An agent-based navigation content generation method and device, an agent, an electronic device, a readable storage medium and a computer program product

Abstract

The present disclosure provides an agent-based navigation content generation method, relates to the field of artificial intelligence, and particularly relates to the fields of large language models, agents, navigation and electronic maps. The agent includes a plurality of candidate navigation content generation components, each of the plurality of candidate navigation content generation components includes at least a function function for generating specific navigation content, the method includes: determining a target navigation content generation component from the plurality of candidate navigation content generation components based on user input information and the function description file; determining a point of interest search area based on the user input information and the state information of the user; determining the point of interest of the user based on the user input information and the point of interest search area by using a large model; and determining the navigation content of the user input information by using the agent to run the function function corresponding to the target navigation content generation component.

Description

Technical Field

[0001] This disclosure relates to the field of artificial intelligence, particularly to the fields of large language models, intelligent agents, navigation, and electronic maps, and specifically to a method, apparatus, intelligent agent, electronic device, readable storage medium, and computer program product for generating navigation content based on an intelligent agent. Background Technology

[0002] Artificial intelligence (AI) is the study of enabling computers to simulate certain human thought processes and intelligent behaviors (such as learning, reasoning, thinking, and planning). It encompasses both hardware and software technologies. AI hardware technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, and big data processing. AI software technologies mainly include computer vision, speech recognition, natural language processing, machine learning / deep learning, big data processing, and knowledge graph technologies.

[0003] Intelligent agents are the manifestation of artificial intelligence in specific task scenarios. Artificial intelligence provides intelligent agents with the technologies and methods to enable them to operate effectively. In recent years, with the development of large language model technology, intelligent agents based on large language models have gradually transformed from simple tools to more human-like partners, capable of deeply understanding human intentions and needs and proactively providing help and support. Currently, intelligent agents have become powerful assistants in human work and life.

[0004] When applying intelligent agents to map navigation services or point-of-interest (POI) recommendation scenarios, they often rely solely on fixed search radii or single text intent recognition, making it difficult to effectively integrate spatial context information such as the user's real-time motion state, client hardware acquisition parameters, and real-time traffic conditions. This results in insufficiently precise determination of POI search areas, and the intelligent agent struggles to efficiently and accurately match and schedule corresponding functional components and functions, thereby reducing the accuracy and timeliness of navigation content generation.

[0005] The methods described in this section are not necessarily methods that had been previously conceived or adopted. Unless otherwise specified, no method described in this section should be assumed to be prior art simply because it is included in this section. Similarly, unless otherwise specified, the issues mentioned in this section should not be considered to be accepted in any prior art. Summary of the Invention

[0006] This disclosure provides a method, apparatus, intelligent agent, electronic device, computer-readable storage medium, and computer program product for generating navigation content based on an intelligent agent.

[0007] According to one aspect of this disclosure, a navigation content generation method based on an intelligent agent is provided, wherein the intelligent agent includes a plurality of candidate navigation content generation components, and wherein each of the plurality of candidate navigation content generation components includes at least a functional description file of the candidate navigation content generation component, a function for generating specific navigation content, and a parameter description file of the function. The method includes: determining a target navigation content generation component from the plurality of candidate navigation content generation components based on user input information and the functional description file; determining a point of interest retrieval region based on the user input information and the user's state information, wherein the state information includes at least the user's current position and transient velocity; determining the user's points of interest using a large model based on the user input information and the point of interest retrieval region; and running the function corresponding to the target navigation content generation component using the intelligent agent based on the user input information, the points of interest, and the parameter description file to determine the navigation content of the user input information.

[0008] According to a second aspect of this disclosure, a navigation content generation apparatus based on an intelligent agent is provided, wherein the intelligent agent includes a plurality of candidate navigation content generation components, and wherein each of the plurality of candidate navigation content generation components includes at least a functional description file of the candidate navigation content generation component, a function for generating specific navigation content, and a parameter description file of the function. The apparatus includes: a target navigation content generation component determination module, configured to determine a target navigation content generation component from the plurality of candidate navigation content generation components based on user input information and the functional description file; an interest point retrieval region determination module, configured to determine an interest point retrieval region based on the user input information and the user's state information, wherein the state information includes at least the user's current position and transient velocity; an interest point determination module, configured to determine the user's interest points using a large model based on the user input information and the interest point retrieval region; and a navigation content determination module, configured to determine the navigation content of the user input information by running the function corresponding to the target navigation content generation component using the intelligent agent, based on the user input information, the interest points, and the parameter description file.

[0009] According to a third aspect of this disclosure, an intelligent agent is provided, comprising: a planner configured to determine a target navigation content generation component based on user input information and the functional description file; and an executor configured to generate navigation content according to the method described in any one of the preceding claims.

[0010] According to a fourth aspect of this disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the methods described above.

[0011] According to a fifth aspect of this disclosure, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, causes the processor to implement the method described above.

[0012] According to a sixth aspect of this disclosure, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the above-described method.

[0013] According to one or more embodiments of this disclosure, a target navigation content generation component is determined based on user input information, and an interest point retrieval region is determined based on state information including the user's current position and transient velocity. This allows the determination of the retrieval region to be combined with the user's real-time motion state, thereby using a large model to determine interest points and generating navigation content through an intelligent agent, improving the accuracy of interest point determination and thus enhancing the adaptability of navigation content.

[0014] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0015] The accompanying drawings exemplify embodiments and form part of the specification, serving together with the textual description to explain exemplary implementations of the embodiments. The illustrated embodiments are for illustrative purposes only and do not limit the scope of the claims. Throughout the drawings, the same reference numerals refer to similar but not necessarily identical elements.

[0016] Figure 1 A schematic diagram of an exemplary system in which the various methods described herein may be implemented according to embodiments of the present disclosure is shown; Figure 2 A flowchart of an agent-based navigation content generation method according to an embodiment of the present disclosure is shown; Figure 3 A flowchart illustrating agent-based navigation content generation according to an embodiment of the present disclosure is shown; Figure 4 A flowchart illustrating agent-based navigation content generation according to an embodiment of the present disclosure is shown; Figure 5A flowchart illustrating agent-based navigation content generation according to an embodiment of the present disclosure is shown; Figure 6 A block diagram of an agent-based navigation content generation apparatus according to an embodiment of the present disclosure is shown; Figure 7 A structural block diagram of an exemplary electronic device that can be used to implement embodiments of the present disclosure is shown. Detailed Implementation

[0017] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0018] In this disclosure, unless otherwise stated, the use of terms such as "first," "second," etc., to describe various elements is not intended to limit the positional, temporal, or importance relationships of these elements; such terms are merely used to distinguish one element from another. In some examples, the first element and the second element may refer to the same instance of that element, while in other cases, based on the context, they may refer to different instances.

