed based on the current location, the intermediate parking position and the destination information.
- the time and energy consumption of multiple driving paths are calculated based on the current traffic conditions, vehicle terminal performance parameters and the number of passengers, and sent to the vehicle terminal so that the passengers can select a path from the multiple driving paths as the global path.
- the vehicle terminal receives a plurality of driving routes and displays the plurality of driving routes, so that the user can select a driving route from the plurality of driving routes to obtain a global route.
- the vehicle terminal sends the global path to the root server.
- the root server divides the global path into multiple subtasks according to the grid covered by the global path.
- Each subtask corresponds to a grid.
- the global path is the driving path of the vehicle terminal.
- the global path is divided into multiple subtasks to obtain a subtask corresponding to each grid.
- the subtask corresponding to each grid generally refers to the task of the vehicle terminal driving within a grid.
- the root server divides the global path into multiple subtasks corresponding to the grids according to the grids covered by the global path, so that one subtask corresponds to one grid. Subsequently, the root server sends the multiple subtasks to the computing centers of each grid respectively.
- the root server sends the multiple subtasks to the middle layer of the computing center of the corresponding grid respectively.
- Each of the multiple subtasks corresponds to a grid, and a computing center middle layer is set in each grid.
- the computing center middle layer After receiving the subtask, the computing center middle layer sends the subtask to at least one computing center connected to the computing center middle layer.
- Each grid is assigned with at least one computing center responsible for local path planning within the grid.
- the backup middle layer when the middle layer of the computing center fails, the backup middle layer is enabled; when the base station cannot establish communication with the middle layer of the grid, the root server takes over the entire grid.
- the subtask is sent to at least one computing center connected to the computing center middle layer, which may be an idle computing center among multiple computing centers connected to the computing center middle layer, or to all computing centers connected to the computing center middle layer.
- the middle layer when the response of a computing center connected to the computing center middle layer times out, the middle layer will allocate resources from other computing centers within the grid; if there are no available computing resources within the grid, computing centers in other grids will provide resources; if all computing centers within the root server coverage area have no available resources, the root server will take over.
- Steps 401-408 complete the assignment of subtasks to the corresponding computing centers, so that the root server assigns the vehicle control rights.
- the computing center of the grid where the vehicle terminal is currently located will bring the vehicle driving data into the entire traffic flow model, perform local path planning for the vehicle terminal in the current grid, and calculate the actuator input parameters to achieve the vehicle's target driving state in real time, control the vehicle terminal to complete the subtask in the current grid; and allocate vehicle control to the computing center that executes the next subtask.
- the computing center determines all current vehicle terminals within the grid where the computing center is located.
- the computing center determines all current vehicle terminals within the grid in order to perform local path planning for all vehicle terminals.
- the computing center obtains the subtask, current location, and current driving status of the vehicle terminal in the grid where the computing center is located.
- the subtask of the computing center in the grid where the computing center is located is used to indicate the position of the vehicle corresponding to the grid entering the grid and the position of the vehicle exiting the grid.
- the subtasks corresponding to the vehicle terminal are segmented to obtain the nodes corresponding to the subtasks.
- the global paths of these vehicles are segmented with the intersection positions and the positions where the vehicles enter and exit the grid as nodes.
- the nodes obtained are generally the nodes corresponding to the intersection positions and the positions where the vehicles enter and exit the grid.
- the positions where the vehicles enter and exit the grid are used to divide the global path into local paths corresponding to the grids, that is, the driving path of the vehicle within a grid; the intersection position is used to divide the path in the same grid.
- the nodes in the same grid the nodes will be arranged according to the order of the vehicle's travel in order to determine the next node at the current location.
- a local path is planned for the vehicle terminal, and the vehicle terminal is controlled to drive along the local path to drive out of the position of the current grid, thereby completing the assigned subtask corresponding to the current grid.
- planning a local path for a vehicle terminal includes: determining a current calculation cycle, the current calculation cycle being a calculation cycle corresponding to the current location of the vehicle terminal to the next node; calculating the theoretical arrival time of each vehicle terminal at the next node according to the current location and current driving status of each vehicle terminal in the current calculation cycle, arranging the order of each vehicle terminal according to the theoretical arrival time, updating the theoretical arrival time of each vehicle terminal, and synchronizing it to the root server; for each vehicle terminal, obtaining the position of the vehicle terminal in real time, calculating the actuator input parameters of the vehicle terminal according to the theoretical arrival time, controlling the vehicle terminal to arrive at the next node within the theoretical time according to the actuator input parameters of the vehicle terminal, and updating the current location of the vehicle terminal; determining whether the next node is the last node corresponding to the subtask, if not, updating the next node of the current location of the vehicle terminal, and executing the step of determining the current calculation cycle; if so, the computing center completes the subtask
- arranging the order of each vehicle terminal according to the theoretical arrival time includes: grouping the vehicle terminals according to the driving direction (straight ahead, left turn or right turn) of each vehicle terminal at the next node, and arranging the order of each group of vehicle terminals according to the theoretical arrival time;
- the computing center determines the actuator input parameters of the vehicle terminal and sends the actuator input parameters to the vehicle terminal.
- the vehicle terminal controls the actuator to operate with the actuator input parameters.
- the vehicle terminal actuator inputs the parameters, it feeds back the offset from the target state to the computing center.
- the computing center brings the vehicle data into the model and recalculates based on the offset, and continuously corrects the actuator parameters until the vehicle terminal is controlled to arrive at the next node according to the theoretical arrival time.
- the vehicle terminal combines GNSS (Global Navigation Satellite System), IMU (inertia measurement unit) track calculation, wheel encoder, etc. to obtain its own real-time position and feed it back to the computing center.
- GNSS Global Navigation Satellite System
- IMU inertia measurement unit
- the step of arranging the order of each vehicle terminal is re-executed.
- the computing center transfers the vehicle control to the computing center of the grid where the next subtask is located until all subtasks corresponding to the global path are completed.
- the process by which the computing center allocates the vehicle control right to the computing center of the grid where the next subtask is located is as follows:
- the current computing center and the next computing center both plan a local path for the vehicle terminal and send actuator input parameters to the vehicle terminal.
- the two adjacent base stations receive the data packets transmitted by the vehicle, send them to the middle layer of their respective grid computing centers, and then send them to the computing center; both computing centers calculate the control parameters based on the real-time information of the vehicle flow in the next grid and the vehicle terminal, and send the control parameters and the request to take over instructions to the vehicle terminal.
- the vehicle terminal receives the calculation results and takeover instructions sent by the two computing centers.
- the vehicle terminal determines the position of the vehicle terminal in the transition area. When the position indicates that it begins to enter the transition area, it still keeps executing the calculation results and takeover instructions sent by the computing center of the current grid. When the position indicates that the vehicle terminal has traveled a set distance in the transition area, it gives priority to executing the input parameters of the actuator that arrives first.
- the vehicle terminal monitors the message delay sent by the two computing centers.
- the packet loss rate and network delay of the message from the computing center of the next grid are less than or equal to the message from the computing center of the current grid, and remain stable for a period of time, feedback information is sent to the current computing center, and the current computing center no longer controls the vehicle terminal, and the handover is completed.
- the packet loss rate and network delay of the computing center remain stable over a period of time, which means that the packet loss rate and network delay are controlled within a certain range respectively to ensure normal communication between the computing center and the vehicle terminal.
- the vehicle After passing the transition area, the vehicle is taken over by a computing center in the new grid, which executes the subtasks assigned by the root server and repeats the above steps 409-413. Finally, the vehicle reaches the destination, enters the parking space, and the autonomous driving process ends.
- the avoidance strategy calculated locally is executed first, including but not limited to: automatic emergency braking, automatic emergency steering, emergency lane keeping, automatic lane change, driving into the emergency lane to wait for rescue, and the results of the vehicle's autonomous avoidance measures are uploaded to the computing center; the computing center replans the nodes and arrival times, updates the order, so that after the danger is lifted, the vehicle resumes remote takeover of the autonomous driving network.
- a distributed computing power layout is realized, which ensures the real-time distribution of control information and vehicle information feedback, and provides a guarantee for the comprehensive popularization of autonomous driving vehicles.
- local path optimization is performed by computing centers in different grids, which is conducive to achieving the best scheduling of computing power and improving the efficiency of information transmission.
- the path of the vehicle terminal in the grid is segmented based on the nodes, and the vehicle terminals in each segment are scheduled and optimized in turn, achieving more refined path optimization.
- the vehicle's driving status is monitored in real time, and real-time corrections are made based on the deviation from the theoretical driving status to ensure that it travels according to the planned path and the theoretical time to reach each node.
