Radar positioning method and system for target AGV

Abstract

The invention provides a radar positioning method and system for a target AGV, and belongs to the technical field of data processing. The method comprises the steps that an arrangement agent receives task description from a target AGV, the task description of the target AGV indicates that radar of the target AGV needs to be positioned, and the radar of the target AGV has been extended out of an initial position by the target AGV; the arrangement agent selects a target positioning agent matched with the space-time prediction state of the target AGV from a plurality of candidate positioning agents according to the task description of the target AGV, and the arrangement agent sends the task description of the arrangement agent to the target positioning agent, the task description of the intelligent agent is arranged to indicate that the radar of the target AGV needs to be positioned in a specified space area; the arrangement intelligent agent receives a positioning result returned by the target positioning intelligent agent and sends the positioning result to the target AGV, and the positioning result is used for determining the position of the radar of the target AGV after the radar of the target AGV is stretched out by the target AGV.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a radar positioning method and system for a target AGV. Background Technology

[0002] When performing cargo handling tasks, automated guided vehicles (AGVs) need to crawl under shelves to lift them up. However, when the robot enters the bottom of the shelf, the lidar mounted on the vehicle can be easily blocked by the shelf above, resulting in insufficient lidar point cloud data, loss of positioning, and even safety hazards.

[0003] To address the aforementioned issues, existing technologies have proposed a scalable lidar solution, which extends the lidar from its initial position above the shelf when an obstruction is detected, thus avoiding the obstruction. However, this solution introduces a new technical problem: after the lidar is extended, its physical position changes, but the positioning system still uses the original position parameters for calculation, leading to a decrease in positioning accuracy. Summary of the Invention

[0004] This invention provides a radar positioning method and system based on a target AGV.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: Firstly, a radar localization method for a target AGV is provided, applied to an orchestration agent. The method includes: the orchestration agent receiving a task description from a target AGV, where the target AGV acts as a working agent; the task description of the target AGV indicates that its radar needs to be located, and the radar of the target AGV has been extended from its initial position by the target AGV; the orchestration agent selecting a target localization agent that matches the spatiotemporal predicted state of the target AGV from multiple candidate localization agents based on the task description of the target AGV; and the orchestration agent sending its own task description to the target localization agent, whereby the task description indicates that the radar of the target AGV needs to be located in a specified spatial region, the specified spatial region being determined based on the spatiotemporal predicted state of the target AGV; and the orchestration agent receiving the localization result returned by the target localization agent and sending the localization result back to the target AGV, the localization result being used by the working agent to determine the position of the target AGV's radar after it has been extended by the target AGV.

[0006] Therefore, this application constructs a three-layer collaborative architecture of "working agent - orchestration agent - positioning agent". After receiving the task description from the target AGV (working agent), the orchestration agent selects the positioning agent based on its spatiotemporal predicted state and sends a task description containing the specified spatial region to the positioning agent, finally returning the positioning result to the target AGV. This architecture enables rapid response and accurate allocation of positioning requests after radar extension, solving the problem of decreased positioning accuracy caused by radar extension in existing technologies.

[0007] Optionally, the orchestration agent selects a target localization agent from multiple candidate localization agents whose spatiotemporal prediction state matches that of the target AGV, based on the task description of the target AGV. This includes: the orchestration agent determining the spatiotemporal prediction state of the target AGV based on the task description of the target AGV; and the orchestration agent determining the spatiotemporal prediction states of multiple candidate localization agents based on the task descriptions of the multiple candidate localization agents; and the orchestration agent selecting a target localization agent from multiple candidate localization agents whose spatiotemporal prediction state matches that of the target AGV, based on the spatiotemporal prediction state of the target AGV and the spatiotemporal prediction states of the multiple candidate localization agents.

[0008] Therefore, the above scheme refines the selection logic of the orchestration agent, clarifying that it needs to simultaneously determine the spatiotemporal prediction state of the target AGV and the spatiotemporal prediction states of multiple candidate localization agents, and make a selection based on the matching of the two. This "two-way prediction" mechanism ensures that the selection decision is not a one-way "who can see the target," but a two-way "who can meet the target at the right time and place." Compared with the one-way selection in the prior art that only considers the target position or only considers resource capabilities, the two-way matching of this application significantly improves the accuracy of the selection, enabling the localization agent to perform observations at the best time and best position, thereby improving the localization success rate.

[0009] Optionally, the task description of the target AGV also indicates the current position and movement trajectory of the target AGV. Based on the task description, the orchestration agent determines the spatiotemporal prediction state of the target AGV, including: the orchestration agent determines the spatial regions occupied by the target AGV at multiple future moments based on the following information: the current position of the target AGV, the movement trajectory of the target AGV, the communication and information processing latency between the target AGV and the orchestration agent, the 3D modeling of the target AGV, and the 3D modeling of the space where the target AGV is located; wherein, the communication and information processing latency is the sum of the transmission latency of the target AGV's task description and the latency required for the orchestration agent to process the target AGV's task description; the transmission latency is the sum of the fixed communication latency and the dynamic communication latency between the orchestration agent and the target AGV; the dynamic communication latency depends on the amount of data in the target AGV's task description; the spatiotemporal prediction state of the target AGV is the spatial region occupied by the target AGV at multiple future moments.

[0010] Therefore, by incorporating "communication and information processing latency" into the spatiotemporal prediction model and clearly defining the composition of latency—the sum of fixed communication latency and dynamic communication latency, where dynamic latency depends on the amount of data in the task description—this design solves the problem of "the location being outdated when information arrives." When the orchestration agent receives the task description, the target AGV has already moved a certain distance; if the current reported position is used directly for prediction, the result will inevitably be biased. This application significantly improves prediction accuracy by using latency correction to make the prediction starting point closer to the real "now." Simultaneously, refining the latency into fixed and dynamic components allows for more precise correction, adapting to latency fluctuations caused by varying data volumes.

[0011] Optionally, the orchestration agent determines the spatial regions occupied by the target AGV at multiple future moments based on the following information: the orchestration agent corrects the current position of the target AGV based on the communication and information processing latency between the target AGV and the orchestration agent and the movement trajectory of the target AGV, obtaining the corrected current position; the orchestration agent determines the spatial region occupied by the target AGV at the corrected current position based on the corrected current position, the 3D model of the target AGV, and the 3D model of the space in which the target AGV is located; the orchestration agent determines the spatial regions occupied by the target AGV at multiple future moments based on the spatial region occupied by the target AGV at the corrected current position, the movement trajectory of the target AGV, the movement speed of the target AGV, the 3D model of the target AGV, and the 3D model of the space in which the target AGV is located, wherein the movement speed of the target AGV is determined based on the task type currently being executed by the target AGV.