[0019] The terminology used in the description of the various examples described in this disclosure is for the purpose of describing particular examples only and is not intended to be limiting. Unless the context explicitly indicates otherwise, an element may be one or more unless the number of elements is specifically limited. Furthermore, the term "and / or" as used in this disclosure covers any one of the listed items and all possible combinations thereof.

[0020] The embodiments of this disclosure will now be described in detail with reference to the accompanying drawings.

[0021] Figure 1 A schematic diagram of an exemplary system 100 in which the various methods and apparatus described herein can be implemented according to embodiments of this disclosure is shown. Reference Figure 1 The system 100 includes one or more client devices 101, 102, 103, 104, 105 and 106, a server 120, and one or more communication networks 110 coupling the one or more client devices to the server 120. The client devices 101, 102, 103, 104, 105 and 106 can be configured to execute one or more applications.

[0022] In embodiments of this disclosure, server 120 may run one or more services or software applications that enable the execution of content recommendation methods.

[0023] In some embodiments, server 120 may also provide other services or software applications, which may include non-virtual and virtual environments. In some embodiments, these services may be provided as web-based services or cloud services, such as to users of client devices 101, 102, 103, 104, 105 and / or 106 under a Software as a Service (SaaS) model.

[0024] exist Figure 1 In the configuration shown, server 120 may include one or more components that implement the functions performed by server 120. These components may include software components, hardware components, or combinations thereof that can be executed by one or more processors. Users operating client devices 101, 102, 103, 104, 105, and / or 106 can sequentially interact with server 120 using one or more client applications to utilize the services provided by these components. It should be understood that various different system configurations are possible and may differ from system 100. Therefore, Figure 1 This is an example of a system used to implement the various methods described herein, and is not intended to be limiting.

[0025] Users can use client devices 101, 102, 103, 104, 105, and / or 106 to recommend content. The client devices can provide an interface that allows users to interact with the client devices. The client devices can also output information to users through this interface. Although... Figure 1 Only six client devices are described, but those skilled in the art will understand that this disclosure can support any number of client devices.

[0026] Client devices 101, 102, 103, 104, 105, and / or 106 may include various types of computer devices, such as portable handheld devices, general-purpose computers (such as personal computers and laptops), workstation computers, wearable devices, smart screen devices, self-service terminal devices, service robots, gaming systems, thin clients, various messaging devices, sensors, or other sensing devices. These computer devices can run various types and versions of software applications and operating systems, such as Microsoft Windows, Apple iOS, UNIX-like operating systems, Linux or Linux-like operating systems (such as Google Chrome OS); or include various mobile operating systems, such as Microsoft Windows Mobile OS, iOS, Windows Phone, and Android. Portable handheld devices may include cellular phones, smartphones, tablets, personal digital assistants (PDAs), etc. Wearable devices may include head-mounted displays (such as smart glasses) and other devices. Gaming systems may include various handheld gaming devices, internet-enabled gaming devices, etc. Client devices are capable of executing various applications, such as various internet-related applications, communication applications (such as email applications), short message service (SMS) applications, and can use various communication protocols.

[0027] Network 110 can be any type of network well known to those skilled in the art, and can support data communication using any of a variety of available protocols (including but not limited to TCP / IP, SNA, IPX, etc.). By way of example only, one or more networks 110 can be a local area network (LAN), an Ethernet-based network, a token ring network, a wide area network (WAN), the Internet, a virtual network, a virtual private network (VPN), an intranet, an extranet, a blockchain network, a public switched telephone network (PSTN), an infrared network, a wireless network (e.g., Bluetooth, WIFI), and / or any combination of these and / or other networks.

[0028] Server 120 may include one or more general-purpose computers, special-purpose server computers (e.g., PC (personal computer) servers, UNIX servers, mid-range servers), blade servers, mainframe computers, server clusters, or any other suitable arrangement and / or combination. Server 120 may include one or more virtual machines running a virtual operating system, or other computing architectures involving virtualization (e.g., one or more flexible pools of logical storage devices that can be virtualized to maintain virtual storage devices for servers). In various embodiments, server 120 may run one or more services or software applications that provide the functionality described below.

[0029] The computing unit in server 120 can run one or more operating systems, including any of the aforementioned operating systems and any commercially available server operating system. Server 120 can also run any of a variety of additional server applications and / or middleware applications, including HTTP servers, FTP servers, CGI servers, JAVA servers, database servers, etc.

[0030] In some implementations, server 120 may include one or more applications to analyze and merge data feeds and / or event updates received from users of client devices 101, 102, 103, 104, 105 and / or 106. Server 120 may also include one or more applications to display data feeds and / or real-time events via one or more display devices of client devices 101, 102, 103, 104, 105 and / or 106.

[0031] In some implementations, server 120 can be a server for a distributed system or a server integrated with blockchain. Server 120 can also be a cloud server, or an intelligent cloud computing server or intelligent cloud host with artificial intelligence technology. A cloud server is a host product in the cloud computing service system, designed to address the shortcomings of traditional physical hosts and Virtual Private Server (VPS) services, such as high management difficulty and weak business scalability.

[0032] System 100 may also include one or more databases 130. In some embodiments, these databases may be used to store data and other information. For example, one or more of the databases 130 may be used to store information such as audio files and video files. Databases 130 may reside in various locations. For example, a database used by server 120 may be local to server 120, or it may be located away from server 120 and may communicate with server 120 via a network-based or dedicated connection. Databases 130 may be of different types. In some embodiments, the database used by server 120 may be, for example, a relational database. One or more of these databases may store, update, and retrieve data from and from the databases in response to commands.

[0033] In some embodiments, one or more of the databases 130 may also be used by an application to store application data. The databases used by the application may be of different types, such as key-value stores, object stores, or regular stores supported by a file system.

[0034] Figure 1The system 100 can be configured and operated in various ways to enable the application of the various methods and apparatus described in this disclosure.

[0035] Figure 2 This is a flowchart illustrating an agent-based navigation content generation method 200 according to an embodiment of the present disclosure. The agent includes a plurality of candidate navigation content generation components, and each candidate navigation content generation component includes at least a functional description file for that component, a function for generating specific navigation content, and a parameter description file for the function, such as... Figure 2 As shown, the method includes: step S202, determining a target navigation content generation component from the plurality of candidate navigation content generation components based on user input information and the function description file; step S204, determining an interest point retrieval region based on the user input information and the user's state information, wherein the state information includes at least the user's current position and transient velocity; step S206, determining the user's interest points using a large model based on the user input information and the interest point retrieval region; and step S208, using the agent to run the function corresponding to the target navigation content generation component based on the user input information, the interest points, and the parameter description file, to determine the navigation content of the user input information.