- the present application also provides a control handover strategy, which enables the computing centers in two adjacent grids to perform analysis in the transition section, send control parameters to the vehicle, and drive based on the control parameters that arrive first.
- the control is completely handed over to prevent problems such as signal interruption or delay when the vehicle terminal crosses the grid boundary.
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Atmospheric Sciences (AREA)
- Traffic Control Systems (AREA)
Abstract
本申请提供了基于中心化计算的全域自动驾驶控制系统和方法，属于自动驾驶控制技术领域。该系统覆盖的区域被划分为多个网格，该系统包括根服务器和多个计算中心；每个网格中包括至少一个计算中心；其中，根服务器，被配置为接收车辆终端发送的出行需求，根据出行需求规划全局路径；根据全局路径覆盖的网格，将全局路径划分为多个子任务，分别发送至各个网格的计算中心；计算中心，被配置为当车辆终端在计算中心所在网格内行驶的过程中，规划局部路径，并控制车辆终端按照局部路径行驶，完成分配的子任务。本申请保证了控制信息下发和车辆信息反馈的实时性，为自动驾驶车辆的全面普及提供了保障。
Description
本申请要求于2022年12月13日提交的申请号为202211594117.8、发明名称为“一种基于中心化计算的全域自动驾驶控制系统”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
本申请的实施例涉及一种基于中心化计算的全域自动驾驶控制系统和方法。
随着技术的不断进步，现今大多数自动驾驶技术都基于“感知—决策—执行”的三步法则来对车辆规划路径，实现了从单一的由人驾驶的车辆逐渐转换为网联自动驾驶车辆。
发明内容
本申请实施例提供了一种基于中心化计算的全域自动驾驶控制系统和方法。本申请的一个或多个实施例提供了如下技术方案：
一种基于中心化计算的全域自动驾驶控制系统，所述系统覆盖的区域被划分为多个网格，所述系统包括根服务器和多个计算中心；每个所述网格中包括至少一个计算中心；其中，
根服务器，被配置为接收车辆终端发送的出行需求，根据所述出行需求规划全局路径；根据所述全局路径覆盖的网格，将所述全局路径划分为多个子任务，分别发送至各个网格的计算中心；
计算中心，被配置为当车辆终端在所述计算中心所在网格内行驶的过程中，规划局部路径，并控制所述车辆终端按照所述局部路径行驶，完成分配的子任务。
可选地，所述系统还包括多个计算中心中间层，与所述多个网格一一对应；每个所述计算中心中间层与所在网格内的一个或多个计算中心连接，并且，每个所述计算中心中间层均连接至所述根服务器。
可选地，所述系统还包括多个无线通信基站，每个网格中包括至少一个所述无线通信基站，每个所述无线通信基站与所在网格内的计算中心中间层连接。
可选地，所述出行需求包括目的地信息；所述根服务器根据所述出行需求规划全局路径包括：
根据所述车辆终端当前所在位置和目的地信息规划全局路径，得到多种行驶路径；
根据当前交通情况、所述车辆终端性能参数和乘员人数，计算所述多种行驶路径的耗时和耗能，并发送至所述车辆终端；
接收乘员对于行驶路径的选择，用于子任务的划分。
可选地，所述计算中心规划局部路径包括：
获取所在网格内当前所有车辆终端的全局路径、当前位置和当前行驶状态；以路口位置、车辆驶入和驶出所述网格的位置为节点，对这些车辆的全局路径进行分段；
以目标车辆终端当前所在位置到下一个节点为一个计算周期，执行以下步骤：
(1)根据当前计算周期内各车辆终端的当前位置和当前行驶状态，计算这些车辆终端到达下一个节点的理论到达时间，根据理论到达时间对这些车辆终端排列先后顺序，并更新这些车辆终端的理论到达时间；
(2)实时获取目标车辆终端位置，根据理论到达时间计算车辆终端的执行器输入参数，控制所述目标车辆终端在理论时间内到达下一个节点，更新目标车辆终端当前所在位置；
重复执行步骤(1)-(2)，直至到达最后一个节点。
可选地，计算中心完成分配的子任务后，将车辆控制权移交至下一个子任务所在网格的计算中心，移交过程为：
在车辆终端向下一个网格行驶的一定过渡区域内，当前计算中心和下一个计算中心均对所述车辆终端规划局部路径，并向所述车辆终端发送执行器输入参数；
所述车辆终端，在刚进入过渡区域时，仍保持执行当前网格的计算中心下达的执行器输入参数，行驶设定路程后，优先执行先到达的执行器输入参数；
所述车辆终端对两个计算中心发送的报文延时进行监测，当来自下一个计算中心的报文丢包率和网络时延均小于或等于当前计算中心报文，且在一段时间内保持稳定时，向当前计算中心发送反馈信息，当前计算中心不再对所述车辆终端进行控制。
本申请一个或多个实施例提供了一种根服务器，用于全域自动驾驶控制，控制区域被划分为多个网格，每个所述网格中包括至少一个计算中心；所述根服务器与每个计算中心均通过计算中心中间层保持通信连接，所述根服务器被配置为执行以下步骤：
接收车辆终端发送的出行需求，根据所述出行需求规划全局路径；
根据所述全局路径覆盖的网格，将所述全局路径划分为多个子任务，分别发送至各个网格的计算中心。
可选地，所述出行需求包括目的地信息；所述根服务器根据所述出行需求规划全局路径包括：
根据所述车辆终端当前所在位置和目的地信息规划全局路径，得到多种行驶路径；
根据当前交通情况、所述车辆终端性能参数和乘员人数，计算所述多种行驶路径的耗时和耗能，并发送至所述车辆终端；
接收乘员对于行驶路径的选择，用于子任务的划分。
可选地，所述根服务器还获取计算中心执行子任务时发送的数据请求，将所述计算中心所在网格内当前所有车辆终端的全局路径、当前位置和当前行驶状态，用于对当前子任务相应车辆终端执行局部路径规划。
一个或多个实施例提供了一种计算中心，与所述根服务器通信连接，所述计算中心被配置为执行以下步骤：
当车辆终端在所述计算中心所在网格内行驶的过程中，规划局部路径，并控制所述车辆终端按照所述局部路径行驶，完成分配的子任务。
可选地，规划局部路径包括：
获取所在网格内当前所有车辆终端的全局路径、当前位置和当前行驶状态；以路口位置、车辆驶入和驶出所述网格的位置为节点，对这些车辆的全局路径进行分段；
以目标车辆终端当前所在位置到下一个节点为一个计算周期，执行以下步骤：
(1)根据当前计算周期内各车辆终端的当前位置和当前行驶状态，计算这些车辆终端到达下一个节点的理论到达时间，根据理论到达时间对这些车辆终端排列先后顺序，并更新这些车辆终端的理论到达时间；
(2)实时获取目标车辆终端位置，根据理论到达时间计算车辆终端的执行器输入参数，控制所述目标车辆终端在理论时间内到达下一个节点，更新目标车辆终端当前所在位置；
重复执行步骤(1)-(2)，直至到达最后一个节点。
可选地，所述计算中心完成分配的子任务后，将车辆控制权移交至下一个子任务所在网格的计算中心，移交过程为：
在车辆终端向下一个网格行驶的一定过渡区域内，当前计算中心和下一个计算中心均对所述车辆终端规划局部路径，并向所述车辆终端发送执行器输入参数；
所述车辆终端，在刚进入过渡区域时，仍保持执行当前网格的计算中心下达的执行器输入参数，行驶设定路程后，优先执行先到达的执行器输入参数；
所述车辆终端对两个计算中心发送的报文延时进行监测，当来自下一个计算中心的报文丢包率和网络时延均小于或等于当前计算中心报文，且在一段时间内保持稳定时，向当前计算中心发送反馈信息，当前计算中心不再对所述车辆终端进行控制。
本申请一个或多个实施例提供了一种基于中心化计算的全域自动驾驶控制方法，应用于基于中心化计算的全域自动驾驶控制系统，所述系统包括根服务器和多个计算中心；所述系统覆盖的区域被划分为多个网格，每个所述网格中包括至少一个计算中心；其中，
所述根服务器接收车辆终端发送的出行需求，根据所述出行需求规划全局路径；根据所述全局路径覆盖的网格，将所述全局路径划分为多个子任务，并将多个子任务分别发送至各个网格的计算中心；
