Map knowledge base construction method, map updating method and related devices

Abstract

The invention discloses a map knowledge base construction method, a map updating method and a related device.The map knowledge base construction method comprises the steps that contour recognition is conducted based on a global scanning map of a target region, and a plurality of target contours and description information of the target contours in the global scanning map are obtained; based on the relative distribution relationship between different target contours and the description information of the target contours, selecting existing scene data matched with the target area as target scene data; based on preset scene data, obtaining a movement attribute of each target contour; and based on the plurality of target contours, the description information of the target contours, the moving attributes of the target contours, the relative distribution relationship between the target contours and the preset scene data, constructing a prior knowledge base of the target area. According to the scheme, the accuracy of map updating strategy generation can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of map updating, in particular to a map knowledge base construction method, a map updating method and related devices. BACKGROUND

[0002] With the development of artificial intelligence in the field of computer vision, mobile intelligent devices are increasingly widely used in various scenarios. These intelligent devices usually rely on map data of the environment they are in to perform tasks such as navigation, positioning and environment perception.

[0003] Since the environment will gradually change over time, in order to maintain the mobility of the device, the map needs to be continuously updated. In the prior art, the map updating operation is usually triggered based on a pre-set fixed time frequency or a threshold of map change rate, so as to cover the existing map data with the latest laser data collected after triggering the map update. However, this updating method is difficult to adapt to the complexity of dynamic changes in the environment, and cannot determine the data that really needs to be updated in the map, thereby causing the problem of unreasonable map update frequency and poor applicability of the updated map. Therefore, how to improve the accuracy of the generated map update strategy has become a problem to be solved. SUMMARY

[0004] The technical problem solved by the present application is to provide a map knowledge base construction method, a map updating method and related devices, which can improve the accuracy of the generated map update strategy.

[0005] To solve the above technical problem, the first aspect of the present application provides a map knowledge base construction method, comprising: performing contour recognition based on a global scanning map of a target area to obtain a plurality of target contours in the global scanning map and description information of the target contours; selecting existing scene data matching the target area as target scene data based on the relative distribution relationship between different target contours and the description information of the target contours; wherein the existing scene data at least includes a preset object in a preset scene, contour description information of the preset object, a relative distribution relationship between the preset objects and a movement attribute of the preset object, the movement attribute representing the variation characteristics of the spatial position of the object; obtaining the movement attribute of each target contour based on the preset scene data; and constructing a prior knowledge base of the target area based on the plurality of target contours, the description information of the target contours, the movement attribute of the target contours, the relative distribution relationship between the target contours and the preset scene data.

[0006] To address the aforementioned technical problems, a second aspect of this application provides a map updating method, comprising: acquiring a currently scanned map collected by a target device while it is running in a target area; performing contour recognition on the currently scanned map to obtain contours of several items within the target scanned map and descriptive information of the item contours; obtaining the movement attributes of the item contours based on a prior knowledge base of the target area and the descriptive information of the item contours, and acquiring the contour changes between the currently scanned map and the global scanned map of the target area; wherein the prior knowledge base of the target area is obtained based on the map knowledge base construction method described in the first aspect above; generating a target strategy based on the descriptive information of the item contours, the movement attributes of the item contours, the contour changes, and the prior knowledge base; wherein the target strategy is used to instruct the update strategy of at least one of the global scanned map and the prior knowledge base.

[0007] To address the aforementioned technical problems, a third aspect of this application provides a map knowledge base construction apparatus, comprising: a first identification module, a scene selection module, an attribute determination module, and a knowledge base construction module. The first identification module is used to perform contour recognition based on a global scan map of the target area to obtain several target contours and their descriptive information within the global scan map. The scene selection module is used to select existing scene data matching the target area as target scene data based on the relative distribution relationship between different target contours and their descriptive information. The existing scene data includes at least preset items within a preset scene, their contour description information, the relative distribution relationship between the preset items, and their movement attributes, where the movement attributes characterize the spatial positional changes of the items. The attribute determination module is used to obtain the movement attributes of each target contour based on the preset scene data. The knowledge base construction module is used to construct a priori knowledge base for the target area based on several target contours, their descriptive information, their movement attributes, the relative distribution relationship between them, and the preset scene data.