[0036] In this embodiment, the target navigation content generation component is determined based on user input information, and the interest point retrieval region is determined based on state information including the user's current position and transient velocity. This allows the determination of the retrieval region to be combined with the user's real-time motion state. In turn, a large model is used to determine the interest points and generate navigation content through an intelligent agent, thereby improving the accuracy of interest point determination and enhancing the adaptability of the navigation content.

[0037] In some embodiments, the functional description file is used to characterize the specific capabilities, applicable scenarios, and execution logic of the corresponding candidate navigation content generation component in natural language or a preset structured format. Specifically, the functional description file may include one or more of the following: component identifier, functional summary text, core intent label, and applicable boundary description. The functional description file may be encapsulated in a structured data format to facilitate the extraction of text features and rapid calculation of semantic similarity with user input information.

[0038] In some embodiments, in step S202, determining the target navigation content generation component from the plurality of candidate navigation content generation components based on user input information and the function description file may include: converting the user input information into a first feature vector using a pre-trained text vectorization model; obtaining the function description files of the plurality of candidate navigation content generation components respectively, and converting them into corresponding second feature vectors using the text vectorization model; calculating the similarity between the first feature vector and each of the second feature vectors; and determining the candidate navigation content generation component as the target navigation content generation component in response to determining that the similarity corresponding to a certain candidate navigation content generation component meets a preset condition, such as the highest similarity or greater than a preset threshold.

[0039] Figure 3 This is a flowchart illustrating an agent-based navigation content generation method 300 according to embodiments of the present disclosure. In some embodiments, such as... Figure 3 As shown, determining the point of interest retrieval area based on the user input information and the user's status information includes: step S302, determining a first retrieval distance based on the user input information; step S304, in response to determining that the transient speed is less than or equal to a first preset value, taking the first retrieval distance as the target retrieval distance; step S306, in response to determining that the transient speed is greater than the first preset value, determining a compensation distance based on the transient speed, and taking the sum of the first retrieval distance and the compensation distance as the target retrieval distance; and step S308, taking the current position as the center point, determining the point of interest retrieval area based on the target retrieval distance.

[0040] In this embodiment, the target retrieval distance is dynamically adjusted based on the relationship between the user's transient speed and the first preset value, thereby matching the point of interest retrieval area with the user's real-time movement state and improving the accuracy of point of interest retrieval.

[0041] In some embodiments, for step S302, the first retrieval distance can be determined by parsing the spatial intent in the user input information. For example, when the user input information contains distance description words such as "nearby", "within one kilometer" or "not far away", these words can be extracted using natural language processing technology and converted into specific physical distance values ​​(e.g., 500 meters) using preset mapping rules, and used as the first retrieval distance.

[0042] In some embodiments, for the spatial distance compensation mechanism targeting user dynamic motion characteristics, the first preset value may be a pre-set speed threshold used to distinguish between a user in a low-speed / stationary state (e.g., standing still, walking slowly) and a high-speed moving state (e.g., cycling, driving). When the transient speed is less than or equal to the first preset value, it indicates that the user's position in physical space is relatively stable, and in this case, the first retrieval distance is directly used as the final target retrieval distance.

[0043] Conversely, when the transient speed is greater than the first preset value, if the user moves faster and the search is performed only according to the original first search distance, the user may have already left or missed the corresponding point of interest by the time the intelligent agent generates and broadcasts the navigation content.

[0044] In some embodiments, the transient speed can be multiplied by a preset coefficient (e.g., response delay time or the user's expected braking / lane change preparation time) to calculate the compensation distance. Then, the first retrieval distance is added to the compensation distance to obtain the dynamically expanded target retrieval distance, thereby providing forward-looking compensation for the spatial retrieval range in high-speed moving scenarios. For example, the target retrieval distance can be calculated using the following formula (1). : (1) in, This is the preset basic search radius; The effective forward look time window parameters are the preset coefficients. The user's current movement speed. The effective forward look-ahead time window parameter. Used to characterize the length of time the system needs to cover in advance. In this embodiment, the effective look-ahead time window parameter... The calculation can be performed using the following formula (2): (2) in, The refresh cycle of the navigation system reflects how often the navigation data is updated. The computation time for the service link of the navigation system; Network transmission and service queuing time for navigation systems, used to reflect jitter in the end-to-end link; To provide a safety margin, it is used to resist speed fluctuations, positioning drift, and system jitter.

[0045] In another possible embodiment, the map navigation scenario or task type can be adjusted accordingly. Implement differentiated adaptation designs. For example, for continuous video frame tracking and close-range recognition scenarios. The refresh rate is 3-8 seconds, emphasizing high refresh rate and focusing on the user's immediate target; this is suitable for typical pedestrian visual point-finding scenarios. The time interval is 8-20 seconds, used to cover one capture, retrieval, and next refresh cycle; this is suitable for POI sparse region retrieval scenarios. For scenarios requiring 20-40 seconds, prioritize improving candidate recall; while in scenarios involving early target acquisition or route planning... The duration should be 30-60 seconds to meet the need for greater forward coverage. For visual point finding and visual question answering tasks in pedestrian navigation, the time can be increased to 30-60 seconds. The initial value is set within the range of 8-20 seconds, and then calibrated using real navigation trajectories and user behavior data.

[0046] In some embodiments, for step S308, after obtaining the target retrieval distance, a continuous geographic coverage area (e.g., a circular area, a rectangular area, or an effective area generated by combining the actual road network topology) can be generated in the geographic coordinate system, using the coordinates representing the current location (e.g., latitude and longitude) as the geometric center point and the target retrieval distance as the spatial retrieval radius or boundary threshold. This coverage area is then determined as the point of interest retrieval area. Through the above dynamic distance adjustment strategy, it can be ensured that regardless of the user's travel status, the candidate points of interest relied upon by the intelligent agent when generating navigation content are all located within a reasonable spatial range that conforms to the user's current physical reach and safe operating procedures, significantly improving navigation accuracy in dynamic navigation scenarios.

[0047] In some embodiments, the method includes: in response to determining that the user input information includes either image information or video information acquired in real time by an image acquisition device based on a user client device, obtaining parameter information of the image acquisition device, wherein the parameter information includes at least the maximum identifiable distance of the image acquisition device, and wherein determining the first retrieval distance based on the user input information includes: using the maximum identifiable distance as the first retrieval distance.