当车辆终端在所述计算中心所在网格内行驶的过程中，所述计算中心规划局部路径，并控制所述车辆终端按照所述局部路径行驶，完成分配的子任务。
可选地，所述将多个子任务分别发送至各个网格的计算中心，包括：
所述根服务器将所述多个子任务分别发送至对应网格的计算中心中间层，所述多个子任务中每个子任务对应一个网格，每个网格内设置有一个计算中心中间层；
所述计算中心中间层接收到子任务后，将所述子任务发送至与所述计算中心中间层连接的至少一个计算中心。
可选地，当车辆终端在所述计算中心所在网格内行驶的过程中，所述计算中心规划局部路径，并控制所述车辆终端按照所述局部路径行驶，完成分配的子任务，包括：
所述计算中心确定所述计算中心所在网格内的当前的所有车辆终端；
对于当前所有车辆终端中的每一车辆终端，所述计算中心获取车辆终端在所述计算中心所在网格内的子任务、当前位置和当前行驶状态；根据预设的路口位置、车辆驶入和驶出所述网格的位置为节点，对车辆终端对应的子任务进行分段，得到子任务对应的节点；根据子任务对应的节点、当前位置和当前行驶状态，对所述车辆终端规划局部路径，并控制所述车辆终端按照所述局部路径行驶以驶出当前网格的位置，完成分配的当前网格对应的子任务。
可选地，所述根据子任务对应的节点、根据子任务对应的节点、当前位置和当前行驶状态，对所述车辆终端规划局部路径，包括：
确定当前计算周期，所述当前计算周期为所述车辆终端当前所在位置到下一个节点对应的一个计算周期；
根据当前计算周期内各个车辆终端的当前位置和当前行驶状态，计算各个车辆终端到达下一个节点的理论到达时间，根据理论到达时间对各个车辆终端排列先后顺序，并更新各车辆终端的理论到达时间；
对于各个车辆终端，实时获取车辆终端位置，根据所述理论到达时间计算车辆终端的执行器输入参数，根据所述执行器输入参数控制所述车辆终端在理论时间内到达下一个节点，并更新所述车辆终端当前所在位置；
判断所述下一个节点是否为所述子任务对应的最后一个节点，若不是，则更新所述车辆终端当前所在位置的下一个节点，执行确定当前计算周期的步骤；若是，则所述计算中心完成对所述车辆终端对应的子任务。
可选地，所述方法还包括：
当所述车辆终端完成当前网格内的子任务，所述计算中心将车辆控制权移交至下一个子任务所在网格的计算中心，直至完成所述全局路径对应的所有子任务。
可选地，所述计算中心将车辆控制权移交至下一个子任务所在网格的计算中心，包括：
在所述车辆终端向下一个网格行驶的一定过渡区域内，当前计算中心和下一个计算中心均对所述车辆终端规划局部路径，并向所述车辆终端发送执行器输入参数；
所述车辆接收当前计算中心和下一个计算中心分别发送的执行器输入参数，所述车辆终端确定所述车辆终端在过渡区域中的位置，根据所述位置从接收到的执行器输入参数中选择
目标执行器参数，以通过所述目标执行器参数对所述目标车辆终端控制。
可选地，所述根据所述位置从接收到的执行器输入参数中选择目标执行器参数，包括：
当所述位置指示所述车辆终端开始进入过渡区域时，所述车辆终端将当前网格的计算中心发送的执行器输入参数作为目标执行器参数；
当所述位置指示所述车辆终端在所述过渡区域行驶设定路程后，所述车辆终端将先到达所述车辆终端的执行器输入参数作为所述目标执行器参数。
可选地，所述方法还包括：
所述车辆终端对两个计算中心发送的报文延时进行监测，当来自下一个计算中心的报文丢包率和网络时延均小于或等于所述当前计算中心的报文丢包率和网络时延，且在一段时间内保持稳定时，向所述当前计算中心发送反馈信息，以使所述当前计算中心不再对所述车辆终端进行控制。
可选为了更清楚地说明本申请实施例中的技术方案，下面将对实施例描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1为本申请实施例提供的一种基于中心化计算的全域自动驾驶控制系统架构图；
图2为本申请实施例提供的根服务器、计算中心和车辆终端的功能框架图；
图3为本申请实施例提供的一种基于中心化计算的全域自动驾驶控制方法的流程图；
图4为本申请实施例提供的一种基于中心化计算的全域自动驾驶控制方法的流程图。
下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
可选在不冲突的情况下，本申请实施例及实施例中的特征可以相互组合。
随着技术的不断进步，现今大多数自动驾驶技术都基于“感知—决策—执行”的三步法则来对车辆规划路径，实现了从单一的由人驾驶的车辆逐渐转换为网联自动驾驶车辆。
随着网联自动驾驶车辆的普及，使得网联自动驾驶车辆的不断增加，而网联自动驾驶车辆是单独控制的，因此需要对网联自动驾驶车辆统一调度指挥。基于此，本申请实施例提供了如下技术方案：
本申请实施例提供了一种基于中心化计算的全域自动驾驶控制系统，该系统用于控制该系统所辖区的封闭道路中的自动驾驶汽车。本申请实施例中的封闭道路的任何节点上不允许有常规道路接入，且在封闭道路的边缘设置物理障碍以避免有任何脱离本系统控制的交通工具驶入封闭道路内，并规定行人在道路标识线划分的指定区域内活动。该系统所辖区的封闭道路为该系统覆盖的区域，该区域会预先被划分为多个网格对区域，后续以每个网格对应的区域作为网格进行说明。
本申请实施例中的自动驾驶汽车是无需人类接管，由自动驾驶系统网络控制的汽车。该汽车通过实时上传车辆状态(车速、横摆角、转向角、加速度、厘米级定位坐标等)，接收计算中心的接管指令来完成由控制器执行驾驶操作。车辆需包括控制器和传感器以采集环境、车辆和行人信息，如超声波探头、摄像头、毫米波雷达等，一方面在路口、狭窄、泊车、信号差等场景下提供周围环境的感知数据优化系统决策；另一方面可作为安全冗余备份，以使车辆实现本地的自动驾驶功能。
下面参照附图1来描述本申请实施例提出的一种基于中心化计算的全域自动驾驶控制系
统。
如图1所示，本申请实施例提供的一种基于中心化计算的全域自动驾驶控制系统包括根服务器、多个计算中心(图1中以9个计算中心为例)。对于系统覆盖的区域划分得到的多个网格中的每一网格，每个网格包括至少一个计算中心(图1中以每个网格内设置有3个计算中心)，所在网格内的所有计算中心均连接至根服务器。
在本申请的另一些实施例中，全域自动驾驶控制系统还包括至少一个计算中心中间层(图1中以3个计算中心中间层为例)。对于系统覆盖的区域划分得到的多个网格中的每一网格，每个网格包括一个计算中心中间层，每个计算中心中间层均连接至根服务器，每个计算中心中间层与该网格内的至少一个计算中心(图1以每个网格内有3个计算中心为例)连接，根服务器和多个计算中心之间经由计算中心中间层进行通信。
其中，根服务器与计算中心中间层、多个计算中心与计算中心中间层通过有线或无线的方式进行通信，根服务器、计算中心和计算中心中间层都是计算机设备，具有处理器、储存器和通信接口。
在本申请的另一些实施例中，全域自动驾驶控制系统还包括多个无线通信基站，每个网格中包括至少一个无线通信基站，每个无线通信基站与所在网格内的计算中心中间层连接。例如，在道路中间(如路灯)部署无线通信基站，该无线通信基站不仅用于将无线通信设备连接到运营商网络，还与该地理位置的网格内计算中心中间层连接，以降低自动驾驶汽车与计算中心的通信延迟。该无线通信基站用于建立车辆终端与根服务器、计算中心之间的通讯。通信基站的部署密度由通讯标准和介质决定，不限于5G、6G或量子通信，通信速度和稳定性直接决定系统的整体性能上限。
其中，根服务器由超大算力的计算机群组成，负责该根服务器所覆盖范围内的区域对应的全域封闭道路的车辆路径规划、道路数据分析、区域流量控制、异常事件处理，从而能够统筹调度交通资源，中转并处理该根服务器所覆盖范围内跨网格计算中心的交通流量信息。此根服务器上还建立了高精地图数据库、车辆数据库和道路基础设施数据库。
计算中心例如为云服务器单元，计算中心运行的自动驾驶系统对所辖区域内的车辆进行建模和运动分析，计算该所辖区域内的车辆的行车轨迹和达到目标状态的各个时间段内车辆底盘执行器的输入参数；另一方面通过计算中心中间层将计算中心的计算结果实时同步到自动驾驶网络根服务器，接收并执行根服务器的交通流量控制策略。计算中心还对基站和路端基础设施的运行状态进行监控和上报。
在本申请实施例中，计算中心是与外界隔离的计算集群，计算中心通过计算中心中间层与其他设备建立通讯，计算中心中间层负责监控计算中心的工作状态、调度网格内各个计算中心的运算资源，同时作为网格化计算中心的外部接口，与根服务器以及其他网格的计算中心进行通信。
图2是本申请实施例提供的根服务器、计算中心和车辆终端的功能框架图，如图2所示，根服务器包括：