[0008] To address the aforementioned technical problems, a fourth aspect of this application provides a map updating apparatus, comprising: a map acquisition module, a second recognition module, a contour change module, and a strategy generation module. The map acquisition module acquires a currently scanned map collected by a target device while it is running in a target area. The second recognition module performs contour recognition on the currently scanned map to obtain contours of several items within the target scanned map and descriptive information of those contours. The contour change module obtains the movement attributes of the item contours based on a priori knowledge base of the target area and the descriptive information of the item contours, and acquires the contour change between the currently scanned map and the global scanned map of the target area. The priori knowledge base of the target area is obtained based on the map knowledge base construction apparatus described in the third aspect above. The strategy generation module generates a target strategy based on the descriptive information of the item contours, the movement attributes of the item contours, the contour change, and the priori knowledge base. The target strategy instructs the update strategy of at least one of the global scanned map and the priori knowledge base.

[0009] To address the aforementioned technical problems, the fifth aspect of this application provides an electronic device, comprising at least a memory and a processor coupled to each other. The memory stores at least program instructions, and the processor executes the program instructions to implement the map knowledge base construction method in the first aspect, or to implement the map update method in the second aspect.

[0010] To address the aforementioned technical problems, a sixth aspect of this application provides a computer-readable storage medium storing program instructions executable by a processor. These program instructions are used to implement the map knowledge base construction method in the first aspect, or to implement the map update method in the second aspect.

[0011] The above scheme performs contour recognition based on a global scan map of the target area, obtaining several target contours and their descriptive information within the global scan map. Based on the relative distribution relationships between different target contours and their descriptive information, existing scene data matching the target area is selected as the target scene data. This existing scene data includes at least preset items within a preset scene, their contour descriptions, the relative distribution relationships between the preset items, and their movement attributes. Movement attributes characterize the spatial positional changes of the items. Based on the preset scene data, the movement attributes of each target contour are obtained. Finally, based on several target contours, their descriptive information, their movement attributes, and the relationships between the target contours... Based on the relative distribution relationships between objects and preset scene data, a prior knowledge base for the target area is constructed. During map updates, the current scan map collected by the target device while it is running in the target area is acquired. Contour recognition is performed on the current scan map to obtain the contours of several objects and their descriptions within the target scan map. Based on the prior knowledge base of the target area and the descriptions of the object contours, the movement attributes of the object contours are obtained, and the contour changes between the current scan map and the global scan map of the target area are acquired. Based on the descriptions of the object contours, their movement attributes, the contour changes, and the prior knowledge base, a target strategy is generated to indicate the update strategy of at least one of the global scan map and the prior knowledge base. On the one hand, based on the analysis of the contour information within the global scan map of the target area, preset scene data matching the target area is obtained. Utilizing the prior knowledge provided by this preset scene data, the movement characteristics of objects corresponding to the target contours within the target area are determined, distinguishing between short-term fluctuations and long-term changes of different objects in the target area. This constructs a more targeted prior knowledge base, providing the most reliable data foundation and decision-making basis possible for map updates of the target area. On the other hand, by integrating the descriptive information, movement attributes, and contour changes of object contours, key data that needs to be updated to the map can be selected, thereby improving the accuracy of map update strategy generation and enhancing the applicability of the updated map. Therefore, it can improve the accuracy of map update strategy generation. Attached Figure Description

[0012] Figure 1 This is a flowchart illustrating an embodiment of the map knowledge base construction method of this application; Figure 2 This is a flowchart illustrating an embodiment of the map updating method of this application; Figure 3 This is a schematic diagram of the framework of an embodiment of the map knowledge base construction device of this application; Figure 4 This is a schematic diagram of the framework of an embodiment of the map updating device of this application; Figure 5 This is a schematic diagram of the framework of an embodiment of the electronic device of this application; Figure 6 This is a schematic diagram of a framework of an embodiment of the computer-readable storage medium of this application. Detailed Implementation

[0013] The embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0014] In the following description, specific details such as particular system architectures, interfaces, and technologies are presented for illustrative purposes rather than for limiting purposes, in order to provide a thorough understanding of this application.

[0015] In this paper, the terms "system" and "network" are often used interchangeably. The term "and / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. Additionally, the slash " / " generally indicates that the preceding and following related objects have an "or" relationship. Furthermore, "many" in this paper indicates two or more objects.