[0012] Therefore, the above scheme further refines the specific steps of prediction: first, the current position is corrected using time delay; then, based on the corrected current position and 3D modeling, the currently occupied spatial area is determined; finally, the spatial area at multiple future moments is predicted by combining the movement rate. The movement rate is determined based on the task type, meaning different tasks (empty, fully loaded, lifting, etc.) have different movement rates, making the prediction more scenario-adaptive. This triple combination of "corrected starting point + task rate + 3D region" upgrades the prediction result from "point position" to "spatial region," and considers the impact of task characteristics on the rate, significantly improving accuracy.

[0013] Optionally, multiple candidate localization agents are deployed on corresponding candidate AGVs, resulting in multiple candidate AGVs. For any one of the candidate AGVs, the task description of the candidate AGV also indicates its current position and movement trajectory. The orchestration agent determines the spatiotemporal prediction state of the multiple candidate localization agents based on their task descriptions, including: the orchestration agent determines the spatial regions occupied by the candidate AGVs at multiple future times based on the following information: the current position of the candidate AGV, the movement trajectory of the candidate AGV, the communication and information processing delay between the candidate AGV and the orchestration agent, the 3D modeling of the candidate AGV, and the 3D modeling of the space where the candidate AGV is located. The space where the candidate AGV is located is the same space as the space where the target AGV is located. The communication and information processing delay is a fraction of the candidate AGV's latency. The transmission delay of the task description, the delay required for the orchestration agent to process the task description of the candidate AGV, the estimated transmission delay of the task description, and the delay required for the localization agent deployed on the candidate AGV to process the estimated task description are all included. The transmission delay of the task description of the candidate AGV is the sum of the fixed communication delay between the orchestration agent and the candidate AGV and the first dynamic communication delay. The first dynamic communication delay depends on the amount of data in the task description of the candidate AGV. The estimated task description is the task description that the orchestration agent estimates needs to send to the candidate AGV. The transmission delay of the estimated task description is the sum of the fixed communication delay between the orchestration agent and the candidate AGV and the second dynamic communication delay. The second dynamic communication delay depends on the amount of data in the estimated task description. The spatiotemporal prediction states of multiple candidate localization agents are the spatial regions occupied by multiple candidate AGVs at multiple future times.

[0014] Therefore, by introducing the concept of "estimated task description" and its transmission latency, the orchestration agent, when predicting the future position of a candidate AGV, needs to consider not only the latency currently reported by the candidate AGV, but also the estimated task description to be sent to that AGV and its transmission latency. This "meta-prediction" mechanism solves the recursive problem that "the selection decision itself also generates latency," meaning that if only the current latency is considered, but the latency of sending instructions after selection is ignored, the prediction result will still be biased. By incorporating "decision latency" into the prediction model, a recursive logic of "prediction-selection-re-prediction" is formed, making the prediction more accurate and the decision more reliable. Furthermore, refining the latency into candidate AGV task description latency and estimated task description latency makes the model more complete.

[0015] Optionally, the orchestration agent determines the spatial regions occupied by the candidate AGV at multiple future moments based on the following information: the orchestration agent corrects the current position of the candidate AGV based on the communication and information processing latency between the candidate AGV and the orchestration agent and the movement trajectory of the candidate AGV, obtaining the corrected current position; the orchestration agent determines the spatial region occupied by the candidate AGV at the corrected current position based on the corrected current position, the 3D model of the candidate AGV, and the 3D model of the space in which the candidate AGV is located; the orchestration agent determines the spatial regions occupied by the candidate AGV at multiple future moments based on the spatial region occupied by the candidate AGV at the corrected current position, the movement trajectory of the candidate AGV, the movement speed of the candidate AGV, the 3D model of the candidate AGV, and the 3D model of the space in which the candidate AGV is located, wherein the movement speed of the candidate AGV is determined based on the task type currently being executed by the candidate AGV.

[0016] Therefore, by correcting the current position through time delay, determining the spatial region through 3D modeling, and determining the rate through task type, the spatial region of the candidate AGV at multiple future time points can be obtained. This design ensures that the prediction accuracy of the candidate positioning agent is at the same level as that of the target AGV, making bidirectional matching more accurate.

[0017] Optionally, the spatiotemporal prediction state of the target AGV refers to the spatial region occupied by the target AGV at multiple future times, and the spatiotemporal prediction state of the multiple candidate positioning agents refers to the spatial region occupied by the multiple candidate AGVs at multiple future times. Based on the spatiotemporal prediction state of the target AGV and the spatiotemporal prediction states of the multiple candidate positioning agents, the orchestration agent selects the target positioning agent that matches the spatiotemporal prediction state of the target AGV from the multiple candidate positioning agents. This includes: for any candidate AGV among the multiple candidate AGVs: the orchestration agent determines the spatial overlap between the spatial region occupied by the candidate AGV and the spatial region occupied by the target AGV at each of the multiple times, obtaining multiple spatial overlaps, and summing the multiple spatial overlaps to obtain the overall spatial overlap between the candidate AGV and the target AGV; the orchestration agent selects the AGV with the highest overall spatial overlap from the multiple candidate AGVs as the selected candidate AGV, and the positioning agent deployed on the selected candidate AGV is the target positioning agent.

[0018] Therefore, the above scheme proposes a method for calculating "spatial overlap," which involves calculating the overlap of the spatial regions occupied by the candidate AGV and the target AGV at each time step, summing the results to obtain the overall spatial overlap, and selecting the candidate AGV with the highest overlap as the target localization agent. This design upgrades the matching from "point-to-point" to "region-to-region" and considers the overall matching degree over time. Compared to "instantaneous matching" that only matches a single moment, the overall spatial overlap reflects the observational fit between the candidate AGV and the target AGV throughout the entire time window, avoiding pseudo-optimal solutions that are "optimal at one moment but completely unobservable at other moments." Furthermore, matching based on "spatial regions" rather than "point positions" better reflects the actual physical reality that AGVs have volume.

[0019] Optionally, the method further includes: the orchestration agent selecting the highest spatial overlap from multiple spatial overlap degrees between the selected candidate AGV and the target AGV, and determining the time corresponding to the highest spatial overlap, the preset position of the target AGV's radar at the time corresponding to the highest spatial overlap, and the spatial position occupied by the selected candidate AGV at the time corresponding to the highest spatial overlap; wherein, the task description of the target AGV further indicates the initial position of the target AGV's radar when the target AGV sends its task description, and the preset position of the target AGV's radar at the time corresponding to the highest spatial overlap is determined by the orchestration agent based on the initial position of the target AGV's radar and the position of the target AGV at the time corresponding to the highest spatial overlap; wherein, the task description of the orchestration agent further indicates the time corresponding to the highest spatial overlap, the preset position of the target AGV's radar at the time corresponding to the highest spatial overlap, and the spatial position occupied by the selected candidate AGV at the time corresponding to the highest spatial overlap.