[0048] In this embodiment, when the user inputs image or video information acquired in real time, the maximum identifiable distance of the image acquisition device is obtained and used as the first retrieval distance. This ensures that the initial retrieval distance matches the actual recognition capability of the image acquisition device, ensuring that the retrieval range of points of interest is within the effective recognition range of the device. This avoids the disconnect between the retrieval distance and the image acquisition capability, thereby improving the rationality and accuracy of determining the retrieval area based on visual input.

[0049] In some embodiments, obtaining parameter information of the image acquisition device can be achieved by calling the underlying hardware interface of the user client device, reading the device driver configuration file, or parsing the metadata (e.g., EXIF ​​information) of the real-time acquired images and videos. The maximum identifiable distance is typically limited by the physical optical characteristics or sensing component capabilities of the image acquisition device, such as the effective focal length of the camera, the resolution of the photosensitive element, or the limiting range of the depth sensor (e.g., a LiDAR lens).

[0050] Taking a specific navigation scenario as an example, when a user initiates navigation by taking a picture of the street scene ahead using an augmented reality (AR) wearable device or mobile terminal, if the underlying interface returns that the maximum recognizable distance of the image acquisition device is 50 meters, then 50 meters is directly determined as the first retrieval distance. Through this mechanism of deep hardware and software linkage, when using a large model combined with visual images to determine target points of interest, the underlying retrieval range is strictly and physically constrained within the user's actual line-of-sight boundary. This not only effectively filters out redundant points of interest beyond the field of view, reducing the recall overhead of the spatial database, but also alleviates the computational burden on the large model when processing multimodal data, further ensuring the real-time generation of navigation content based on intelligent agents.

[0051] It should be noted that the image or video information included in the user input information is not captured in real time by the image acquisition device of the user client device (for example, the image or video is read from a local media library, shared by a third-party social application, or downloaded from the network), and no longer directly relies on the physical hardware parameters of the current client device to set the first retrieval distance.

[0052] Specifically, when processing such non-real-time acquired visual content, the original metadata embedded in the image or video file can be parsed first (e.g., the original shooting focal length, depth markers, or estimated shooting distance recorded in EXIF ​​information). If parsing is successful, it is converted into the first retrieval distance. If the media file does not contain usable metadata or the metadata has been erased, a semantic understanding-based auxiliary ranging strategy will be switched to. That is, the intelligent agent or underlying large model focuses on analyzing the natural language instructions input synchronously with the non-real-time image or video, and combines the intent type and preset mapping rules described in the foregoing embodiments to re-infer and determine the first retrieval distance.

[0053] In some embodiments, the parameter information further includes the field of view of the image acquisition device, which is the maximum spatial angle range within which the image acquisition device can receive optical signals and form an image. Furthermore, determining the point of interest retrieval region based on the target retrieval distance and using the current position as the center point includes: determining the point of interest retrieval region based on the target retrieval distance and the field of view, using the current position as the center point.

[0054] In this embodiment, by further combining the field of view of the image acquisition device when determining the point of interest retrieval area, the range of the retrieval area is limited not only by distance but also by angle range, thereby making the point of interest retrieval area more closely match the actual imaging space range of the image acquisition device and improving the accuracy of point of interest retrieval.

[0055] In some embodiments, after obtaining the field of view of the image acquisition device, a fan-shaped region or a three-dimensional frustum space can be defined in a geographic coordinate system or a spatial coordinate system, using the current position as the geometric vertex, the target retrieval distance as the radial boundary, and the field of view as the subtended angle. This fan-shaped region or three-dimensional frustum space is then determined as the point of interest retrieval region. For example, when the client device is a smart wearable AR glasses with a 120-degree horizontal field of view and the determined target retrieval distance is 50 meters, the constructed point of interest retrieval region is a fan-shaped coverage area with the current device position as the vertex, facing the center of the screen, with an angle of 120 degrees and a radius of 50 meters.

[0056] By introducing a field of view for spatial geometric constraints, irrelevant geographic entities located in imaging blind spots (such as outside the edge of the image or directly behind the user) can be effectively eliminated during the recall phase. This processing logic, which maps physical optical parameters to digital retrieval boundaries, not only reduces the amount of redundant data input to large models for multimodal inference from the source, significantly reducing underlying computational overhead and memory usage, but also ensures that the pool of points of interest that the agent focuses on when generating navigation content maintains spatial correspondence and synchronization with the user's current field of view.

[0057] In some embodiments, the method further includes: acquiring spatial pose information when the image acquisition device acquires the image information or the video information, wherein the spatial pose information includes at least orientation information when the image acquisition device acquires the image information or the video information, and wherein determining the point of interest retrieval region based on the target retrieval distance and the field of view, with the current position as the center point, includes: determining the point of interest retrieval region based on the target retrieval distance, the field of view, and the orientation information, with the current position as the center point.

[0058] In this embodiment, by further combining the orientation information of the image acquisition device when determining the point of interest retrieval area, the direction of the retrieval area is made consistent with the actual orientation of the user when taking the picture, thereby limiting the retrieval range to the direction that the user is actually interested in, effectively eliminating irrelevant interference in non-orientation areas, and further improving the accuracy of point of interest retrieval and the matching with user intent.

[0059] In some embodiments, the spatial attitude information can be obtained by accessing data from sensing modules such as an inertial measurement unit, gyroscope, gravity sensor, or electronic compass built into the user client device. The orientation information obtained thereby can accurately characterize the three-dimensional spatial pose of the image acquisition device when acquiring images or videos, such as its pitch angle, yaw angle, or roll angle in physical space, as well as the specific deflection angle of its lens optical axis relative to geographic true north.

[0060] When constructing the spatial retrieval boundary, the current position can be used as the geometric vertex, and the physical spatial direction indicated by the orientation information can be used as the central axis (or the symmetrical bisector of the angle) of the retrieval area. Combining the aforementioned field of view angle as the spatial angle and the target retrieval distance as the radial boundary, a sector or three-dimensional frustum coverage area with clear directionality is delineated in the geographic coordinate system, and this area is determined as the point of interest retrieval area.