[0016] Please see Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of the map knowledge base construction method of this application. Specifically, it may include the following steps: Step S11: Perform contour recognition based on the global scan map of the target area to obtain several target contours and their descriptive information within the global scan map.

[0017] In this embodiment, the global scan map of the target area can be acquired by sensors such as LiDAR and depth cameras. This global scan map reflects the overall environmental layout of the target area at the time of acquisition. Specifically, the global scan map can be a point cloud map or a raster map, etc., and is not limited thereto in this application.

[0018] In one implementation scenario, image data of the target area is acquired, and contour segmentation is performed based on a global scan map to obtain target contours belonging to various items. The image data usually contains information such as the color, texture, and shape of the items in the target area. Based on the image data, item features related to the target contour are extracted, such as color and item category. Based on the target contour, geometric features of the target contour are extracted. Based on the item features and geometric features of the target contour, descriptive information of the target contour is obtained.

[0019] In a specific implementation scenario, a global scan map of the target area can be obtained by scanning with a lidar device, and image data of the target area can be collected based on a camera device.

[0020] In a specific implementation scenario, based on the positioning relationship between image data and a global scan map, the features and geometric characteristics of items belonging to the same target contour are associated to form complete descriptive information of the target contour. For example, if the image data shows that the item corresponding to a certain target contour is a red cuboid box, and its geometric dimensions are determined to be 1 meter long, 0.5 meters wide, and 0.8 meters high through the global scan map, then by integrating information such as the color "red," the item category "cuboid box," and the geometric dimensions "1 meter long, 0.5 meters wide, and 0.8 meters high," a detailed descriptive information of the target contour is obtained.

[0021] In another implementation scenario, when a global scan map of the target area is obtained by scanning with a LiDAR device, the global scan map can be converted into an image. Then, a visual model such as a convolutional network can be used to extract the geometric features of the contour, such as shape, perimeter, area, roundness, rectangularity, minimum outer rectangle, and minimum outer circle, as descriptive information of the target contour. The shape can be a straight line, arc, circle, broken line, right angle, etc.

[0022] In a specific implementation scenario, segmentation algorithms such as U-Net depth method and region growing method are used to segment the contours in the global scan map to separate the contours belonging to the same item and obtain several target contours. For example, multiple load-bearing columns in the target area are divided into multiple independent target contours, and a shelf in the target area is divided into a separate target contour.

[0023] Step S12: Based on the relative distribution relationship between different target contours and the descriptive information of the target contours, select existing scene data that matches the target region as the target scene data.

[0024] In this embodiment of the disclosure, the existing scene data includes at least preset items in the preset scene, outline description information of the preset items, relative distribution relationship between preset items, and movement attributes of preset items. Movement attributes characterize the change characteristics of the spatial position of the items. The preset scene is usually a specific application space, such as a warehouse or factory workshop. The preset items, item distribution, and movement attributes are different in different preset scenes. Taking movement attributes as an example, shelf A in warehouse A is used for long-term storage of goods. In the scene corresponding to warehouse A, the movement attribute of shelf A is long-term stationary. However, in warehouse B, shelf A is used as a temporary stacking rack for goods and moves frequently due to the turnover of goods. Therefore, in the scene corresponding to warehouse B, the movement attribute of shelf A is short-term stationary.

[0025] In an implementation scenario, movement attributes include, but are not limited to, long-term stationary, short-term stationary, periodic movement, and random movement. Long-term stationary items, such as fixed shelves in a warehouse or load-bearing columns in a building, maintain their spatial position over a relatively long period. Short-term stationary items, such as promotional displays or temporarily stacked goods, have relatively fixed spatial positions in the short term, but these positions may change due to the end of an activity or goods turnover. Periodically moving items, such as transport vehicles on a factory production line or sorting robots in a logistics warehouse, change their spatial positions according to certain time patterns or task requirements. Randomly moving items, such as pedestrians in crowded places or vehicles in parking lots, exhibit uncertain spatial position changes.

[0026] In one implementation scenario, the description information and relative distribution of the contours of each target in the target area are compared with existing scene data in the scene database. By calculating similarity, the existing scene data with the highest similarity is selected as the target scene data. For example, if there are multiple neatly arranged cuboid shelves in the target area, and the aisle distribution between the shelves is similar to the shelf distribution and description information of a warehouse scene in the preset scene database, then the existing scene data of that warehouse scene is selected as the target scene data.