[0020] Therefore, the above scheme selects the moment with the highest overlap from the candidate AGVs with the highest overall spatial overlap, and determines the preset position of the target AGV radar and the spatial position occupied by the candidate AGV at that moment. This "optimal moment selection" mechanism enables the orchestration agent to not only know "who to select," but also "when to observe" and "where to observe." With this information included in the task description, the localization agent does not need to determine the observation timing and location itself, reducing the computational burden on the localization agent while ensuring accurate synchronization of observations. In particular, the preset position of the target AGV radar is determined based on its initial position and its position at that moment, taking into account the offset after the radar extends, making the observation commands more precise.

[0021] Optionally, the positioning results include: the positional relationship between the radar of the selected candidate AGV and the radar of the target AGV at the moment corresponding to the highest spatial overlap, and the positional relationship between the radar of the selected candidate AGV and its initial position when the target AGV sends its task description.

[0022] Therefore, the above scheme clarifies the specific content of the positioning results, including the positional relationship between the radar of the selected candidate AGV and the radar of the target AGV at the optimal observation time, and the positional relationship between the radar of the candidate AGV and the initial position of the radar of the target AGV when the target AGV sends the task description. This "dual positional relationship" design provides the target AGV with complete coordinate transformation information. The target AGV not only knows the relationship between the current radar and the observation radar, but also the radar positional relationship between the observation time and the request time, thus enabling it to accurately calculate the actual position of the radar after it extends from its initial position.

[0023] In a second aspect, a radar positioning system for a target AGV is provided, the system including an orchestration agent for performing the method provided in the first aspect.

[0024] Thirdly, a processing apparatus is provided, comprising: a processor and a memory; the memory is used to store a computer program, which, when executed by the processor, causes the processing apparatus to perform the method described in the first aspect.

[0025] In one possible design, the processing device described in the third aspect may further include a transceiver. This transceiver may be a transceiver circuit or an interface circuit. The transceiver can be used for communication between the processing device described in the third aspect and other processing devices.

[0026] In this embodiment of the invention, the processing device described in the third aspect may be a terminal, or a chip (system) or other component or assembly disposed in the terminal, or a system containing the terminal. Attached Figure Description

[0027] Figure 1 A schematic diagram of the architecture of a radar positioning system for a target AGV provided in an embodiment of the present invention; Figure 2 A schematic diagram of a target AGV in a radar positioning system for a target AGV provided in an embodiment of the present invention; Figure 3 A schematic flowchart illustrating a radar positioning method for a target AGV provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of the processing device provided in an embodiment of the present invention. Detailed Implementation

[0028] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0029] In this embodiment of the invention, "instruction" can include direct and indirect instructions, as well as explicit and implicit instructions. The information indicated by a certain piece of information is called the information to be instructed. In specific implementation, there are many ways to instruct the information to be instructed, such as, but not limited to, directly instructing the information to be instructed, such as the information to be instructed itself or its index. It can also indirectly instruct the information to be instructed by instructing other information, where there is a correlation between the other information and the information to be instructed. It can also instruct only a part of the information to be instructed, while the other parts are known or pre-agreed upon. For example, the instruction of specific information can be achieved by using a pre-agreed (e.g., protocol-defined) arrangement of various pieces of information, thereby reducing instruction overhead to some extent. Simultaneously, common parts of various pieces of information can be identified and uniformly indicated to reduce the instruction overhead caused by individually indicating the same information.

[0030] Furthermore, the specific indication method can also be any existing indication method, such as, but not limited to, the above-mentioned indication methods and their various combinations. Specific details of various indication methods can be found in existing technologies, and will not be elaborated upon here. As described above, for example, when multiple pieces of information of the same type need to be indicated, the indication methods for different pieces of information may differ. In specific implementation, the required indication method can be selected according to specific needs. This embodiment of the invention does not limit the selected indication method; therefore, the indication methods involved in this embodiment of the invention should be understood to cover various methods that enable the party to be indicated to obtain the information to be indicated.

[0031] In this embodiment of the invention, descriptions such as "when," "under the circumstances," "if," and "if" all refer to the device making corresponding processing under certain objective circumstances, and are not limited to a specific time. They do not require the device to make a judgment action during implementation, nor do they imply any other limitations.

[0032] In the description of the embodiments of the present invention, unless otherwise stated, " / " indicates that the objects before and after are in an "or" relationship. For example, A / B can represent A or B. "And / or" in the embodiments of the present invention is merely a description of the relationship between the related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, and B alone, where A and B can be singular or plural. Furthermore, in the description of the embodiments of the present invention, unless otherwise stated, "multiple" refers to two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple. Additionally, to facilitate a clear description of the technical solutions of the embodiments of the present invention, the terms "first" and "second" are used in the embodiments of the present invention to distinguish identical or similar items with essentially the same function and effect. Those skilled in the art will understand that the terms "first," "second," etc., do not limit the quantity or order of execution, and that "first," "second," etc., are not necessarily different. Furthermore, in the embodiments of this invention, words such as "exemplary" or "for example" are used to indicate that something is being described as an example, illustration, or illustration. Any embodiment or design described as "exemplary" or "for example" in the embodiments of this invention should not be construed as being more preferred or advantageous than other embodiments or design solutions. Specifically, the use of words such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner for ease of understanding.

[0033] To facilitate understanding of the embodiments of the present invention, firstly, let's take... Figure 1 The radar positioning system of the target AGV shown in the figure is used as an example to illustrate in detail the radar positioning method of the target AGV applicable to the embodiments of the present invention.

[0034] For example, Figure 1 This is a schematic diagram of the architecture of a radar positioning system for a target AGV provided in an embodiment of the present invention. Figure 1 As shown, the system mainly includes three types of intelligent agents: Orchestration Agent, Worker Agent, and Localization Agent. These agents are interconnected via a wireless communication network to collaboratively achieve precise positioning of the target AGV after it extends beyond the radar.

[0035] The orchestration agent is the core decision-making unit of the system, typically deployed on a central server, edge computing nodes, or cloud platform, possessing strong computing power and storage resources. Its form can be an industrial server, industrial control computer, or virtualized instance. The orchestration agent is responsible for receiving positioning requests from worker agents. Based on information such as current pose, movement trajectory, and task type carried in the request, and combined with a pre-built environmental map and dynamic resource pool, it performs spatiotemporal prediction and intelligent matching, selecting the optimal agent from multiple candidate positioning agents, and generating a task description containing the specified observation time and spatial region. The orchestration agent communicates bidirectionally with worker agents and positioning agents via low-latency wireless networks (such as Wi-Fi 6 or 5G) to ensure real-time transmission of instructions and data.