[0061] For example, if a user holds a mobile device and takes a picture of commercial buildings in front of them, pointing due east, with a horizontal field of view of 90 degrees, the determined point of interest retrieval area will be a fan-shaped space extending 45 degrees northeast and southeast from the user's current location as the vertex, due east as the central axis, and eastward and southeastward respectively. This processing mechanism, which deeply integrates the 3D pose data of hardware sensors with geospatial retrieval, ensures that the candidate points of interest processed by the large model and the agent are consistent with the real physical horizon captured by the camera. Based on this, invalid landmark data located behind or to the side of the user is avoided at the physical level, reducing the computational redundancy of the multimodal large model when processing complex street scenes and navigation intentions, and improving the response speed of content generation.

[0062] In some embodiments, the method includes: in response to determining that the image information exists in the user input information, determining the user's points of interest using a large model based on the user input information and the point of interest retrieval region includes: determining the user's points of interest using the large model based at least on the image information and the point of interest retrieval region.

[0063] In this embodiment, when the user input includes image information, the image information and the interest point retrieval area are input together into the large model to determine the interest point, thereby realizing the fusion analysis of visual content and spatial range information and improving the accuracy of interest point recognition.

[0064] In some embodiments, the large model is a multimodal large language model with visual understanding capabilities, which can jointly use the image information and the geographic boundary data representing the point of interest retrieval region as input to the large model. On the one hand, the large model can perform deep visual understanding of the image information, extracting visual semantic features such as the outlines of landmark buildings, shop sign text (OCR recognition), brand logos, or the decoration style of specific stores. On the other hand, the large model uses the point of interest retrieval region as a hard constraint condition in geospatial terms.

[0065] For example, in real-world pedestrian navigation scenarios, when a user takes an image showing the exterior of a coffee shop but not its name, a large model might match hundreds or thousands of similar-looking coffee shops in its global database based solely on the image. However, using the method in this embodiment, the large model first extracts visual features from the image, such as "dark green awnings and retro brick facades." Then, it cross-references known physical entities within a defined area of ​​interest (e.g., a fan-shaped area within 50 meters directly in front of the user) defined by the aforementioned steps. Due to the strict constraints of physical spatial boundaries, the large model effectively filters out interference from similar shops outside this area, accurately matching the target store the user is pointing to and identifying it as the target point of interest. This processing mechanism, which deeply couples multimodal visual features with physical spatial boundaries, reduces the risk of spatial illusion in the large model and significantly enhances the uniqueness and reliability of point-of-interest identification in complex street scenes.

[0066] In some embodiments, the method includes: in response to determining that the video information exists in the user input information, determining keyframe image information of the video information based on the large model, wherein determining the user's points of interest based on the user input information and the point of interest retrieval region using the large model includes: determining the user's points of interest based at least on the keyframe image information and the point of interest retrieval region using the large model.

[0067] In this embodiment, for cases where user input includes video information, keyframe image information in the video information is determined by using a large model, and points of interest are determined based on the keyframe image information and the point of interest retrieval region. This allows for the selection of representative image content in the video, avoiding redundant processing of the full video frame data and reducing the data processing load. At the same time, the combination of spatial retrieval region ensures the accuracy of point of interest determination.

[0068] In some embodiments, when processing the video information, the large model is a multimodal video large model with temporal visual understanding capabilities. Specifically, the large model can comprehensively analyze the video stream based on the spatiotemporal dynamic features in the video information and the user input information (such as voice commands input synchronously with video recording), identify visually significant changes in the video frame, camera focusing actions (such as zooming in or pausing on an object for a long time), the sharpness of the main subject, or the alignment relationship between the video frame and the user's voice intent on the timeline, thereby automatically extracting single or multiple frames containing a clear directional intent or the clearest representation of the target landmark from the continuous video stream, and using them as the keyframe image information.

[0069] Taking driving or cycling scenarios as an example, when a user records a dynamic video of a street scene using a vehicle camera or smart helmet and simultaneously issues a voice request, "Navigate me to the fast food restaurant with the red sign ahead," directly processing all video frames is not only extremely time-consuming due to the device being in motion, but also prone to recognition errors due to image jitter. The method in this embodiment allows the large model to perceive the moment the voice command occurs and, combined with visual analysis of the video stream, accurately extract a frame with a clear sign and no motion blur as the keyframe image information. Subsequently, this high-quality keyframe image information, along with the point of interest retrieval region defined by the user's current location and transient speed, is input together. This mechanism combines temporal feature dimensionality reduction (keyframe extraction) with retrieval region constraints, enabling the large model to concentrate all its computing power on the spatial comparison of high-value visual data, quickly locating the target fast food restaurant in a short time, thus improving the efficiency of intent parsing and navigation response speed in dynamic continuous visual input scenarios.

[0070] Figure 4 This is a flowchart illustrating an agent-based navigation content generation method 400 according to embodiments of the present disclosure. In some embodiments, such as... Figure 4 As shown, the method further includes: step S402, determining the user's displacement information during the video recording based on the non-keyframe image information in the video information; and step S404, updating the point of interest retrieval area in real time based on the displacement information.

[0071] In this embodiment, the displacement information of the user during shooting is determined based on the non-keyframe image information in the video information, and the point of interest retrieval area is updated in real time based on the displacement information. This allows the retrieval area to be dynamically adjusted as the user moves during the shooting process, ensuring that the retrieval range always matches the user's actual shooting position, thereby improving the accuracy and real-time performance of point of interest retrieval in mobile shooting scenarios.

[0072] In some embodiments, for step S402, since the non-keyframe image information constitutes a continuous time series in the video information, it contains rich clues to background motion changes. In this embodiment, computer vision tracking algorithms such as visual odometry or real-time localization and mapping can be used to extract feature points and perform inter-frame matching on the continuous non-keyframe image information. By calculating the pixel-level positional offset of fixed reference objects (e.g., road markings, static building outlines, trees) in the background environment between adjacent non-keyframes, and combining this with the intrinsic and extrinsic parameter matrices of the image acquisition device, the motion trajectory vector of the image acquisition device (i.e., the user) in the real three-dimensional physical space can be calculated in reverse, thereby obtaining extremely accurate displacement information.

[0073] Furthermore, regarding step S404, in complex mobile recording scenarios (e.g., a user walking or cycling while holding a device and recording a video lasting several seconds while looking around), the user's physical position at the end of video recording or when a keyframe is extracted has typically shifted significantly compared to the recording start point. If the initial position coordinates are still used as the spatial reference, it will cause lag errors in the spatial coordinate system when the large model analyzes the keyframes. Therefore, the displacement information (including the direction and magnitude of the displacement vector) can be used to perform geometric translation or dynamic boundary correction on the pre-determined point of interest retrieval region. For example, the geometric origin of the original point of interest retrieval region can be translated along the direction of the displacement vector by a real physical distance proportional to the motion trajectory to generate a real-time updated point of interest retrieval region.