[0027] In a specific implementation scenario, if the distribution of items within the target area is cluttered or the target area contains multiple different functional zones, and only some item descriptions and distributions are similar to those in a pre-defined scenario, multiple similarity calculation methods can be combined to select the best-matching existing scene data. For example, both the similarity of item outline shapes and the similarity of relative distances between items can be considered to ensure that the target scene data best matches the actual situation of the target area. Alternatively, the global scan map can be divided into multiple sub-regions, and existing scene data matching each sub-region can be selected as sub-target scene data. Then, the sub-target scene data can be integrated and processed to obtain target scene data applicable to the entire target area.

[0028] In one implementation scenario, a global scan map is divided into several sub-regions. Based on the relative distribution relationship between different target contours within each sub-region and the descriptive information of the target contours within the sub-region, a first prompt instruction is constructed. This first prompt instruction instructs a first large model to select existing scene data that meets the similarity criteria based on the matching similarity between existing scene data and the sub-regions. The output data of the first large model in response to the first prompt instruction is then obtained as the preset scene data. Specifically, the first large model can be a large language model, a pre-trained artificial intelligence model with scene matching capabilities, etc. The specific type of the first large model is not limited in this application. This model has been trained by matching a large amount of existing scene data with scan data from different regions, and can output the most matching existing scene data based on the sub-region information, including the relative distribution relationship and descriptive information of the target contours.

[0029] In a specific implementation scenario, the similarity condition is defined as: the matching similarity is not lower than the preset threshold and the similarity is the highest among all matching items. Specifically, the preset threshold can be flexibly set according to the accuracy requirements of the actual application scenario. For example, in the warehouse management scenario, if the matching accuracy of item distribution is high, the preset threshold can be set to 90%, that is, only existing scenario data with a matching similarity of 90% or higher will be selected.

[0030] In a specific implementation scenario, different area sizes are pre-set, such as 10*10m, 20*20m, etc. The global scan map is divided by traversing areas of different sizes to obtain several sub-regions. For example, when there are warehouse storage areas and office areas with different sizes within the target area, the arrangement of items in the warehouse storage area differs significantly from that in the office area. These two functional areas can be divided according to different area sizes. By dividing and matching these areas by different sizes, existing scene data mainly consisting of shelves and goods storage can be matched to the warehouse storage area, while existing scene data mainly consisting of desks, chairs, and filing cabinets can be matched to the office area. This solution can more flexibly adapt to the complex distribution of different functional areas or items within the target area and can more accurately obtain pre-set scene data that matches the actual environment of the sub-region.

[0031] Step S13: Based on the preset scene data, obtain the movement attributes of each target contour.

[0032] In one implementation scenario, based on the movement attributes of preset items in the preset scenario data and the matching relationship between target contours and preset items, the movement attributes corresponding to each target contour are determined. For example, when the preset scenario data indicates that the movement attribute of a preset item is long-term stationary, such as a heavy-duty shelf in a warehouse, if the target contour matches the preset item successfully, the movement attribute of the target contour is determined to be long-term stationary; if the movement attribute of a preset item in the preset scenario data is short-term movement, such as a promotional display stand, the movement attribute of the matched target contour is correspondingly determined to be short-term movement.

[0033] In a specific implementation scenario, when the first model detects that no existing scene data meets the similarity conditions with the target sub-region, the descriptive information of the target contour within at least one candidate sub-region is obtained as the first descriptive information, and the descriptive information of the target contour within the target sub-region is obtained as the second descriptive information. The target sub-region is one of several sub-regions, and the candidate sub-regions are sub-regions in the globally scanned map that are adjacent to the target sub-region and have existing scene data that meets the similarity conditions with the candidate sub-regions. Based on the similarity result between the first and second descriptive information and the existing scene data that matches the candidate sub-regions, movement attributes are matched for the target contours within the target sub-region. This scheme, when faced with a situation where there is no directly matching existing scene data for the target sub-region, can indirectly infer the possible movement attributes of items within the target sub-region by introducing information from adjacent candidate sub-regions with matching data.