[0036] The working agent is deployed on an Automated Guided Vehicle (AGV), which is a lurking, lifting mobile robot equipped with a retractable LiDAR. The working agent typically exists in the form of an embedded controller or industrial PC, integrated with the AGV's motion control unit, radar drive unit, and communication module. Figure 2 This is a schematic diagram of a target AGV in a radar positioning system for a target AGV, provided as an embodiment of the present invention. Figure 2 As shown, the working agent is responsible for real-time monitoring of whether the radar is obstructed by the shelf. When obstruction is detected, it controls the radar to extend from its initial position. The AGV's radar is initially located at position A (retracted state). When obstruction is detected, the radar extends to position A' (extended state), and the extension height can be dynamically adjusted according to the degree of obstruction. The radar's extension mechanism is driven by a motor and equipped with a position sensor, which can provide real-time feedback on the extension height for the working agent to use. At this time, the working agent can immediately generate a positioning request and send it to the orchestration agent via a wireless network. The request includes information such as the AGV's identifier, current position, movement trajectory, task type, movement speed, and radar extension status. The working agent then waits for the positioning result returned by the orchestration agent and uses the result to correct its own positioning algorithm to ensure the continuity and accuracy of positioning.

[0037] The positioning agent is the system's observation execution unit and is also deployed on the AGVs. In other words, these AGVs play a dual role in the system: they can act as working agents during their own tasks, and switch to positioning agents when idle or when tasks permit, providing observation services to other AGVs. This dynamic role switching mechanism is uniformly scheduled by the orchestration agent, dynamically allocated based on the real-time status and task load of each AGV. The positioning agent includes a lidar sensor, a local processor, and a communication module. It can receive task descriptions (including observation time, target spatial area, etc.) from the orchestration agent, accurately observe the target AGV at the specified time, calculate the relative positional relationship between itself and the target AGV's radar, and return the positioning results to the orchestration agent. The positioning agent's local processor is responsible for computational tasks such as point cloud acquisition, feature extraction, and coordinate transformation to reduce the burden on the central node and lower communication latency.

[0038] Each AGV registers with the orchestration agent during initialization and periodically reports its own status (such as current position, speed, task type, radar status, remaining battery power, etc.). The orchestration agent maintains a dynamic resource pool, marking each AGV as "working mode," "idle mode," or "location mode" based on its status. When an AGV needs location services, the orchestration agent selects AGVs from the resource pool that meet the spatiotemporal matching conditions as candidate location agents and issues them a task description. The selected AGV temporarily switches to the location agent role to perform the observation task; after completing the task, it can revert to its original role or remain idle and wait for the next scheduling. This dynamic role allocation mechanism eliminates the need for additional dedicated location equipment, fully utilizes existing AGV resources, and reduces system costs.

[0039] The communication network serves as the link connecting the various intelligent agents. This embodiment employs industrial-grade wireless LAN standards, such as IEEE 802.11ax (Wi-Fi 6) or 5G URLLC (Ultra-Reliable Low-Latency Communication), to ensure millisecond-level end-to-end latency and communication reliability. A bidirectional message queue is established between the orchestration agent and each AGV, using Message Queuing Telemetry Transport (MQTT) or Data Distribution Service (DDS) protocols for data exchange, supporting publish / subscribe modes for convenient real-time status updates and task distribution. Through this architecture, this invention enables the orchestration agent to uniformly schedule and locate the intelligent agents after radar extension obstruction, accurately matching observation timing and spatial region based on spatiotemporal prediction, and ultimately feeding back the positioning results to the working intelligent agents in real time, effectively solving the problem of positioning loss for lurking mobile robots in shelf obstruction scenarios.

[0040] Figure 3This is a flowchart illustrating a radar positioning method for a target AGV provided in an embodiment of the present invention. This method is applicable to the aforementioned system.

[0041] The specific process is as follows: S301, the orchestration agent receives the task description from the target AGV.

[0042] The target AGV, acting as a working intelligent agent, extends its LiDAR from its initial position to a second position via a telescopic mechanism after detecting that its LiDAR is obstructed by the shelf. At this point, the target AGV needs to obtain the precise position of the extended LiDAR to maintain positioning accuracy. Therefore, it generates a positioning request and sends it to the orchestration agent via a wireless network. This positioning request is the task description described in this step.

[0043] The task description for the target AGV indicates that its radar needs to be located, and that the radar has been extended from its initial position. For example, the task description might include a "Request Type" field with a value of "Radar Location" and a "Radar Status" field with a value of "Extended," clearly specifying the purpose of the request and the current radar status. The task description also indicates the target AGV's identifier, such as the device ID "AGV-001," used to uniquely identify the source of the request for the orchestrated agent. Furthermore, the task description indicates the target AGV's current position and its movement trajectory. The current position is typically represented as coordinates (x, y, yaw) in the global coordinate system, such as (12.34 meters, 56.78 meters, 1.57 radians); the movement trajectory can be a sequence of pose points sampled over a past period, such as recording a position point every 0.5 seconds over the past 2 seconds, reflecting the AGV's movement trend. The task description also indicates the initial position of the radar when the target AGV sends the task description. This initial position is the original installation position of the radar before it extends, which is fixed and known relative to the AGV body coordinate system. For example, it is expressed as a three-dimensional offset (0.15 m, 0.00 m, 0.50 m) relative to the center point of the AGV.

[0044] The target AGV encapsulates the above information into a structured data packet conforming to a preset protocol and sends it to the orchestration agent via the onboard communication module. The orchestration agent, acting as the central node, continuously listens to the designated port, receives and parses the task description, extracts various information from it, stores it in memory, and provides input for subsequent spatiotemporal prediction and localization agent selection.

[0045] S302, the orchestration agent selects a target localization agent from multiple candidate localization agents based on the task description of the target AGV, which matches the spatiotemporal predicted state of the target AGV, and the orchestration agent sends the task description of the orchestration agent to the target localization agent.

[0046] This step further includes the following steps 1 and 2.

[0047] Step 1: The orchestration agent determines the spatiotemporal prediction state of the target AGV based on the task description of the target AGV, and the orchestration agent determines the spatiotemporal prediction state of multiple candidate localization agents based on the task descriptions of multiple candidate localization agents.