基础数据管理模块，用于管理区域内的地图数据和道路基础设施；
车辆管理模块，用于管理区域内的自动驾驶车辆终端信息和对应乘员身份信息，用于驾车时的身份认证；
行驶授权模块，用于接收乘员启动车辆终端时发送的身份认证请求，并进行认证，若认证通过，进入全局路径规划模块；
全局路径规划模块，用于接收车辆终端发送的出行需求，根据出行需求规划全局路径。具体地，出行需求包括目的地信息；根据出行需求规划全局路径包括：
根据车辆终端当前所在位置和目的地信息规划全局路径，得到多种行驶路径；
根据当前交通情况、车辆终端性能参数和乘员人数，计算多种行驶路径的耗时和耗能，并发送至车辆终端；
接收乘员对于行驶路径的选择，用于子任务的划分。
其中乘员选择的行驶路径作为路径规划的全局路径，将该全局路径划分为多个子任务，每个子任务一般指的是车辆在一个网格内行驶的任务。
子任务分配模块，用于根据全局路径覆盖的网格，将全局路径划分为多个子任务，分别发送至各个网格的计算中心中间层，继而由计算中心中间层分配至所在网格内的计算中心。
子任务跟踪模块，用于跟踪所有子任务的实际完成情况，并根据任务完成的结果和偏差评估对全局规划的影响，以请求子任务重新分配。
车辆状态管理模块，用于获取所有车辆终端的当前位置、当前行驶状态、所在网格，以及当前控制车辆的计算中心等信息；以及，接收计算中心执行子任务时发送的数据请求，将计算中心所在网格内当前所有车辆终端的全局路径、当前位置和当前行驶状态，用于对当前子任务相应车辆终端执行局部路径规划。
计算中心，包括：
局部路径规划模块，用于当车辆终端在计算中心所在网格内行驶的过程中，规划局部路径；具体包括：
向根服务器发送数据请求，自根服务器获取所在网格内当前所有车辆终端的全局路径、当前位置和当前行驶状态；以路口位置、车辆驶入和驶出网格的位置为节点，对这些车辆的全局路径进行分段；
其中，以路口位置、车辆驶入和驶出网格的位置为节点，对这些车辆的全局路径进行分段，得到的节点一般是路口位置、车辆驶入和驶出网格的位置对应的节点，而车辆驶入和驶出网格的位置用于将全局路径划分为网格对应的局部路径，即：车辆在一个网格内的行驶路径；路口位置是用于划分同一网格中的路径。对于同一网格内的节点，会按照车辆的行驶的先后顺序将节点进行排列，以便确定当前所在位置的下一个节点。
以目标车辆终端当前所在位置到下一个节点为一个计算周期，执行以下步骤：
(1)根据当前计算周期内各车辆终端的当前位置和当前行驶状态，计算这些车辆终端到达下一个节点的理论到达时间，根据理论到达时间对这些车辆终端排列先后顺序，并更新这些车辆终端的理论到达时间；
(2)实时获取目标车辆终端位置，根据理论到达时间计算车辆终端的执行器输入参数，根据车辆终端的执行器输入参数控制目标车辆终端在理论时间内到达下一个节点，更新目标车辆终端当前所在位置；即：计算中心会确定车辆终端的执行器输入参数，并将该执行器输入参数发送至车辆终端，该车辆终端控制执行器以执行器输入参数运行，从而控制目标车辆终端在理论时间内到达下一个节点。
重复执行步骤(1)-(2)，直至到达最后一个节点。
车辆行驶控制模块，用于控制车辆终端按照局部路径行驶，完成分配的子任务。
车辆控制权移交模块，用于在该计算中心完成分配的子任务后，将车辆控制权移交至下一个子任务所在网格的计算中心，移交过程为：
在车辆终端向下一个网格行驶的一定过渡区域内，当前计算中心和下一个计算中心均对车辆终端规划局部路径，并向车辆终端发送执行器输入参数；
车辆终端，在刚进入过渡区域时，仍保持执行当前网格的计算中心下达的执行器输入参数，行驶设定路程后，优先执行先到达的执行器输入参数；
车辆终端对两个计算中心发送的报文延时进行监测，当来自下一个网格的计算中心的报文丢包率和网络时延均小于或等于当前网格的计算中心报文，且在一段时间内保持稳定时，向当前计算中心发送反馈信息，当前计算中心不再对车辆终端进行控制。
其中，计算中心的报文丢包率和网络时延在一段时间内保持稳定是指报文丢包率和网络时延分别控制在一定范围内，以保证计算中心与车辆终端之间正常通信。
车辆终端，包括：
当前状态获取模块，用于实时获取自身当前位置和行驶状态，并发送至根服务器；
控制系统登录模块，用于获取乘员的身份信息，并发送至根服务器；
执行控制模块，用于获取计算中心发送的执行器输入参数，控制车辆终端行驶。
此外，系统还设有突发情况的处理策略：
(1)当与基站失去通讯时，
若车辆未启动，尝试重连无效时，退出自动驾驶网络；
若车辆已启动，尝试重连无效时，从自动驾驶降级到有限自动驾驶，由车辆外部传感器感知环境，本地计算行驶路径，自动驶入临时车道等待救援。
(2)当前计算中心响应超时，
由中间层分配本区域内其他计算中心资源；
若网格内无可用计算资源时，由其他网格内计算中心提供资源；
所有计算中心均无可用资源时，由根服务器接管。
(3)当中间层宕机时，
启用备份中间层；
基站与所在网格的中间层无法建立通讯，改由根服务器接管整个所在网格。
(4)当车辆有碰撞危险时，
由车辆外部的传感器感知的危险，必须采取避险措施时，优先执行本地计算的避险策略，包括且不限于：自动紧急制动、自动紧急转向、紧急车道保持、自动变道、驶入应急车道等待救援。
将车辆自主避险措施的结果上传到计算中心；
计算中心重新规划节点目标和到达时间，更新排序；
危险解除后，车辆恢复自动驾驶网络的远程接管。
上述自动驾驶控制系统所执行的控制方法包括以下步骤：
自动驾驶车辆登录步骤：自动驾驶车辆在泊车位被启动时，获取乘员身份信息并发送至根服务器，根服务器对身份信息进行认证，认证通过后，自动驾驶车辆登录自动驾驶控制系统。
其中，登录自动驾驶控制系统具体包括：
(1)车辆解锁上电；
(2)对车主进行生物识别认证；即：采用生物特征识别技术，如指纹、面部、虹膜等，对车主进行身份认证。
(3)车辆自检查，诊断所有控制器状态，记录诊断结果，将检测数据打包上传到根服务器数据库；
(4)根服务器对车辆上传的数据进行故障检查，若无问题允许登录，若存在低风险故障，允许有条件登录，若存在严重故障则不允许登录。
车辆自检成功后，登陆自动驾驶网络，取得行驶授权并进入远程接管模式。
出行请求获取步骤：登录自动驾驶网络后，自动驾驶车辆获取乘员输入的出行请求并发送至根服务器，出行请求包括目的地信息，更为具体地，还包括最终停车区域。全局路径规划步骤：根服务器根据当前所在位置和目的地信息进行全局路径规划，并按照全局路径覆盖的网格，将全局路径分解为多个子任务，将多个子任务分别发送至对应网格的计算中心中间层。
其中，根服务器进行全局路径规划包括：
(1)根服务器收到车辆终端的出行请求；
(2)根据当前所在位置和目的地信息进行全局路径规划，得到多种行驶路径；
(3)根据当前交通流量情况、车辆性能参数、乘员人数计算每种行驶路径的耗时和能耗，发送至车辆终端并进行显示；
(4)接收乘员对于行驶路径的选择，将行驶路径根据其覆盖的网格，划分为多个子任务，每个子任务对应一个网格；
(5)将多个子任务分别发送至对应网格的计算中心中间层；
(6)计算中心中间层接收到子任务后，发送至与其连接的其中一个计算中心。
根服务器获取出行请求后还向车辆终端发送询问信息，询问信息包括是否有中途停车需求，若获取了中途泊车位置，就根据当前所在位置、中途泊车位置和目的地信息进行全局路径规划。
局域行驶控制步骤：根服务器将车辆控制权分配给目标车辆终端当前位置所在网格的计算中心，该计算中心将车辆行驶数据带入整个交通流量模型中，针对目标车辆终端在当前网格内进行局部路径规划，并实时计算达到车辆目标行驶状态的执行器输入参数，控制目标车辆终端完成当前网格内的子任务；将车辆控制权分配给执行下一个子任务的计算中心。
计算中心被配置为执行以下步骤：
(1)获取所在网格内当前所有车辆终端的全局路径、当前位置信息和当前行驶状态信息；
(2)将这些车辆终端的全局路径均进行分段，例如以路口位置、车辆驶入和驶出网格的位置为分段节点；
(3)以目标车辆终端当前所在位置到下一个节点为一个计算周期，根据当前计算周期内各车辆终端的当前位置和当前行驶状态，计算这些车辆终端到达下一个节点的理论到达时间；
(4)根据这些车辆终端在下一个节点的行驶方向(直行、左转或右转)，对这些车辆终端进行分组，对于每组车辆终端，根据理论到达时间排列先后顺序；更新车辆终端的理论到达时间，并同步至根服务器；
(5)实时获取目标车辆终端位置，根据理论到达时间计算车辆终端的执行器输入参数，保证车辆在理论时间内到达下一个节点；行驶过程中，目标车辆终端执行器输入参数，并向计算中心反馈与目标状态的偏移量，计算中心根据偏移量，把车辆数据代入模型中重新计算，不断修正执行器参数，直至到达下一个节点。