[0074] This closed-loop spatial correction mechanism based on continuous visual flow motion estimation does not rely on external sensor signals. Instead, it uses non-keyframe data to compensate for the positioning drift, refresh delay, or signal blockage problems that easily occur when relying solely on traditional methods such as the Global Positioning System (GPS) in short-distance or high-frequency movement. This ensures that the agent can accurately anchor the multimodal visual features in the keyframes to the user's current real-time physical spacetime range.

[0075] In some embodiments, the method includes: in response to determining that the user input information does not include image information and video information acquired in real time by an image acquisition device based on a user client device, determining the user's intent type based on the user input information using the large model, wherein determining the first retrieval distance based on the user input information includes: determining the first retrieval distance based on the intent type using a preset mapping rule, wherein the mapping rule is used to describe the correspondence between the intent type and the first retrieval distance.

[0076] In this embodiment, when the user input information does not contain image or video information, the user's intent type is determined by using a large model, and the first retrieval distance is determined based on a preset mapping rule between the intent type and the first retrieval distance. This allows the determination of the first retrieval distance to adapt to the user's specific semantic intent, realizing intelligent setting of the retrieval distance in scenarios without visual input, and improving the rationality and accuracy of the determination of the point of interest retrieval region.

[0077] In some embodiments, when a user interacts solely through plain text or voice, the user input information is sent to the large model for deep natural language understanding and intent slot extraction. The large model can combine contextual semantics, interjections, and even input time to perform refined classification of the user's intent type. For example, the intent types may include, but are not limited to: emergency convenience (e.g., finding a pharmacy or emergency room), leisure and consumption (e.g., finding a coffee shop, a popular restaurant, or a movie theater), and long-distance planning (e.g., finding a gas station, charging station, or highway service area).

[0078] Furthermore, the preset mapping rules can be represented as a configuration table or a piece of mapping logic code. The core of this approach is to assign different spatial scales based on the expected physical accessibility and time sensitivity behind different intent types. For example, if the large model determines the user's intent type to be emergency convenience, since such needs often require extremely high walking accessibility and immediacy, the mapping rules can set the first retrieval distance to a shorter value (e.g., 500 meters). If the intent type is determined to be leisure consumption, the first retrieval distance can be set to a medium value (e.g., 3 kilometers). When the intent type is long-distance planning, it means the user usually has transportation and a high tolerance for distance, so the first retrieval distance can be mapped to a larger value (e.g., 15 kilometers).

[0079] This flexible distance mapping mechanism, based on large-model semantic parsing, effectively compensates for the spatial scale blind spots caused by the lack of prior camera visual parameters (such as maximum recognizable distance or field of view). By deeply mining the urgency and life experience logic hidden in users' natural language expressions, it can avoid the extreme problems caused by using a globally fixed search radius, such as recalling too many useless remote POIs (Points of Interest) for urgent needs or empty results for long-distance needs.

[0080] Figure 5 This is a flowchart illustrating an agent-based navigation content generation method 500 according to embodiments of the present disclosure. In some embodiments, such as... Figure 5As shown, the method further includes: step S502, obtaining traffic information of the current location; and step S504, adjusting the center point of the point of interest retrieval area within a preset distance range of the current location based on the traffic information.

[0081] In this embodiment, by obtaining the road condition information of the current location and adjusting the center point of the point of interest retrieval area within a preset distance range based on the road condition information, the center point of the retrieval area can be made to fit the actual road distribution, thereby improving the rationality and accuracy of the determination of the point of interest retrieval area.

[0082] In some embodiments, for step S502, obtaining the traffic information can be achieved by calling the real-time traffic data interface of a third-party map service, receiving broadcast beacons from intelligent transportation base stations, or parsing vehicle-to-everything (V2X) cloud data. The traffic information may include, but is not limited to: real-time congestion index of the surrounding road network, average traffic speed, road closure status, one-way street direction, and traffic control events.

[0083] In step S504, considering that absolute physical distance in geographic space is not equivalent to actual traffic accessibility cost, strictly using the user's current absolute physical coordinates as the center point for retrieval may result in a large number of points of interest located in severely congested road sections, one-way streets, or closed areas, making the navigation content generated by the agent infeasible in actual execution. Therefore, by combining the traffic information, a virtual anchor point with optimal traffic efficiency can be calculated within the preset distance range (e.g., a maximum tolerance radius of 200 meters around the current location), and this virtual anchor point can replace the original coordinates as the center point of the adjusted point of interest retrieval area.

[0084] For example, when the traffic information indicates that a traffic accident has occurred directly ahead of the user's current road, causing severe congestion and the queue length exceeds a certain threshold, a position offset vector can be calculated pointing to the nearest nearby, smoothly passing branch road or parallel auxiliary road. Using this offset vector, the center point of the point of interest retrieval area is shifted from the core of the congestion to the smooth, adjacent road.

[0085] This centerpoint offset mechanism, based on real-time dynamic traffic conditions, transforms the static geographic geometry into a traffic topology space with time weights. This ensures that the generated navigation points of interest physically match distance expectations and guarantees their fastest accessibility in the traffic dimension, enhancing the actual service efficiency of navigation agents in complex urban road network environments.

[0086] In some embodiments, in response to determining that the target navigation content generation component includes multiple function functions for completing a specific task, the function description file further includes descriptive information about the dependencies of the multiple function functions, and wherein, the step of using the agent to run the function functions based on the user input information, the points of interest, and the parameter description file to determine the navigation content of the user input information includes: using the agent to run the multiple function functions based on the user input information, the points of interest, the parameter description file, and the dependencies to determine the navigation content of the user input information.

[0087] In this embodiment, for complex tasks that require multiple functional functions to work together, by clarifying the dependencies between functional functions, the agent can schedule and execute multiple functional functions in an orderly manner according to the dependencies, ensuring the correctness of data transmission and logical coherence between functions, thereby improving the success rate of processing complex user intentions and the accuracy of navigation content generation.

[0088] In some embodiments, the description information of the dependencies may include the execution topology between the various functional functions, pre- and post-constraints, and mapping rules for input / output parameters. Specifically, the orchestration module within the intelligent agent can parse the description information and construct the multiple functional functions into a directed acyclic graph (DAG) or a multi-step execution chain. In this execution chain, the output results of the pre-functions (e.g., JSON-formatted return data) are automatically extracted by the agent and mapped to fill the parameter slots defined in the parameter description file for the post-functions, thereby achieving seamless connection of the underlying data.