[0048] Spatiotemporal prediction of the target AGV: Since the target AGV's task description also indicates its current position and movement trajectory, the orchestration agent can determine the spatial regions occupied by the target AGV at multiple future moments based on the following information: the target AGV's current position, its movement trajectory, the communication and information processing latency between the target AGV and the orchestration agent, the 3D modeling of the target AGV, and the 3D modeling of the space in which the target AGV is located. The communication and information processing latency is the sum of the transmission latency of the target AGV's task description and the latency required for the orchestration agent to process the target AGV's task description. The transmission latency is the sum of the fixed communication latency and the dynamic communication latency between the orchestration agent and the target AGV, with the dynamic communication latency depending on the amount of data in the target AGV's task description. The spatiotemporal prediction state of the target AGV is the spatial region occupied by the target AGV at multiple future moments.

[0049] Based on this, the specific implementation process of the orchestration agent to determine the spatiotemporal prediction state of the target AGV is as follows.

[0050] The orchestration agent first corrects the target AGV's current position based on the communication and information processing latency between the target AGV and the agent, combined with the target AGV's movement trajectory, to obtain the corrected current position. As mentioned earlier, there is a time difference between the moment the target AGV sends the task description and the moment it is received and processed by the orchestration agent; during this time, the target AGV continues to move. Without correction, directly using the "past" position carried in the task description for prediction would lead to a prediction result deviating from the actual situation. Therefore, the orchestration agent combines the communication and information processing latency with the target AGV's movement trajectory to calculate the target AGV's displacement within this time delay, and then extrapolates the reported current position backward to obtain a corrected current position that is closer to the actual current position. For example, assuming the target AGV reports a position of (12.34, 56.78), and the movement trajectory shows that it is moving eastward at a speed of 1.5 m / s, with a communication and processing delay of 80 milliseconds, then the corrected current position is approximately (12.34 + 1.5 × 0.08, 56.78) = (12.46, 56.78).

[0051] The orchestration agent then determines the spatial region occupied by the target AGV at its corrected current position, based on the corrected current position, the 3D model of the target AGV, and the 3D model of the space in which the target AGV is located. The 3D model of the target AGV describes its own geometry, such as its length, width, height, and the radar's installation location; the 3D model of the space in which the target AGV is located describes static objects in the environment, such as obstacles, shelves, and walls. The orchestration agent uses the corrected current position as the coordinates of the AGV's center point, combined with its 3D model, to calculate the 3D spatial region occupied by the AGV at that position. This region can be represented by a 3D bounding box, for example, the minimum and maximum coordinate range in the global coordinate system.

[0052] The orchestration agent further determines the spatial regions occupied by the target AGV at multiple future time points based on the spatial region occupied by the target AGV at its corrected current position, the target AGV's movement trajectory, the target AGV's movement speed, the target AGV's 3D model, and the 3D model of the space in which the target AGV is located. The target AGV's movement speed is determined based on the type of task it is currently performing. For example, when the AGV is performing an empty transport task, its movement speed might be 1.5 m / s; when performing a fully loaded lifting task, its movement speed might decrease to 0.8 m / s. Starting from the corrected current position, the orchestration agent extrapolates forward along the direction indicated by the movement trajectory, at the speed determined by the task type, to obtain the AGV's center point position at a series of future time points (e.g., 0.1 seconds, 0.2 seconds, 0.3 seconds, etc.). Combining 3D modeling and environmental modeling, it calculates the spatial region occupied by the AGV at each time point. These spatial regions constitute the spatiotemporal predicted state of the target AGV at multiple future time points.

[0053] Spatiotemporal prediction of state for multiple candidate localization agents: Multiple candidate positioning agents are deployed on corresponding candidate AGVs, resulting in a total of multiple candidate AGVs. For any candidate AGV among these multiple candidate AGVs, and given the candidate AGV's task description indicating its current position and movement trajectory, the orchestration agent determines the spatial regions occupied by the candidate AGVs at multiple future moments (these moments are the same as the target AGV's multiple future moments) based on the following information: the candidate AGV's current position, its movement trajectory, the communication and information processing latency between the candidate AGV and the orchestration agent, the candidate AGV's 3D model, and the 3D model of the space where the candidate AGV is located. The space where the candidate AGV is located is the same as the space where the target AGV is located.

[0054] The composition of communication and information processing latency is more complex. Specifically, this latency is the sum of the transmission latency of the candidate AGV's task description, the latency required for the orchestration agent to process the candidate AGV's task description, the estimated transmission latency of the task description, and the latency required for the positioning agent deployed by the candidate AGV to process the estimated task description. The transmission latency of the candidate AGV's task description is the sum of the fixed communication latency between the orchestration agent and the candidate AGV and the first dynamic communication latency, the first dynamic communication latency depending on the amount of data in the candidate AGV's task description. The estimated task description is the task description that the orchestration agent estimates needs to send to the candidate AGV; the transmission latency of the estimated task description is the sum of the fixed communication latency between the orchestration agent and the candidate AGV and the second dynamic communication latency, the second dynamic communication latency depending on the amount of data in the estimated task description. The spatiotemporal prediction states of multiple candidate positioning agents represent the spatial regions occupied by multiple candidate AGVs at multiple future times.

[0055] The reason for this complex latency is that when the orchestration agent predicts the future position of a candidate AGV, it needs to consider not only the latency caused by the candidate AGV reporting its own state, but also the latency caused by sending positioning task commands to the candidate AGV in the future. Since the candidate AGV itself is also in motion, if its future position is predicted only based on its current state, but the time required to send the command is ignored, there will still be a discrepancy between the predicted result and the actual position when the command arrives. Therefore, the orchestration agent adopts a "meta-prediction" approach, incorporating the transmission and processing latency of the estimated task description into the prediction model.

[0056] Based on this, the specific implementation process of the orchestration agent to determine the spatiotemporal prediction state of candidate AGVs is as follows.

[0057] The orchestration agent first corrects the current position of the candidate AGV based on the communication and information processing delay between the candidate AGV and the agent, combined with the candidate AGV's movement trajectory, to obtain the corrected current position. Similar to the principle of correcting the target AGV's current position based on delay, the orchestration agent combines the aforementioned complex communication and information processing delay with the candidate AGV's movement trajectory to calculate the displacement of the candidate AGV within this delay, and then extrapolates the current position reported by the candidate AGV backward to obtain the corrected current position.

[0058] The orchestration agent then determines the spatial region occupied by the candidate AGV at its corrected current position based on the corrected current position, the 3D model of the candidate AGV, and the 3D model of the space in which the candidate AGV is located. This process is the same as that for the target AGV, which calculates the bounding box of the candidate AGV at its current corrected position using the 3D model.