其中，车辆终端结合GNSS(全球导航卫星系统，Global Navigation Satellite System)、IMU(惯性传感器，inertia measurement unit)航迹推算、车轮编码器等获得自身实时位置并反馈至计算中心。
(6)重复执行步骤(3)-(5)，直至到达最后一个节点，即车辆驶出当前网格的位置，完成目标车辆终端在当前网格内的子任务。
若存在子任务变更(乘员变更目的地)或新增子任务(有新的车辆在当前网格行驶)，重新执行排序。
(7)计算中心将车辆控制权分配给下一个子任务所在网格的计算中心，重复执行步骤(1)-(6)，直至完成全部子任务。
其中，步骤(7)中计算中心将车辆控制权分配给下一个子任务所在网格的计算中心的过程如下：
(7-1)在车辆终端向下一个网格行驶的一定过渡区域内，当前计算中心和下一个计算中心均对车辆终端规划局部路径，并向车辆终端发送执行器输入参数。具体地，驶入过渡路段，相邻两个基站均接收车辆传送的数据包，发送到各自的网格计算中心中间层，继而被发送至计算中心；两个计算中心均基于下一个网格内车流量和目标车辆终端的实时信息，计算控制参数，将控制参数和请求接管指令发送给车辆终端。
通过基站接收车辆信息，保证了数据传输的实时性。
(7-2)车辆终端接收两个计算中心发送的计算结果和接管指令，在刚进入过渡区域内时，仍保持执行当前网格的计算中心发送的计算结果和接管指令，行驶一定路程后，优先执
行先到达的执行器输入参数。
(7-3)车辆终端对两个计算中心发送的报文延时进行监测，当来自下一个网格的计算中心的报文丢包率和网络时延均小于或等于当前网格的计算中心报文，且在一段时间内保持稳定时，向当前计算中心发送反馈信息，当前计算中心不再对车辆终端进行控制，交接结束。
渡过交接期后，车辆被新的网格中的某个计算中心接管，该计算中心执行根服务器分配的子任务，重复上述过程，最终到达目的地，车辆进入泊车位，自动驾驶过程结束。
若上述出行请求获取步骤中未获取最终停车区域，例如乘员没有固定泊车位的情形，获取出行请求后还向车辆终端发送询问信息，询问信息包括是否有泊车位置偏好。进行全局路径规划后，对于每种行驶路径，还根据预期到达时间和泊车位置偏好计算出多个推荐泊车位。将每种行驶路径的耗时、能耗和推荐泊车位发送至车辆终端。在局域行驶控制步骤执行过程中，通过不断修正预期到达时间，从多个推荐泊车位中逐步锁定目标泊车位，最终控制车辆终端进入目标泊车位。此处需要注意的是，若存在多个车辆终端锁定同一目标泊车位，根服务器对多个车辆终端预估到达时间，并进行排序，优先将车位分配给先到达者。
本申请实施例提供的一种基于中心化计算的全域自动驾驶控制方法，应用于基于中心化计算的全域自动驾驶控制系统，基于中心化计算的全域自动驾驶控制系统中的根服务器、计算中心中间层和计算中心均能够执行本申请实施例提供的基于中心化计算的全域自动驾驶控制方法中的一个或多个步骤。
图3是本申请实施例提供的一种基于中心化计算的全域自动驾驶控制方法，应用于基于中心化计算的全域自动驾驶控制系统，该方法包括：
301、根服务器接收车辆终端发送的出行需求，根据出行需求规划全局路径；根据全局路径覆盖的网格，将全局路径划分为多个子任务。
302、将多个子任务分别发送至各个网格的计算中心。
303、计算中心当车辆终端在计算中心所在网格内行驶的过程中，规划局部路径，并控制车辆终端按照局部路径行驶，完成分配的子任务。
本申请实施例通过对全域自动驾驶道路进行地理网格划分，在网格内设置一个或多个计算中心，以及设置集中式根服务器的方式，实现了分布式的算力布局，保证了控制信息下发和车辆信息反馈的实时性，为自动驾驶车辆的全面普及提供了保障。并且，通过对车辆的全局路径划分为子任务，由不同网格内的计算中心执行局部路径优化，有利于实现算力的最佳调度，并提高信息传递的效率；此外，每个网格内进行局部路径规划时，还对车辆终端在该网格内的路径基于节点进行分段，依次对每个分段内的车辆终端进行调度优化，实现了更为精细化的路径优化。图4是本申请实施例提供的一种基于中心化计算的全域自动驾驶控制方法，应用于基于中心化计算的全域自动驾驶控制系统，该方法包括：
401、车辆终端向根服务器发送出行需求。
其中，出行需求包括车辆终端的当前所在位置和目的地信息。
在本申请的实施例中，出行需求还可以包括目的地信息对应的停车区域。
在车辆终端向根服务器发送出行需求之前需要执行自动驾驶车辆登录步骤，自动驾驶车辆登录步骤如下：
当自动驾驶车辆在泊车位被启动时，车辆终端获取乘员身份信息并发送至根服务器，根服务器对乘员身份信息进行认证，在认证通过后，自动驾驶车辆登录自动驾驶控制系统。其中，登录自动驾驶控制系统具体包括：
(1)车辆解锁上电；
(2)对车主进行生物识别认证；即：采用生物特征识别技术，如指纹、面部、虹膜等，对车主进行身份认证。
(3)车辆自检查，诊断车辆的所有控制器状态，记录诊断结果，将检测数据打包上传到根服务器数据库；
(4)根服务器对车辆上传的数据进行故障检查，若无问题允许登录，若存在低风险故障，允许有条件登录，若存在严重故障则不允许登录。
当车辆自检成功后，登陆自动驾驶网络，取得行驶授权并进入远程接管模式，以通过基于中心化计算的全域自动驾驶控制方法对车辆进行控制。
当基于中心化计算的全域自动驾驶控制方法对车辆进行控制时，若车辆未启动，尝试与基站重连无效时，退出自动驾驶网络；若车辆已启动，尝试与基站重连无效时，从自动驾驶降级到有限自动驾驶，由车辆外部传感器感知环境，本地计算行驶路径，自动驶入临时车道等待救援。
402、根服务器接收车辆终端发送的出行需求，并根据出行需求规划全局路径，得到多种行驶路径。
当出行需求包括目的地信息对应的停车区域时，则根据出行需求规划全局路径；当出行需求不包括目的地信息对应的停车区域时，根据行驶至停车区域所需要的预测出行时间和出行人对停车区域的历史偏好，确定目的地信息对应的停车区域，根据目的地信息、目的地信息对应的停车区域、当前所在位置规划全局路径。
在本申请的实施例中，根服务器在接收到出行需求后，会向车辆终端发送询问信息，询问信息包括是否有中途停车需求，若获取了中途停车需求对应的中途泊车位置，则根据当前所在位置、中途泊车位置和目的地信息进行全局路径规划。
在本申请的实施例中，根据当前交通情况、车辆终端性能参数和乘员人数，计算多种行驶路径的耗时和耗能，并发送至车辆终端，以便乘员从多种行驶路径中选择一个路径作为全局路径。
403、将多种行驶路径发送至车辆终端。
404、车辆终端接收多种行驶路径并显示多种行驶路径，以使用户从多种行驶路径中选择行驶路径，得到全局路径。
405、车辆终端将全局路径发送至根服务器。
406、根服务器根据全局路径覆盖的网格，将全局路径划分为多个子任务。
其中，每个子任务对应一个网格。该全局路径为该车辆终端的行驶路径，将全局路径划分为多个子任务，得到每个网格对应的子任务，每个网格对应的子任务一般是指车辆终端在一个网格内行驶的任务。
在本申请的实施例中，根服务器根据全局路径覆盖的网格，将全局路径划分为网格对应的多个子任务，以使一个子任务对应一个网格，后续根服务器将多个子任务分别发送至各个网格的计算中心。
407、根服务器将多个子任务分别发送至对应网格的计算中心中间层。
其中，多个子任务中每个子任务对应一个网格，每个网格内设置有一个计算中心中间层。
408、计算中心中间层接收到子任务后，将子任务发送至与计算中心中间层连接的至少一个计算中心。
其中，每个网格分配有至少一个计算中心来负责该网格内的局部路径规划。
在本申请的实施例中，当计算中心中间层宕机时，启用备份中间层；当基站与所在网格的中间层无法建立通讯，由根服务器接管整个所在网格。
在本申请的实施例中，将子任务发送至与计算中心中间层连接的至少一个计算中心可以为与计算中心中间层连接的多个计算中心中的空闲的计算中心，也可以为与计算中心中间层连接的所有计算中心。
在本申请的实施例中，在与计算中心中间层连接的一个计算中心的响应超时，由中间层分配本网格内其他计算中心资源；若网格内无可用计算资源时，由其他网格内计算中心提供资源；若根服务器覆盖区域内的所有计算中心均无可用资源时，由根服务器接管。
通过步骤401-408完成子任务分配至对应的计算中心，从而根服务器将车辆控制权分配
给车辆终端当前位置所在网格的计算中心，计算中心将车辆行驶数据带入整个交通流量模型中，针对车辆终端在当前网格内进行局部路径规划，并实时计算达到车辆目标行驶状态的执行器输入参数，控制车辆终端完成当前网格内的子任务；将车辆控制权分配给执行下一个子任务的计算中心。
409、计算中心确定计算中心所在网格内的当前的所有车辆终端。