[0089] For example, when a user's intention is to navigate to the shopping mall ahead and check for available parking spaces with charging stations, the determined target navigation content generation component may include three functions that need to coordinate: a first function "POI basic information query function," a second function "parking lot real-time status query function," and a third function "end-point route planning function." Based on the description of the dependencies, the agent first runs the first function, combining it with the points of interest determined in previous steps to obtain the unique entity identifier of the shopping mall; subsequently, the agent automatically passes this entity identifier as an input parameter to the second function to query the availability of charging stations and the specific coordinates of the underground entrance in the shopping mall's associated parking lot; finally, the agent passes the underground entrance coordinates as a target point parameter to the third function, thereby generating the final navigation content containing precise parking guidance.

[0090] Through this dependency-based chained invocation and automated parameter orchestration mechanism, the intelligent agent breaks through the capability bottleneck of a single tool interface. It can not only handle complex combined intents involving cross-business domains (such as map retrieval, vehicle network query, and route calculation), but also effectively avoid function call crashes caused by missing parameters or reversed execution order, giving navigation applications based on large models the ability to handle deep logical reasoning and complex tasks.

[0091] It should be noted that the threshold values ​​involved in the technical solution disclosed herein can be adaptively adjusted according to the actual application scenario.

[0092] The collection, storage, use, processing, transmission, provision, and disclosure of any type of information, such as user personal information, in this technical solution comply with relevant laws and regulations and do not violate public order and good morals.

[0093] Figure 6 This is a block diagram illustrating an agent-based navigation content generation apparatus 600 according to an embodiment of the present disclosure, wherein the agent includes a plurality of candidate navigation content generation components, and wherein each of the plurality of candidate navigation content generation components includes at least a functional description file of the candidate navigation content generation component, a function for generating specific navigation content, and a parameter description file of the function. The apparatus includes: a target navigation content generation component determination module 602, configured to determine a target navigation content generation component from the plurality of candidate navigation content generation components based on user input information and the functional description file; The point of interest retrieval region determination module 604 is configured to determine the point of interest retrieval region based on the user input information and the user's state information, wherein the state information includes at least the user's current position and transient velocity; the point of interest determination module 606 is configured to determine the user's points of interest using a large model based on the user input information and the point of interest retrieval region; and the navigation content determination module 608 is configured to determine the navigation content of the user input information by using the intelligent agent to run the function corresponding to the target navigation content generation component based on the user input information, the points of interest, and the parameter description file.

[0094] It should be noted that, Figure 6 The various modules of the device 600 shown can be connected to the reference. Figure 2 The steps in method 200 described correspond to each other. Therefore, the operations, features, and advantages described above for method 200 also apply to apparatus 600 and its included modules and units. For the sake of brevity, some operations, features, and advantages will not be repeated here.

[0095] According to embodiments of this disclosure, an intelligent agent is provided, comprising: a planner configured to determine a target navigation content generation component based on user input information and the functional description file; and an executor configured to generate navigation content according to the method described in any one of the preceding embodiments.

[0096] In some embodiments, the planner essentially acts as the central brain of the intelligent agent. Its core relies on a large language model, focusing on deep semantic understanding of unstructured natural language requests, intent feature parsing, and dynamic distribution and routing of global tasks. Specifically, upon receiving user input and optional multimodal context (such as image features or video keyframes), the planner combines the user's state information to perform multi-step logical decomposition of the user's complex, non-standard initial request. Subsequently, by comparing the decomposed sub-task intents with the functional description files of the multiple candidate navigation content generation components, the planner autonomously determines the most suitable target navigation content generation component and, based on the dependencies between multiple functions within the component, orchestrates and generates a global call plan including execution order, logical branches, and parameter passing paths.

[0097] In this embodiment, the executor acts as the underlying driving engine for the implementation of agent actions and data processing. It strictly follows the decision results issued by the planner and is responsible for specific underlying data retrieval and rule calculations. Specifically, based on the global call plan issued by the planner and the determined target navigation content generation component, the executor performs substantive spatial retrieval engine calls and cross-module data flow within the constrained point of interest retrieval area. In specific operation, the executor strictly adheres to the parameter description file and dependencies, accurately extracting the current environmental context (such as road condition information, field of view parameters, displacement vectors, etc.), formatting it, and injecting it into the parameter slots of the corresponding functional functions. Subsequently, the executor actually interfaces with the underlying geographic information database, hardware sensor interface, or third-party network service API, capturing the return results of each functional function, and finally assembling and converting these discrete spatial calculations and road network topology data into logically coherent navigation content that can be directly presented or broadcast, thus completing the entire closed-loop process from high-level abstract decision-making of the large model to the accurate implementation of physical data.

[0098] According to embodiments of this disclosure, an electronic device, a readable storage medium, and a computer program product are also provided.

[0099] refer to Figure 7The present invention describes a structural block diagram of an electronic device 700 that can serve as a server or client of the present disclosure, which is an example of a hardware device that can be applied to various aspects of the present disclosure. The electronic device is intended to represent various forms of digital electronic computer devices, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0100] like Figure 7 As shown, the electronic device 700 includes a computing unit 701, which can perform various appropriate actions and processes based on a computer program stored in a read-only memory (ROM) 702 or a computer program loaded from a storage unit 708 into a random access memory (RAM) 703. The RAM 703 may also store various programs and data required for the operation of the electronic device 700. The computing unit 701, ROM 702, and RAM 703 are interconnected via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0101] Multiple components in electronic device 700 are connected to I / O interface 705, including: input unit 706, output unit 707, storage unit 708, and communication unit 709. Input unit 706 can be any type of device capable of inputting information to electronic device 700. Input unit 706 can receive input digital or character information and generate key signal inputs related to user settings and / or function control of electronic device, and may include, but is not limited to, a mouse, keyboard, touchscreen, trackpad, trackball, joystick, microphone, and / or remote control. Output unit 707 can be any type of device capable of presenting information, and may include, but is not limited to, a monitor, speaker, video / audio output terminal, vibrator, and / or printer. Storage unit 708 may include, but is not limited to, disk and optical disk. Communication unit 709 allows electronic device 700 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks, and may include, but is not limited to, modems, network cards, infrared communication devices, wireless communication transceivers, and / or chipsets, such as Bluetooth devices, 802.11 devices, WiFi devices, WiMax devices, cellular communication devices, and / or the like.