[0059] The orchestration agent further determines the spatial regions occupied by the candidate AGV at multiple future time points based on the spatial region occupied by the candidate AGV at its corrected current position, the candidate AGV's movement trajectory, the candidate AGV's movement speed, the candidate AGV's 3D model, and the 3D model of the space in which the candidate AGV is located. The candidate AGV's movement speed is determined based on the task type it is currently performing; for example, the speed is faster when idle and slower when fully loaded. Starting from the corrected current position, the orchestration agent extrapolates forward along the direction indicated by the movement trajectory, at the speed determined by the task type, to obtain the center point position of the candidate AGV at each future time point, the same as the target AGV, and then calculates the spatial region occupied by the candidate AGV at each time point. These spatial regions constitute the spatiotemporal predicted state of the candidate AGV at multiple future time points.

[0060] Through the above processing, the orchestration agent obtains the spatial region sequence of the target AGV at each future time, as well as the spatial region sequence of each candidate AGV at the same time sequence, laying the foundation for subsequent matching and selection.

[0061] Step 2: The orchestration agent selects the target positioning agent that matches the spatiotemporal prediction state of the target AGV based on the spatiotemporal prediction state of the target AGV and the spatiotemporal prediction states of multiple candidate positioning agents, and sends the task description of the orchestration agent to the target positioning agent.

[0062] As described above, the spatiotemporal prediction state of the target AGV refers to the spatial regions occupied by the target AGV at multiple future moments, and the spatiotemporal prediction states of multiple candidate localization agents refer to the spatial regions occupied by multiple candidate AGVs at multiple future moments. Therefore, the orchestration agent needs to evaluate the spatiotemporal matching degree between the target AGV and each candidate AGV in order to select the optimal observer executor.

[0063] For any one of multiple candidate AGVs, the orchestration agent determines the spatial overlap between the spatial region occupied by the candidate AGV and the spatial region occupied by the target AGV at each of multiple time points. Spatial overlap measures whether two AGVs can form an effective observation relationship at the same time—when the spatial regions of two AGVs are close or overlap, it means that the candidate AGV has a higher probability of observing the target AGV. There are various ways to calculate spatial overlap, such as calculating the distance between the center points of the two spatial regions, the intersection volume, or the minimum observation angle. In one embodiment, spatial overlap is defined as whether two spatial regions intersect; if they intersect, the overlap is 1, otherwise it is 0. In another embodiment, spatial overlap is quantified as the reciprocal of the distance between the center points of the two regions; the closer the distance, the higher the overlap.

[0064] The orchestration agent calculates a spatial overlap degree at each time point, resulting in a set of spatial overlap degree values ​​over a time series. For example, overlap degrees c1, c2, c3, etc., are obtained at times t1, t2, t3, etc. The orchestration agent sums these spatial overlap degrees to obtain the overall spatial overlap degree between the candidate AGV and the target AGV. The overall spatial overlap degree reflects the overall matching degree between the candidate AGV and the target AGV over the entire future time window.

[0065] The orchestration agent repeats the above calculations for each candidate AGV to obtain the overall spatial overlap degree of each candidate AGV. Then, the orchestration agent selects the AGV with the highest overall spatial overlap degree from among the multiple candidate AGVs as the selected candidate AGV. The localization agent deployed on this selected candidate AGV is the target localization agent. Selecting the AGV with the highest overall spatial overlap degree means that this AGV has the highest spatial proximity to the target AGV within the future time window, and is most likely to be in a suitable observation position when the target AGV needs to be observed.

[0066] Therefore, the orchestration agent identifies the target localization agent and sends a task description to it. This task description instructs the orchestration agent to locate the target AGV's radar within a specified spatial region, determined based on the target AGV's spatiotemporally predicted state. For example, the specified spatial region could be the area occupied by the target AGV at one or more future moments, or the union or envelope of these spatial regions. The task description may also include information such as the target AGV's identifier, the optimal observation time, and the preset location of the target AGV's radar to guide the localization agent in accurate observation.

[0067] S303, the orchestration agent receives the positioning result returned by the target positioning agent and sends the positioning result to the target AGV.

[0068] The positioning results are used by the working agent to determine the position of the target AGV's radar after the target AGV extends.

[0069] In step S302 above, the orchestration agent selects the highest spatial overlap from multiple spatial overlap values ​​between the selected candidate AGVs and the target AGV, and determines the time corresponding to this highest spatial overlap, the preset position of the target AGV's radar at that time, and the spatial position occupied by the selected candidate AGV at that time (i.e., the designated spatial area mentioned above). The target AGV's task description indicates the initial position of its radar when the target AGV sends the task description. The preset position of the target AGV's radar at the time corresponding to the highest spatial overlap is determined by the orchestration agent based on the initial position of the target AGV's radar and the target AGV's position at that time. It should be noted that this preset position is not the actual position after the radar extends, but rather refers to the spatial position that the target AGV's radar should be in if it is not extended and is still in its initial installation state, when the target AGV moves to that position at that time. In other words, the preset position is the result of the radar's initial installation position shifting with the movement of the AGV itself; it is equivalent to a virtual reference point used for subsequent calculations of the actual position after the radar extends.

[0070] The orchestration agent incorporates the above information into the task description sent to the target localization agent. This task description indicates at least: the time corresponding to the highest spatial overlap, the preset radar position of the target AGV at that time, and the spatial position occupied by the selected candidate AGV at that time. The spatial position occupied by the selected candidate AGV is the spatial region indicated by the candidate AGV's spatiotemporal prediction state at that time, typically represented by the AGV's center point coordinates or a three-dimensional bounding box.

[0071] After receiving the task description from the orchestration agent, the target localization agent performs observation at the specified time. Specifically, at that time, the target localization agent activates its lidar to scan the spatial area where the target AGV is located. Since the task description has already indicated the preset position of the target AGV at that time, as well as the spatial position of the candidate AGV itself, the target localization agent can adjust the radar's scanning angle or focus area accordingly to ensure that the radar can be effectively captured after the target AGV extends.

[0072] After observation, the target localization agent performs localization calculations based on the collected point cloud data. The calculation process includes two aspects: First, the target localization agent determines the positional relationship between its own radar and the target AGV's radar at the moment corresponding to the highest spatial overlap. Since the spatial position of the target localization agent at that moment is known (i.e., the spatial position occupied by the candidate AGV at that moment as indicated in the task description), and the actual position of the target AGV's radar after it extends can be extracted from feature points in the point cloud data (such as radar reflectors or mounting brackets), the target localization agent can calculate the relative displacement vector and orientation relationship between its own radar and the radar after the target AGV extends, for example, expressed as the offset (Δx1, Δy1, Δz1) relative to the candidate AGV's vehicle coordinate system.

[0073] Second, the target localization agent determines the positional relationship between its own radar and the initial position of the target AGV's radar at the moment the target AGV sends its task description. Here, the target localization agent utilizes the initial position information of the target AGV's radar provided in the task description, as well as its own spatial position at the time of transmission (this position can be obtained from the candidate AGV's own historical trajectory record). The target localization agent uses the initial position as a virtual reference point for the target AGV's radar and calculates the relative displacement vector of its own radar to this virtual reference point at that historical moment, for example, represented as (Δx2, Δy2, Δz2). These two positional relationships together constitute the complete localization result.