在本申请实施例中，计算中心会确定该网格内的当前的所有车辆终端，以便对所有车辆终端进行局部路径规划。
410、对于当前所有车辆终端中的每一车辆终端，计算中心获取车辆终端在计算中心所在网格内的子任务、当前位置和当前行驶状态。
需要说明的是，该计算中心在计算中心所在网格内的子任务用于表示该网格对应的车辆驶入该网格的位置和车辆驶出该网格的位置。
411、根据预设的路口位置、车辆驶入和驶出网格的位置为节点，对车辆终端对应的子任务进行分段，得到子任务对应的节点。
其中，以路口位置、车辆驶入和驶出网格的位置为节点，对这些车辆的全局路径进行分段，得到的节点一般是路口位置、车辆驶入和驶出网格的位置对应的节点，而车辆驶入和驶出网格的位置用于将全局路径划分为网格对应的局部路径，即：车辆在一个网格内的行驶路径；路口位置是用于划分同一网格中的路径。对于同一网格内的节点，会按照车辆的行驶的先后顺序将节点进行排列，以便确定当前所在位置的下一个节点。
412、根据子任务对应的节点、当前位置和当前行驶状态，对车辆终端规划局部路径，并控制车辆终端按照局部路径行驶以驶出当前网格的位置，完成分配的当前网格对应的子任务。
在本申请的实施例中，对车辆终端规划局部路径包括：确定当前计算周期，当前计算周期为车辆终端当前所在位置到下一个节点对应的一个计算周期；根据当前计算周期内各个车辆终端的当前位置和当前行驶状态，计算各个车辆终端到达下一个节点的理论到达时间，根据理论到达时间对各个车辆终端排列先后顺序，并更新各车辆终端的理论到达时间，并同步至根服务器；对于各个车辆终端，实时获取车辆终端位置，根据理论到达时间计算车辆终端的执行器输入参数，根据车辆终端的执行器输入参数控制车辆终端在理论时间内到达下一个节点，并更新车辆终端当前所在位置；判断下一个节点是否为子任务对应的最后一个节点，若不是，则更新车辆终端当前所在位置的下一个节点，执行确定当前计算周期的步骤；若是，则计算中心完成对车辆终端对应的子任务。
在本申请的实施例中，根据理论到达时间对各个车辆终端排列先后顺序包括：根据各车辆终端在下一个节点的行驶方向(直行、左转或右转)，对这些车辆终端进行分组，对于每组车辆终端，根据理论到达时间排列先后顺序；
在本申请的实施例中，计算中心会确定车辆终端的执行器输入参数，并将该执行器输入参数发送至车辆终端，该车辆终端控制执行器以执行器输入参数运行，当车辆终端执行器输入参数时，并向计算中心反馈与目标状态的偏移量，计算中心根据偏移量，把车辆数据带入模型中重新计算，不断修正执行器参数，直至控制车辆终端按理论到达时间到达下一个节点。
其中，车辆终端结合GNSS(全球导航卫星系统，Global Navigation Satellite System)、IMU(惯性传感器，inertia measurement unit)航迹推算、车轮编码器等获得自身实时位置并反馈至计算中心。
在本申请的实施例中，若存在子任务变更(例如乘员变更目的地)或新增子任务(例如有新的车辆在当前网格行驶)，重新执行对各个车辆终端排列先后顺序的步骤。
413、当车辆终端完成当前网格内的子任务，计算中心将车辆控制权移交至下一个子任务所在网格的计算中心，直至完成全局路径对应的所有子任务。
其中，计算中心将车辆控制权分配给下一个子任务所在网格的计算中心的过程如下：在
车辆终端向下一个网格行驶的一定过渡区域内，当前计算中心和下一个计算中心均对车辆终端规划局部路径，并向车辆终端发送执行器输入参数。具体地，驶入过渡区域，相邻两个基站均接收车辆传送的数据包，发送到各自的网格计算中心中间层，继而被发送至计算中心；两个计算中心均基于下一个网格内车流量和车辆终端的实时信息，计算控制参数，将控制参数和请求接管指令发送给车辆终端。
车辆终端接收两个计算中心发送的计算结果和接管指令，车辆终端确定车辆终端在过渡区域中的位置，当位置指示开始进入过渡区域内时，仍保持执行当前网格的计算中心发送的计算结果和接管指令，当位置指示车辆终端在过渡区域行驶设定路程后，优先执行先到达的执行器输入参数。
在本申请的实施例中，车辆终端对两个计算中心发送的报文延时进行监测，当来自下一个网格的计算中心的报文丢包率和网络时延均小于或等于当前网格的计算中心报文，且在一段时间内保持稳定时，向当前计算中心发送反馈信息，当前计算中心不再对车辆终端进行控制，交接结束。
其中，计算中心的报文丢包率和网络时延在一段时间内保持稳定是指报文丢包率和网络时延分别控制在一定范围内，以保证计算中心与车辆终端之间正常通信。
在渡过过渡区域后，车辆被新的网格中的某个计算中心接管，该计算中心执行根服务器分配的子任务，重复上述步骤409-413，最终到达目的地，车辆进入泊车位，自动驾驶过程结束。
在本申请实施例中，当车辆终端由车辆外部的传感器感知有碰撞的危险，必须采取避险措施时，优先执行本地计算的避险策略，包括且不限于：自动紧急制动、自动紧急转向、紧急车道保持、自动变道、驶入应急车道等待救援，并将车辆自主避险措施的结果上传到计算中心；计算中心重新规划节点和到达时间，更新排序，从而在危险解除后，车辆恢复自动驾驶网络的远程接管。通过对全域自动驾驶道路进行地理网格划分，在网格内设置一个或多个计算中心，以及设置集中式根服务器的方式，实现了分布式的算力布局，保证了控制信息下发和车辆信息反馈的实时性，为自动驾驶车辆的全面普及提供了保障。
通过对车辆的全局路径划分为子任务，由不同网格内的计算中心执行局部路径优化，有利于实现算力的最佳调度，并提高信息传递的效率；此外，每个网格内进行局部路径规划时，还对车辆终端在该网格内的路径基于节点进行分段，依次对每个分段内的车辆终端进行调度优化，实现了更为精细化的路径优化。
同时，在得到了局部优化的路径后，还实时监测车辆的行驶状态，根据与理论行驶状态的偏移进行实时修正，保证其按照规划的路径和到达每个节点的理论时间行驶。
为了克服相邻网格内计算中心车辆控制器移交的问题，本申请还提供了控制权移交策略，在过渡段上使得相邻两个网格内的计算中心均执行分析，均将控制参数发送至车辆，基于优先到达的控制参数进行行驶，待某一计算中心发送的数据稳定时完全移交控制权，防止车辆终端在越过网格交界线时发生信号中断或延迟等问题。
上述虽然结合附图对本申请的具体实施方式进行了描述，但并非对本申请保护范围的限制，所属领域技术人员应该明白，在本申请的技术方案的基础上，本领域技术人员不需要付出创造性劳动即可做出的各种修改或变形仍在本申请的保护范围以内。
在本申请中，应该理解到，术语“第一”、“第二”等仅用于描述目的，而不能理解为指示或暗示相对重要性或者隐含指明所指示的技术特征的数量。
本领域技术人员在考虑说明书及实践这里公开的发明后，将容易想到本申请的其它实施方案。本申请旨在涵盖本申请的任何变型、用途或者适应性变化，这些变型、用途或者适应性变化遵循本申请的一般性原理并包括本申请未公开的本技术领域中的公知常识或惯用技术手段。说明书和实施例仅被视为示例性的。
应当理解的是，本申请并不局限于上面已经描述并在附图中示出的精确结构，并且可以在不脱离其范围进行各种修改和改变。本申请的范围仅由所附的权利要求来限制。
以上所述仅是为了便于本领域的技术人员理解本申请的技术方案，并不用以限制本申请。凡在本申请的精神和原则之内，所作的任何修改、等同替换、改进等，均应包含在本申请的保护范围。
Claims (20)
- 一种基于中心化计算的全域自动驾驶控制系统，其特征在于，所述系统覆盖的区域被划分为多个网格，所述系统包括根服务器和多个计算中心；每个所述网格中包括至少一个计算中心；其中，根服务器，被配置为接收车辆终端发送的出行需求，根据所述出行需求规划全局路径；根据所述全局路径覆盖的网格，将所述全局路径划分为多个子任务，分别发送至各个网格的计算中心；计算中心，被配置为当车辆终端在所述计算中心所在网格内行驶的过程中，规划局部路径，并控制所述车辆终端按照所述局部路径行驶，完成分配的子任务。
- 如权利要求1所述的基于中心化计算的全域自动驾驶控制系统，其特征在于，所述系统还包括多个计算中心中间层，与所述多个网格一一对应；每个所述计算中心中间层与所在网格内的一个或多个计算中心连接，并且，每个所述计算中心中间层均连接至所述根服务器。
- 如权利要求1或2所述的基于中心化计算的全域自动驾驶控制系统，其特征在于，所述系统还包括多个无线通信基站，每个网格中包括至少一个所述无线通信基站，每个所述无线通信基站与所在网格内的计算中心中间层连接。
- 如权利要求1或2所述的基于中心化计算的全域自动驾驶控制系统，其特征在于，所述出行需求包括目的地信息；所述根服务器根据所述出行需求规划全局路径包括：根据所述车辆终端当前所在位置和目的地信息规划全局路径，得到多种行驶路径；根据当前交通情况、所述车辆终端性能参数和乘员人数，计算所述多种行驶路径的耗时和耗能，并发送至所述车辆终端；接收乘员对于行驶路径的选择，用于子任务的划分。