[0102] The computing unit 701 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 701 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 701 performs the various methods and processes described above, such as method 200. For example, in some embodiments, method 200 may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 708. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 700 via ROM 702 and / or communication unit 709. When the computer program is loaded into RAM 703 and executed by the computing unit 701, one or more steps of method 200 described above may be performed. Alternatively, in other embodiments, the computing unit 701 may be configured to perform method 200 by any other suitable means (e.g., by means of firmware).

[0103] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0104] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0105] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0106] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0107] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), the Internet, and blockchain networks.

[0108] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0109] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0110] While embodiments or examples of this disclosure have been described with reference to the accompanying drawings, it should be understood that the methods, systems, and devices described above are merely exemplary embodiments or examples, and the scope of the invention is not limited by these embodiments or examples, but only by the granted claims and their equivalents. Various elements in the embodiments or examples may be omitted or replaced by their equivalents. Furthermore, the steps may be performed in a different order than that described in this disclosure. Further, various elements in the embodiments or examples may be combined in various ways. Importantly, as the technology evolves, many elements described herein can be replaced by equivalents that appear after this disclosure.

Claims

1. A method for generating navigation content based on intelligent agents, wherein, The intelligent agent includes multiple candidate navigation content generation components, and each candidate navigation content generation component includes at least a functional description file for that candidate navigation content generation component, a function for generating specific navigation content, and a parameter description file for the function. The method includes: Based on user input information and the function description file, a target navigation content generation component is determined from the plurality of candidate navigation content generation components; Based on the user input information and the user's status information, an interest point retrieval area is determined, wherein the status information includes at least the user's current location and transient velocity; Based on the user input information and the interest point retrieval region, the user's interest points are determined using a large model; and Based on the user input information, the points of interest, and the parameter description file, the agent runs the function corresponding to the target navigation content generation component to determine the navigation content of the user input information.

2. The method as described in claim 1, wherein, The process of determining the point of interest retrieval region based on the user input information and the user's status information includes: Determine the first search distance based on user input information; In response to determining that the transient velocity is less than or equal to a first preset value, the first retrieval distance is taken as the target retrieval distance; In response to determining that the transient velocity is greater than a first preset value, a compensation distance is determined based on the transient velocity, and the sum of the first retrieval distance and the compensation distance is used as the target retrieval distance; and Using the current location as the center point, the point of interest retrieval area is determined based on the target retrieval distance.

3. The method as described in claim 2, wherein, The method includes: In response to determining that the user input information includes either image information or video information acquired in real time by an image acquisition device based on a user client device, parameter information of the image acquisition device is obtained, wherein the parameter information includes at least the maximum recognizable distance of the image acquisition device, and wherein determining the first retrieval distance based on the user input information includes: The maximum identifiable distance is taken as the first retrieval distance.

4. The method of claim 3, wherein, The parameter information also includes the field of view of the image acquisition device, which is the maximum spatial angle range within which the image acquisition device can receive optical signals and form an image. Furthermore, determining the point of interest retrieval region based on the target retrieval distance and using the current position as the center point includes: Using the current location as the center point, the point of interest retrieval area is determined based on the target retrieval distance and the field of view.

5. The method of claim 4, wherein, The method further includes: Acquire spatial pose information when the image acquisition device acquires the image information or the video information, wherein the spatial pose information includes at least the orientation information when the image acquisition device acquires the image information or the video information, and wherein determining the point of interest retrieval region based on the target retrieval distance and the field of view, with the current position as the center point, includes: Using the current location as the center point, the point of interest retrieval area is determined based on the target retrieval distance, the field of view, and the orientation information.

6. The method according to any one of claims 3-5, wherein, The method includes: In response to determining that the image information exists in the user input information, the step of determining the user's interest points based on the user input information and the interest point retrieval region using a large model includes: Based at least on the image information and the point of interest retrieval region, the user's points of interest are determined using the large model.

7. The method according to any one of claims 3-5, wherein, The method includes: In response to determining that the video information exists in the user input information, keyframe image information of the video information is determined based on the large model, wherein determining the user's points of interest using the large model based on the user input information and the point of interest retrieval region includes: Based at least on the keyframe image information and the interest point retrieval region, the user's interest points are determined using the large model.

8. The method of claim 7, wherein, The method further includes: Based on the non-keyframe image information in the video information, determine the user's displacement information during the video recording period; and The point of interest retrieval area is updated in real time based on the displacement information.

9. The method of claim 2, wherein, The method includes: In response to determining that the user input information does not include image and video information acquired in real time by the image acquisition device on the user's client device, the user's intent type is determined based on the user input information using the large model, wherein determining the first retrieval distance based on the user input information includes: Based on the intent type, the first retrieval distance is determined using a preset mapping rule, wherein the mapping rule is used to describe the correspondence between the intent type and the first retrieval distance.

10. The method according to any one of claims 2-9, wherein, The method further includes: Obtain the traffic information of the current location; and Based on the road condition information, the center point of the point of interest retrieval area is adjusted within a preset distance range of the current location.

11. The method according to any one of claims 1-10, wherein, In response to determining that the target navigation content generation component includes multiple function functions for completing a specific task, the function description file further includes descriptive information about the dependencies between the multiple function functions, and wherein, based on the user input information, the points of interest, and the parameter description file, the agent runs the function functions to determine the navigation content of the user input information, including: Based on the user input information, the points of interest, the parameter description file, and the dependencies, the agent runs the multiple function functions to determine the navigation content of the user input information.

12. A navigation content generation device based on intelligent agents, wherein, The intelligent agent includes multiple candidate navigation content generation components, and each candidate navigation content generation component includes at least a functional description file for that candidate navigation content generation component, a function for generating specific navigation content, and a parameter description file for the function. The device includes: The target navigation content generation component determination module is configured to determine the target navigation content generation component from the plurality of candidate navigation content generation components based on user input information and the function description file. The point of interest retrieval region determination module is configured to determine the point of interest retrieval region based on the user input information and the user's status information, wherein the status information includes at least the user's current location and transient velocity; The point of interest determination module is configured to determine the user's points of interest based on the user input information and the point of interest retrieval region, using a large model; and The navigation content determination module is configured to determine the navigation content of the user input information by using the intelligent agent to run the function corresponding to the target navigation content generation component based on the user input information, the points of interest, and the parameter description file.

13. An intelligent agent, comprising: The planner is configured to determine the target navigation content generation component based on user input information and the function description file; as well as An executor is configured to generate navigation content according to the method of any one of claims 1-11.

14. An electronic device comprising: At least one processor; as well as A memory that is communicatively connected to the at least one processor; in The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-11.

15. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-11.

16. A computer program product comprising a computer program, wherein, When the computer program is executed by a processor, it implements the method of any one of claims 1-11.

Images