[0074] The target localization agent encapsulates the two sets of positional relationships calculated above into a localization result and returns it to the orchestration agent via a wireless network. Upon receiving the localization result, the orchestration agent forwards it to the target AGV.

[0075] As a working intelligent agent, the target AGV, upon receiving the positioning result, uses the information to determine its precise position after its radar is extended. Specifically, the target AGV knows its own radar's initial position (fixed relative to the vehicle coordinate system) at the time the task description was sent, and the positional relationship between the radar of the candidate positioning agent and that initial position at that time (i.e., Δx2, Δy2, Δz2). Simultaneously, the target AGV knows the positional relationship between the radar of the candidate positioning agent and its own extended radar at the observation time (i.e., Δx1, Δy1, Δz1). Through coordinate transformation and relative relationship calculations, the target AGV can deduce the absolute spatial position of the extended radar at the current moment, thereby correcting its own positioning parameters to ensure high-precision positioning even with the radar extended.

[0076] In this way, the target AGV determines its specific position after its radar extends at the moment corresponding to the highest spatial overlap, based on the positioning results, that is, according to the above two positional relationships.

[0077] In summary, this application constructs a three-layer collaborative architecture consisting of a "working agent, an orchestration agent, and a positioning agent." After receiving the task description from the target AGV (working agent), the orchestration agent selects a positioning agent based on its spatiotemporal predicted state and sends a task description containing a specified spatial region to the positioning agent. Finally, it returns the positioning result to the target AGV. This architecture enables rapid response and accurate allocation of positioning requests after radar extension, solving the problem of decreased positioning accuracy caused by radar extension in existing technologies.

[0078] Figure 4 This is a schematic diagram of the structure of a processing device provided in an embodiment of the present invention. Exemplarily, this processing device may be a network device, or a chip (system) or other component or assembly that can be disposed in a network device. Figure 4 As shown, the processing device 400 may include a processor 401. Optionally, the processing device 400 may also include a memory 402 and / or a transceiver 403. The processor 401 is coupled to the memory 402 and the transceiver 403, for example, via a communication bus.

[0079] The following is combined Figure 4 A detailed description of each component of the processing equipment 400 is provided below: The processor 401 is the control center of the processing device 400. It can be a single processor or a collective term for multiple processing elements. For example, the processor 401 can be one or more central processing units (CPUs), application-specific integrated circuits (ASICs), or one or more integrated circuits configured to implement embodiments of the present invention, such as one or more digital signal processors (DSPs), or one or more field-programmable gate arrays (FPGAs).

[0080] Optionally, the processor 401 can perform various functions of the processing device 400 by running or executing software programs stored in the memory 402 and calling data stored in the memory 402.

[0081] In a specific implementation, as one example, processor 401 may include one or more CPUs, for example... Figure 4 CPU0 and CPU1 are shown in the diagram.

[0082] In a specific implementation, as one embodiment, the processing device 400 may also include multiple processors. Each of these processors may be a single-core processor (single-CPU) or a multi-core processor (multi-CPU). Here, a processor may refer to one or more devices, circuits, and / or processing cores used for processing data (e.g., computer program instructions).

[0083] The memory 402 is used to store the software program that executes the solution of the present invention, and is controlled by the processor 401 to execute it. The specific implementation method can be referred to the above method embodiment, and will not be repeated here.

[0084] Optionally, the memory 402 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. The memory 402 may be integrated with the processor 401 or exist independently, and may be connected via the interface circuit of the processing device 400. Figure 4 (Not shown in the image) is coupled to processor 401, and this embodiment of the invention does not specifically limit this.

[0085] Transceiver 403 is used for communication with other processing devices. For example, if processing device 400 is a terminal, transceiver 403 can be used to communicate with a network device or with another terminal device. As another example, if processing device 400 is a network device, transceiver 403 can be used to communicate with a terminal or with another network device.

[0086] Alternatively, transceiver 403 may include a receiver and a transmitter. Figure 4 (Not shown separately). The receiver is used to implement the receiving function, and the transmitter is used to implement the transmitting function.

[0087] Alternatively, the transceiver 403 can be integrated with the processor 401, or it can exist independently and be connected via the interface circuit of the processing device 400. Figure 4 (Not shown in the image) is coupled to processor 401, and this embodiment of the invention does not specifically limit this.

[0088] Understandable Figure 4 The structure of the processing device 400 shown does not constitute a limitation on the processing device. Actual processing devices may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0089] Furthermore, the technical effects of the processing device 400 can be referred to the technical effects of the method described in the above method embodiments, and will not be repeated here.

[0090] It should be understood that the processor in the embodiments of the present invention can be a central processing unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0091] It should also be understood that the memory in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced synchronous DRAM (ESDRAM), synchronous linked DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0092] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A radar positioning method for a target AGV, characterized in that, Applied to orchestrating intelligent agents, the method includes: The orchestration agent receives a task description from the target AGV, which acts as the working agent. The task description of the target AGV indicates that the radar of the target AGV needs to be located, and the radar of the target AGV has been extended from the initial position by the target AGV. The orchestration agent selects a target positioning agent from multiple candidate positioning agents based on the task description of the target AGV, which matches the spatiotemporal prediction state of the target AGV. The orchestration agent also sends its task description to the target positioning agent, wherein the task description indicates that the radar of the target AGV needs to be located in a specified spatial area, and the specified spatial area is determined based on the spatiotemporal prediction state of the target AGV. The orchestration agent receives the positioning result returned by the target positioning agent and sends the positioning result to the target AGV. The positioning result is used by the working agent to determine the position of the target AGV's radar after the target AGV extends.

2. The method according to claim 1, characterized in that, The orchestration agent selects a target localization agent from multiple candidate localization agents based on the task description of the target AGV, which matches the spatiotemporal predicted state of the target AGV. This includes: The orchestration agent determines the spatiotemporal prediction state of the target AGV based on the task description of the target AGV, and the orchestration agent determines the spatiotemporal prediction state of the multiple candidate positioning agents based on the task descriptions of the multiple candidate positioning agents. The orchestration agent selects the target positioning agent that matches the spatiotemporal prediction state of the target AGV from among the multiple candidate positioning agents, based on the spatiotemporal prediction state of the target AGV and the spatiotemporal prediction states of the multiple candidate positioning agents.