- 如权利要求4所述的基于中心化计算的全域自动驾驶控制系统，其特征在于，所述计算中心规划局部路径包括：获取所在网格内当前所有车辆终端的全局路径、当前位置和当前行驶状态；以路口位置、车辆驶入和驶出所述网格的位置为节点，对这些车辆的全局路径进行分段；以目标车辆终端当前所在位置到下一个节点为一个计算周期，执行以下步骤：(1)根据当前计算周期内各车辆终端的当前位置和当前行驶状态，计算这些车辆终端到达下一个节点的理论到达时间，根据理论到达时间对这些车辆终端排列先后顺序，并更新这些车辆终端的理论到达时间；(2)实时获取目标车辆终端位置，根据理论到达时间计算车辆终端的执行器输入参数，控制所述目标车辆终端在理论时间内到达下一个节点，更新目标车辆终端当前所在位置；重复执行步骤(1)-(2)，直至到达最后一个节点。
- 如权利要求3所述的基于中心化计算的全域自动驾驶控制系统，其特征在于，计算中心完成分配的子任务后，将车辆控制权移交至下一个子任务所在网格的计算中心，移交过程为：在车辆终端向下一个网格行驶的一定过渡区域内，当前计算中心和下一个计算中心均对所述车辆终端规划局部路径，并向所述车辆终端发送执行器输入参数；所述车辆终端，在刚进入过渡区域时，仍保持执行当前网格的计算中心下达的执行器输入参数，行驶设定路程后，优先执行先到达的执行器输入参数；所述车辆终端对两个计算中心发送的报文延时进行监测，当来自下一个计算中心的报文丢包率和网络时延均小于或等于当前计算中心报文，且在一段时间内保持稳定时，向当前计算中心发送反馈信息，当前计算中心不再对所述车辆终端进行控制。
- 一种根服务器，用于全域自动驾驶控制，其特征在于，控制区域被划分为多个网格， 每个所述网格中包括至少一个计算中心；所述根服务器与每个计算中心均通过计算中心中间层保持通信连接，所述根服务器被配置为执行以下步骤：接收车辆终端发送的出行需求，根据所述出行需求规划全局路径；根据所述全局路径覆盖的网格，将所述全局路径划分为多个子任务，分别发送至各个网格的计算中心。
- 如权利要求7所述的一种根服务器，其特征在于，所述出行需求包括目的地信息；所述根服务器根据所述出行需求规划全局路径包括：根据所述车辆终端当前所在位置和目的地信息规划全局路径，得到多种行驶路径；根据当前交通情况、所述车辆终端性能参数和乘员人数，计算所述多种行驶路径的耗时和耗能，并发送至所述车辆终端；接收乘员对于行驶路径的选择，用于子任务的划分。
- 如权利要求7所述的一种根服务器，其特征在于，所述根服务器还获取计算中心执行子任务时发送的数据请求，将所述计算中心所在网格内当前所有车辆终端的全局路径、当前位置和当前行驶状态，用于对当前子任务相应车辆终端执行局部路径规划。
- 一种计算中心，与如权利要求7-9任一项所述根服务器通信连接，其特征在于，所述计算中心被配置为执行以下步骤：当车辆终端在所述计算中心所在网格内行驶的过程中，规划局部路径，并控制所述车辆终端按照所述局部路径行驶，完成分配的子任务。
- 如权利要求9所述的计算中心，其特征在于，规划局部路径包括：获取所在网格内当前所有车辆终端的全局路径、当前位置和当前行驶状态；以路口位置、车辆驶入和驶出所述网格的位置为节点，对这些车辆的全局路径进行分段；以目标车辆终端当前所在位置到下一个节点为一个计算周期，执行以下步骤：(1)根据当前计算周期内各车辆终端的当前位置和当前行驶状态，计算这些车辆终端到达下一个节点的理论到达时间，根据理论到达时间对这些车辆终端排列先后顺序，并更新这些车辆终端的理论到达时间；(2)实时获取目标车辆终端位置，根据理论到达时间计算车辆终端的执行器输入参数，控制所述目标车辆终端在理论时间内到达下一个节点，更新目标车辆终端当前所在位置；重复执行步骤(1)-(2)，直至到达最后一个节点。
- 如权利要求9所述的计算中心，其特征在于，所述计算中心完成分配的子任务后，将车辆控制权移交至下一个子任务所在网格的计算中心，移交过程为：在车辆终端向下一个网格行驶的一定过渡区域内，当前计算中心和下一个计算中心均对所述车辆终端规划局部路径，并向所述车辆终端发送执行器输入参数；所述车辆终端，在刚进入过渡区域时，仍保持执行当前网格的计算中心下达的执行器输入参数，行驶设定路程后，优先执行先到达的执行器输入参数；所述车辆终端对两个计算中心发送的报文延时进行监测，当来自下一个计算中心的报文丢包率和网络时延均小于或等于当前计算中心报文，且在一段时间内保持稳定时，向当前计算中心发送反馈信息，当前计算中心不再对所述车辆终端进行控制。
- 一种基于中心化计算的全域自动驾驶控制方法，其特征在于，应用于基于中心化计算的全域自动驾驶控制系统，所述系统包括根服务器和多个计算中心，所述系统覆盖的区域被划分为多个网格，每个所述网格中包括至少一个计算中心；其中，所述根服务器接收车辆终端发送的出行需求，根据所述出行需求规划全局路径；根据所述全局路径覆盖的网格，将所述全局路径划分为多个子任务，并将多个子任务分别发送至各个网格的计算中心；当车辆终端在所述计算中心所在网格内行驶的过程中，所述计算中心规划局部路径，并控制所述车辆终端按照所述局部路径行驶，完成分配的子任务。
- 如权利要求13所述的方法，其特征在于，所述将多个子任务分别发送至各个网格的计算中心，包括：所述根服务器将所述多个子任务分别发送至对应网格的计算中心中间层，所述多个子任务中每个子任务对应一个网格，每个网格内设置有一个计算中心中间层；所述计算中心中间层接收到子任务后，将所述子任务发送至与所述计算中心中间层连接的至少一个计算中心。
- 如权利要求13所述的方法，其特征在于，当车辆终端在所述计算中心所在网格内行驶的过程中，所述计算中心规划局部路径，并控制所述车辆终端按照所述局部路径行驶，完成分配的子任务，包括：所述计算中心确定所述计算中心所在网格内的当前的所有车辆终端；对于当前所有车辆终端中的每一车辆终端，所述计算中心获取车辆终端在所述计算中心所在网格内的子任务、当前位置和当前行驶状态；根据预设的路口位置、车辆驶入和驶出所述网格的位置为节点，对车辆终端对应的子任务进行分段，得到子任务对应的节点；根据子任务对应的节点、当前位置和当前行驶状态，对所述车辆终端规划局部路径，并控制所述车辆终端按照所述局部路径行驶以驶出当前网格的位置，完成分配的当前网格对应的子任务。
- 如权利要求15所述的方法，其特征在于，所述根据子任务对应的节点、根据子任务对应的节点、当前位置和当前行驶状态，对所述车辆终端规划局部路径，包括：确定当前计算周期，所述当前计算周期为所述车辆终端当前所在位置到下一个节点对应的一个计算周期；根据当前计算周期内各个车辆终端的当前位置和当前行驶状态，计算各个车辆终端到达下一个节点的理论到达时间，根据理论到达时间对各个车辆终端排列先后顺序，并更新各车辆终端的理论到达时间；对于各个车辆终端，实时获取车辆终端位置，根据所述理论到达时间计算车辆终端的执行器输入参数，根据所述执行器输入参数控制所述车辆终端在理论时间内到达下一个节点，并更新所述车辆终端当前所在位置；判断所述下一个节点是否为所述子任务对应的最后一个节点，若不是，则更新所述车辆终端当前所在位置的下一个节点，执行确定当前计算周期的步骤；若是，则所述计算中心完成对所述车辆终端对应的子任务。
- 如权利要求15所述的方法，其特征在于，所述方法还包括：当所述车辆终端完成当前网格内的子任务，所述计算中心将车辆控制权移交至下一个子任务所在网格的计算中心，直至完成所述全局路径对应的所有子任务。
- 如权利要求17所述的方法，其特征在于，所述计算中心将车辆控制权移交至下一个子任务所在网格的计算中心，包括：在所述车辆终端向下一个网格行驶的一定过渡区域内，当前计算中心和下一个计算中心均对所述车辆终端规划局部路径，并向所述车辆终端发送执行器输入参数；所述车辆接收当前计算中心和下一个计算中心分别发送的执行器输入参数，所述车辆终端确定所述车辆终端在过渡区域中的位置，根据所述位置从接收到的执行器输入参数中选择目标执行器参数，以通过所述目标执行器参数对所述目标车辆终端控制。
- 如权利要求18所述的方法，其特征在于，所述根据所述位置从接收到的执行器输入参数中选择目标执行器参数，包括：当所述位置指示所述车辆终端开始进入过渡区域时，所述车辆终端将当前网格的计算中心发送的执行器输入参数作为目标执行器参数；当所述位置指示所述车辆终端在所述过渡区域行驶设定路程后，所述车辆终端将先到达所述车辆终端的执行器输入参数作为所述目标执行器参数。