[0034] In a specific implementation scenario, the similarity between the target sub-region and the target contour description information in adjacent candidate sub-regions is calculated. This similarity analysis may involve comparing features such as the shape, size, and category of the items. If the description information of both is greater than a preset threshold, it can be reasonably inferred that the items in the target sub-region may match the candidate sub-regions in terms of movement attributes. That is, if the items in the candidate sub-regions are mostly stationary for a long time, then the same or similar items in the target sub-region may also have similar long-term stationary characteristics.

[0035] In a specific implementation scenario, when multiple candidate sub-regions exist and the item movement attributes in their respective matched existing scene data are inconsistent, a comprehensive movement attribute can be assigned to items within the target sub-region by combining the similarity weights between each candidate sub-region and the target sub-region, as well as the prevalence or importance of the corresponding item movement attributes in each candidate sub-region. For example, if in two adjacent candidate sub-regions, one region's items are generally stationary for a long time, while the other is stationary for a short time, and the target sub-region has a higher similarity to the long-term stationary region, then the long-term stationary movement attribute might be preferentially assigned to items within the target sub-region. Alternatively, the movement attribute could be assigned based on the movement attribute possessed by the majority of items in the candidate sub-regions, or a relatively reasonable movement attribute could be determined by the median or average of the movement attribute distribution. After matching the movement attribute to the target contour within the target sub-region, the corresponding scene for the target sub-region is defined based on the relative distribution relationship between different target contours within the target sub-region.

[0036] In one implementation scenario, after obtaining the movement attributes of each target contour, each target contour is encoded to obtain a unique identifier. This unique identifier can be in the form of numbers, letters, or a combination thereof, ensuring that each target contour has a unique ID in the map knowledge base. Encoding facilitates subsequent querying, tracking, and management of target contours.

[0037] Step S14: Based on several target contours, the description information of the target contours, the movement attributes of the target contours, the relative distribution relationship between the target contours, and the preset scene data, construct a prior knowledge base for the target region.

[0038] In an implementation scenario, when constructing a prior knowledge base for the target area, the unique identifiers, descriptions, and movement attributes of each target contour within the global scan map are first stored in a structured manner, for example, using database tables or plain text. Each row represents a target contour, and each target contour stores its identifier, description, movement attributes, and related information from the preset scene data. For example, the prior knowledge base stores: "Wall, permanently stationary, is a straight line segment with a length exceeding 3m or an arc segment with a circumference exceeding 2m (its minimum circumscribed circle diameter is approximately 2m)." The building's supporting columns are square with sides approximately 0.5m in length and remain stationary for extended periods. The machine is stationary for a long time and is rectangular in shape, about 2*3m in size. A small cart for moving goods, movable at any time, approximately 1*0.5m rectangular; The pipes are randomly distributed, stationary, and unstructured. The smallest circumcircle of the outline is less than 5 cm. "A reflective pillar, stationary for a long period, a circle with a diameter of 10cm." Specifically, the preset scene data includes the type of preset scene, a list of preset items and their corresponding outline descriptions and movement attributes. The preset scene data is then associated and stored with the target scene data obtained by matching the actual target area. This approach allows the prior knowledge base to not only reflect the actual situation of the current target area but also, with the help of the rich information in the preset scene data, provide a wider range of references for subsequent map analysis and applications.

[0039] In a specific implementation scenario, based on the relative distribution relationships between target contours, regularly distributed target contours are associated and stored. For example, the contours of various shelves on the same row of shelves are linked together, or the contours of related items within the same functional area are linked together. This associative storage method more clearly presents the layout logic and functional zoning of items within the target area, facilitating subsequent queries and use of the map knowledge base. For instance, in the prior knowledge base of a factory workshop, the contours of various equipment on the same production line are associated and stored. When querying equipment information for that production line, descriptions and movement attributes of all related equipment can be retrieved at once, improving query efficiency. Simultaneously, this associative storage also facilitates subsequent updates and maintenance of the map knowledge base; when the layout of items in a certain area changes, the associated contour information that needs modification can be located more accurately.

[0040] In a specific implementation scenario, the prior knowledge base can be categorized and stored according to the movement attributes of the target contours. Long-term static item contours, short-term static item contours, periodically moving item contours, and randomly moving item contours can be stored in different data tables or file partitions. This allows for rapid filtering of item contour information with corresponding movement attributes during subsequent map analysis, based on different needs. For example, when conducting warehouse inventory checks, information on shelves and fixed storage equipment can be prioritized from long-term static item contour data; when analyzing personnel and vehicle flow, relevant data can be primarily obtained from randomly moving item contour data. This approach further improves the efficiency and relevance of the prior knowledge base, providing stronger support for map analysis and applications.