3. The method according to claim 2, characterized in that, The task description of the target AGV also indicates the current position and movement trajectory of the target AGV. Based on the task description, the orchestration agent determines the spatiotemporal prediction state of the target AGV, including: The orchestration agent determines the spatial regions occupied by the target AGV at multiple future moments based on the following information: the current position of the target AGV, the movement trajectory of the target AGV, the communication and information processing delay between the target AGV and the orchestration agent, the 3D modeling of the target AGV, and the 3D modeling of the space where the target AGV is located. Wherein, the communication and information processing latency is the sum of the transmission latency of the target AGV's task description and the latency required for the orchestration agent to process the target AGV's task description; the transmission latency is the sum of the fixed communication latency and dynamic communication latency between the orchestration agent and the target AGV; the dynamic communication latency depends on the amount of data in the target AGV's task description; the spatiotemporal prediction state of the target AGV is the spatial region occupied by the target AGV at multiple future moments.

4. The method according to claim 3, characterized in that, The orchestration agent determines the spatial regions occupied by the target AGV at multiple future moments based on the following information: The orchestration agent corrects the current position of the target AGV based on the communication and information processing delay between the target AGV and the orchestration agent and the movement trajectory of the target AGV, thus obtaining the corrected current position. The orchestration agent determines the spatial area occupied by the target AGV at the corrected current position based on the corrected current position, the 3D model of the target AGV, and the 3D model of the space where the target AGV is located. The orchestration agent determines the spatial regions occupied by the target AGV at multiple future moments based on the spatial region occupied by the target AGV at the corrected current position, the target AGV's movement trajectory, the target AGV's movement speed, the target AGV's 3D model, and the 3D model of the space where the target AGV is located. The movement speed of the target AGV is determined based on the task type currently being executed by the target AGV.

5. The method according to claim 2, characterized in that, The plurality of candidate positioning agents are deployed on corresponding candidate AGVs, with a total of plurality of candidate AGVs. For any one of the plurality of candidate AGVs, the task description of the candidate AGV also indicates the current position and the movement trajectory of the candidate AGV. The orchestration agent determines the spatiotemporal prediction state of the plurality of candidate positioning agents based on the task descriptions of the plurality of candidate positioning agents, including: The orchestration agent determines the spatial regions occupied by the candidate AGV at multiple future moments based on the following information: the current position of the candidate AGV, the movement trajectory of the candidate AGV, the communication and information processing delay between the candidate AGV and the orchestration agent, the 3D model of the candidate AGV, and the 3D model of the space where the candidate AGV is located, wherein the space where the candidate AGV is located is the same space as the space where the target AGV is located. The communication and information processing latency is the sum of the transmission latency of the candidate AGV's task description, the latency required for the orchestration agent to process the candidate AGV's task description, the estimated transmission latency of the task description, and the latency required for the positioning agent deployed on the candidate AGV to process the estimated task description. The transmission latency of the candidate AGV's task description is the sum of the fixed communication latency between the orchestration agent and the candidate AGV and the first dynamic communication latency, where the first dynamic communication latency depends on the data volume of the candidate AGV's task description. The estimated task description is the task description that the orchestration agent estimates needs to send to the candidate AGV. The transmission latency of the estimated task description is the sum of the fixed communication latency between the orchestration agent and the candidate AGV and the second dynamic communication latency, where the second dynamic communication latency depends on the data volume of the estimated task description. The spatiotemporal prediction states of the multiple candidate positioning agents are the spatial regions occupied by the multiple candidate AGVs at multiple future times.

6. The method according to claim 5, characterized in that, The orchestration agent determines the spatial regions to be occupied by the candidate AGVs at multiple future times based on the following information: The orchestration agent corrects the current position of the candidate AGV based on the communication and information processing delay between the candidate AGV and the orchestration agent and the movement trajectory of the candidate AGV, thus obtaining the corrected current position. The orchestration agent determines the spatial region occupied by the candidate AGV at the corrected current position based on the corrected current position, the 3D model of the candidate AGV, and the 3D model of the space where the candidate AGV is located. The orchestration agent determines the spatial regions occupied by the candidate AGV at multiple future moments based on the spatial region occupied by the candidate AGV at the corrected current position, the movement trajectory of the candidate AGV, the movement speed of the candidate AGV, the 3D model of the candidate AGV, and the 3D model of the space where the candidate AGV is located. The movement speed of the candidate AGV is determined based on the task type currently being executed by the candidate AGV.

7. The method according to any one of claims 2-6, characterized in that, The spatiotemporal prediction state of the target AGV refers to the spatial regions occupied by the target AGV at multiple future times. The spatiotemporal prediction states of the multiple candidate positioning agents refer to the spatial regions occupied by the multiple candidate AGVs at the multiple future times. The orchestration agent selects the target positioning agent that matches the spatiotemporal prediction state of the target AGV from the multiple candidate positioning agents based on the spatiotemporal prediction states of the target AGV and the multiple candidate positioning agents, including: For any one of the multiple candidate AGVs: the orchestration agent determines the degree of spatial overlap between the spatial region occupied by the candidate AGV and the spatial region occupied by the target AGV at each of the multiple time points, obtains multiple degrees of spatial overlap, and sums the multiple degrees of spatial overlap to obtain the overall degree of spatial overlap between the candidate AGV and the target AGV; The orchestration agent selects the AGV with the highest overall spatial overlap from the multiple candidate AGVs as the selected candidate AGV, and the positioning agent deployed on the selected candidate AGV is the target positioning agent.

8. The method according to claim 7, characterized in that, The method further includes: The orchestration agent selects the highest spatial overlap from multiple spatial overlaps between the selected candidate AGV and the target AGV, and determines the time corresponding to the highest spatial overlap, the preset position of the radar of the target AGV at the time corresponding to the highest spatial overlap, and the spatial position occupied by the selected candidate AGV at the time corresponding to the highest spatial overlap. The task description of the target AGV also indicates the initial position of the radar of the target AGV when the target AGV sends the task description. The preset position of the radar of the target AGV at the time corresponding to the highest spatial overlap is determined by the orchestration agent based on the initial position of the radar of the target AGV and the position of the target AGV at the time corresponding to the highest spatial overlap. The task description of the orchestration agent also indicates the time corresponding to the highest spatial overlap, the preset position of the radar of the target AGV at the time corresponding to the highest spatial overlap, and the spatial position occupied by the selected candidate AGV at the time corresponding to the highest spatial overlap.

9. The method according to claim 8, characterized in that, The positioning results include: the positional relationship between the radar of the selected candidate AGV and the radar of the target AGV at the time corresponding to the highest spatial overlap, and the positional relationship between the radar of the selected candidate AGV and the initial position when the target AGV sends the task description of the target AGV.

10. A radar positioning system for a target AGV, characterized in that, The system includes an orchestration agent for performing the method as described in any one of claims 1-9.

Images