- 如权利要求18所述的方法，其特征在于，所述方法还包括：所述车辆终端对两个计算中心发送的报文延时进行监测，当来自下一个计算中心的报文丢包率和网络时延均小于或等于所述当前计算中心的报文丢包率和网络时延，且在一段时间内保持稳定时，向所述当前计算中心发送反馈信息，以使所述当前计算中心不再对所述车辆终端进行控制。
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211594117.8A CN115953911A (zh) | 2022-12-13 | 2022-12-13 | 一种基于中心化计算的全域自动驾驶控制系统 |
| CN202211594117.8 | 2022-12-13 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2024125529A1 true WO2024125529A1 (zh) | 2024-06-20 |
Family
ID=87296081
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2023/138258 WO2024125529A1 (zh) | 2022-12-13 | 2023-12-12 | 基于中心化计算的全域自动驾驶控制系统和方法 |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN115953911A (zh) |
| WO (1) | WO2024125529A1 (zh) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115953911A (zh) * | 2022-12-13 | 2023-04-11 | 奇瑞汽车股份有限公司 | 一种基于中心化计算的全域自动驾驶控制系统 |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108931971A (zh) * | 2018-05-24 | 2018-12-04 | 奇瑞汽车股份有限公司 | 用于无人驾驶的移动终端、车辆、服务器及无人驾驶系统 |
| CN110210806A (zh) * | 2019-05-27 | 2019-09-06 | 大连理工大学 | 一种5g边缘计算的云基无人车架构及其控制评价方法 |
| CN112258864A (zh) * | 2020-10-19 | 2021-01-22 | 广西大学 | 基于顺序选择的自动驾驶车辆路口调度方法及系统 |
| CN113867354A (zh) * | 2021-10-11 | 2021-12-31 | 电子科技大学 | 一种自动驾驶多车智能协同的区域交通流导引方法 |
| CN115953911A (zh) * | 2022-12-13 | 2023-04-11 | 奇瑞汽车股份有限公司 | 一种基于中心化计算的全域自动驾驶控制系统 |
-
2022
- 2022-12-13 CN CN202211594117.8A patent/CN115953911A/zh active Pending
-
2023
- 2023-12-12 WO PCT/CN2023/138258 patent/WO2024125529A1/zh unknown
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108931971A (zh) * | 2018-05-24 | 2018-12-04 | 奇瑞汽车股份有限公司 | 用于无人驾驶的移动终端、车辆、服务器及无人驾驶系统 |
| CN110210806A (zh) * | 2019-05-27 | 2019-09-06 | 大连理工大学 | 一种5g边缘计算的云基无人车架构及其控制评价方法 |
| CN112258864A (zh) * | 2020-10-19 | 2021-01-22 | 广西大学 | 基于顺序选择的自动驾驶车辆路口调度方法及系统 |
| CN113867354A (zh) * | 2021-10-11 | 2021-12-31 | 电子科技大学 | 一种自动驾驶多车智能协同的区域交通流导引方法 |
| CN115953911A (zh) * | 2022-12-13 | 2023-04-11 | 奇瑞汽车股份有限公司 | 一种基于中心化计算的全域自动驾驶控制系统 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115953911A (zh) | 2023-04-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113486293B (zh) | 全自动化边装卸集装箱码头的智能水平运输系统及方法 | |
| WO2018036197A1 (zh) | 一种无人驾驶车辆控制方法、终端、服务器及系统 | |
| JP6732129B2 (ja) | 問題の状況に対処するための自律走行車の遠隔操作 | |
| US10345805B2 (en) | System for and method of maximizing utilization of a closed transport system in an on-demand network | |
| JP7214017B2 (ja) | 異なる交差道路を走行する車両の共同制御 | |
| CN109285373B (zh) | 一种面向整体道路网的智能网联交通系统 | |
| CN108011947B (zh) | 一种车辆协作式编队行驶系统 | |
| US11378952B2 (en) | Autonomous vehicle remote support mapping interface | |
| CN106651175B (zh) | 无人驾驶车辆运营管理系统、总控平台、分控平台、车载计算装置和计算机可读存储介质 | |
| CN101859494B (zh) | 车队车辆管理 | |
| WO2019233153A1 (zh) | 一种基于车车协作的列车移动授权方法 | |
| WO2023109495A1 (zh) | 一种tacs系统的防死锁方法、装置、设备及介质 | |
| JP2019537159A5 (zh) | ||
| CN104269048A (zh) | 智能公交系统的动态调度和时刻管理 | |
| CN104298236A (zh) | 智能公交系统的行程规划和管理方法 | |
| JP2019185279A (ja) | 管制装置 | |
| WO2024125529A1 (zh) | 基于中心化计算的全域自动驾驶控制系统和方法 | |
| JP2021028747A (ja) | 車両貸出システム及び車両貸出方法 | |
| US20240278817A1 (en) | System and/or method for remote operation of a rail vehicle | |
| CN117764331A (zh) | 一种机场飞机牵引车有人和无人驾驶混行调度优化方法 | |
| RU2753778C1 (ru) | Устройство управления движением и маневрированием группы роботизированных и автономных наземных транспортных средств на основе применения многосвязной адаптивной системы управления | |
| JP7469979B2 (ja) | 車両制御システム | |
| US20240311943A1 (en) | Apparatus and methods for point-to-point transportation | |
| US20240300556A1 (en) | Rail authority system and/or method | |
| CN116784727A (zh) | 一种基于滑动窗口的agv与清洁机器人协同控制方法 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 23902725 Country of ref document: EP Kind code of ref document: A1 |