[0041] In one implementation scenario, the prior knowledge base also includes: reference item types used to indicate the area positioning within the target area during map updates. These reference item types are usually selected from items within the target area that have significant features and relatively fixed positions. For example, the prior knowledge base also stores: "long-term stationary objects". Structured geometric contours: straight lines, broken lines, arcs, circles; not a collection of small, scattered dots (e.g., circles with a diameter less than 5cm); "Man-made markers such as reflective pillars and reflective strips." The above solution, by clearly defining the types of these reference items in the prior knowledge base, can utilize the stable positional information of these reference items during subsequent map updates to more accurately locate and calibrate other items and areas on the map, thereby effectively improving the accuracy and reliability of map updates.

[0042] In one implementation scenario, before or after building the prior knowledge base for the target area, a prompt message is generated to guide the target object to correct or confirm the prior knowledge base. Specifically, the target object could be personnel responsible for map construction and maintenance. This prompt message can be sent to the target object through various methods such as system interface pop-ups, email notifications, and mobile application push notifications. The content of the prompt message can include potential inaccuracies in the prior knowledge base, item outline information requiring further confirmation, and movement attributes. For example, the prompt message could be, "In the prior knowledge base, the movement attribute of a shelf in a certain area is marked as short-term stationary, but on-site observation may reveal a misjudgment; please verify and correct it," or "In the newly built prior knowledge base, the descriptions of some item outlines are not detailed enough; please supplement and improve them." After receiving the prompt message, the target object can correct or confirm the prior knowledge base according to the actual situation, ensuring the accuracy and completeness of the prior knowledge base and providing a reliable data foundation for subsequent map analysis and applications.

[0043] In a specific implementation scenario, the target object can further supplement the prior knowledge base by adding outlines and their descriptions that are not marked in the global scan map. For example, if the global scan map fails to fully capture the outlines of certain small objects or special structures within the target area due to resolution limitations or scanning angle issues, the target object can supplement the prior knowledge base with this missing outline information and its detailed descriptions through on-site investigation or by referring to other reliable data. The supplementary content can include the object's name, shape, size, material, and possible movement attributes, thereby further enriching the content of the prior knowledge base.

[0044] The above scheme performs contour recognition based on a global scan map of the target area, obtaining several target contours and their descriptive information within the global scan map. Based on the relative distribution relationships between different target contours and their descriptive information, existing scene data matching the target area is selected as the target scene data. This existing scene data includes at least preset items within a preset scene, their contour descriptions, relative distribution relationships between preset items, and their movement attributes. Movement attributes characterize the spatial positional changes of the items. Based on the preset scene data, the movement attributes of each target contour are obtained. A prior knowledge base for the target area is constructed based on several target contours, their descriptive information, their movement attributes, relative distribution relationships, and the preset scene data. Therefore, by analyzing the contour information within the global scan map of the target area, obtaining preset scene data matching the target area, and utilizing the prior knowledge provided by the preset scene data, the movement characteristics of items corresponding to the target contours within the target area are determined. This distinguishes between short-term fluctuations and long-term changes of different items within the target area, thus constructing a more targeted prior knowledge base and providing a reliable data foundation and decision-making basis for map updates. This improves the accuracy of map update strategy generation.

[0045] Please see Figure 2 , Figure 2 This is a flowchart illustrating an embodiment of the map updating method of this application. Specifically, it may include the following steps: Step S21: Obtain the current scan map collected by the target device when it is running in the target area.

[0046] In this embodiment, the target device can be a mobile robot, drone, or AGV equipped with scanning sensors. These target devices can move within the target area according to a preset path or autonomous navigation, and collect the current scan map in real time during the movement. The current scan map contains the real-time location, shape, size, and other contour information of items within the target area, which can be in the form of digital images or point cloud data. For example, in a large warehouse, a mobile robot can move along the aisles between shelves, using its onboard LiDAR and other sensors to collect the contour information of shelves, goods, and other facilities in the warehouse in real time, generating the current scan map.

[0047] In a specific implementation scenario, after acquiring the currently scanned map, preprocessing operations are performed on it. These preprocessing operations can include steps such as noise reduction, filtering, and coordinate transformation to eliminate errors and interference that may have been introduced during the acquisition process, thereby improving the accuracy and reliability of the map data. For example, for point cloud data acquired by LiDAR, filtering algorithms can be used to remove outliers and noise points, while coordinate transformations are performed to unify the point cloud data to the same coordinate system for comparison and analysis with the global scanned map.

[0048] In a specific implementation scenario, when acquiring the current scan map, the target device may only perform a detailed scan of a sub-region within the target area due to factors such as power limitations, task planning, or environmental obstacles. In this case, the acquired current scan map only covers a portion of the target area, rather than a complete global view. For example, in a complex factory environment, an AGV might only scan a specific production workshop or warehouse area to generate a current scan map, which represents only a subset of the entire factory area. Therefore, in subsequent map updates, it is necessary to obtain the spatial correspondence between the current scan map and the global scan map. Specifically, methods such as coordinate matching, feature point alignment, or semantic association can be used to associate and compare the item outlines and descriptions in the current scan map with the corresponding parts in the global scan map.

[0049] Step S22: Perform contour recognition on the current scanned map to obtain the contours of several items within the target scanned map and their descriptive information.

[0050] In one implementation scenario, contour segmentation is performed on the currently scanned map to obtain several object contours. Then, descriptive information of the object contours is extracted using methods such as visual models. Specifically, the descriptive information includes at least geometric features and may also include object features. For details, please refer to the detailed steps in the aforementioned embodiments. For the sake of brevity, they will not be repeated here.

[0051] In a specific implementation scenario, the outlines of objects in the currently scanned map are encoded to correspond to descriptive information. The specific encoding method can be selected according to actual needs, and this application does not impose any restrictions.

[0052] Step S23: Based on the prior knowledge base of the target area and the description information of the item outline, obtain the movement attributes of the item outline and obtain the outline change between the current scan map and the global scan map of the target area.

[0053] In this embodiment of the disclosure, the prior knowledge base of the target area is obtained by the map knowledge base construction method in any of the foregoing embodiments. For details, please refer to the detailed description of the foregoing embodiments. For the sake of brevity, it will not be repeated here.

[0054] In one implementation scenario, based on a prior knowledge base of the target area and the description information of the object's outline, the movement attributes of the object's outline can be obtained by querying entries in the prior knowledge base that match the description information of the current object's outline. For example, if the description information of an object's outline identified in the current scanned map is "a rectangle of approximately 2*3m", entries with similar description information can be queried in the prior knowledge base to determine the movement attributes of the object's outline. Simultaneously, by comparing the outlines of objects in the same or similar areas in the current scanned map with those in the global scanned map, changes in the outline can be obtained, such as the addition, deletion, movement, or shape change of the object.

[0055] In a specific implementation scenario, if the outline of an item in the current scan map does not exist in the global scan map, its corresponding movement attribute can be initially inferred based on the movement attributes of similar item outlines in the prior knowledge base or the movement attributes of target outlines belonging to the same scene. Alternatively, it can be marked as a newly appeared item, and staff can be prompted for further confirmation. For example, if a "1*0.5m rectangle" item outline appears in the current scan map but does not appear in the global scan map, while the prior knowledge base contains the description "a small cart for transporting goods, movable at any time, a rectangle of about 1*0.5m", the movement attribute of the new item outline can be initially inferred as "movable at any time", and a prompt message can be sent to staff requesting verification of the item's actual movement attribute.

[0056] In a specific implementation scenario, techniques such as image comparison or feature point matching can be used to obtain information about contour changes. For example, the current scanned map can be compared with the global scanned map at the pixel level to find the areas of difference between the two, or feature points of the object contour, such as corner points and edge points, can be extracted, and the contour can be determined by comparing the position and number of feature points.

[0057] In a specific implementation scenario, reference item types stored in a prior knowledge base can be used as a benchmark when acquiring contour changes. This allows for more accurate identification and judgment of contour changes in other items. For example, if reflective pillars are specified as reference item types in the prior knowledge base, and their precise locations in the global scan map are known, then when comparing the current scan map with the global scan map, the positions of these reflective pillars can be located first to confirm whether they have moved. This information can then be used as a reference to determine the relative changes in the contours of surrounding items, such as whether a shelf has shifted relative to the reflective pillars. This approach effectively improves the accuracy and reliability of contour change judgment, providing a more precise basis for subsequent map update strategies.

[0058] In a specific implementation scenario, based on the contour changes, it is possible to determine the new contours of the current scanned map relative to the global scanned map, the disappearing contours of the current scanned map relative to the global scanned map, and the changes in the overlapping areas of the contours of the current scanned map relative to the global scanned map, such as changes in position and angle, contour geometric features, object attributes, and other features.

[0059] In a specific implementation scenario, given a sub-region of the target area corresponding to the current scan map collected by the target device, a local scan map consistent with the current scan map area is determined in the global scan map. Based on the circumscribed graphic of the object outline, the spatial range of the object outline is determined. Based on the spatial range of the object outline and the target outline within the local scan map, the corresponding region of the object outline in the local scan map is determined. Based on the corresponding region and the spatial range of the target outline within the local scan map, the contour change between the object outline and the target outline is calculated. Based on the contour change, the contour change between the current scan map and the global scan map is determined.

[0060] Step S24: Generate the target strategy based on the description information of the item outline, the movement attributes of the item outline, the outline change situation, and the prior knowledge base.

[0061] In this embodiment of the disclosure, the target strategy is used to indicate the update strategy of at least one of the global scan map and the prior knowledge base. The update strategy includes, but is not limited to, updating, not updating, updating the outline of an item to the global scan map, updating the description information of the outline of an item to the prior knowledge base, etc., and is not limited in this application.

[0062] In one implementation scenario, a second prompt instruction is constructed based on the description information of the object's outline, its movement attributes, outline changes, and a prior knowledge base. This second prompt instruction instructs the second main model to determine which data in the global scan map and prior knowledge base has a long-term static movement attribute and needs to be updated. The output data of the second main model in response to the second prompt instruction is then obtained as the target strategy. For example, the second prompt instruction could be "#role". You are an excellent laser positioning map expert, capable of making the most reliable choice regarding map updates when receiving the currently scanned map based on skill information.

[0063] Skill ##Reference item types suitable for area positioning - Fixed, stationary objects For objects with vertical surfaces (AMR laser scanning height), this avoids abrupt changes in scanning information due to uneven ground or different laser tilt angles. Structured geometric contours: straight lines, broken lines, arcs, circles: not a collection of small, scattered dots (circles smaller than 5cm in diameter). Artificially placed reflective pillars, reflective strips, and other markers ##Target Environment Information - This is a factory scene where only the walls, factory pillars (50cm square), and machines (2m x 2m x 2m) are stationary; everything else is movable. #User-provided information ##Currently scanned map information - The mounting height of the 2D laser in the scanned AMR is 25cm. • New additions compared to the global scan map: -Structured outlines include: approximately straight line segments longer than 3m, square outlines around 50cm, right-angled outlines around 2m x 1.5m, circular outlines around 10cm in diameter, and zigzag outlines within 30cm. - Unstructured contours: Multiple scattered regions with a distribution range of approximately 2m x 0.5m, and multiple rectangular chaotic regions with a distribution range of approximately 40cm x 30cm. • Disappears relative to the global scan map: - A rectangular outline approximately 1m x 1m in size near the wall. • Changes in the outline within the same space compared to the old drawing: The approximately 1.8m x 3.5m right-angled outline was shifted about 5cm to the lower left corner. - Multiple rectangular outlines, approximately 1.2 x 0.8 m in size, can be moved by about 0.5 m and rotated within 10 degrees. ##question 1. Please help me select a suitable outline from the currently scanned map and update it to the global scanned map that is in use; 2. Tell me which outlines in the global scan map need to be erased; 3. What data in the map knowledge base needs to be updated?

[0064] The above scheme, during the map update process, acquires the current scan map collected by the target device while it is running in the target area, performs contour recognition on the current scan map to obtain the contours of several items in the target scan map and their descriptions, obtains the movement attributes of the item contours based on the prior knowledge base of the target area and the descriptions of the item contours, and obtains the contour changes between the current scan map and the global scan map of the target area. Based on the descriptions of the item contours, their movement attributes, the contour changes, and the prior knowledge base, a target strategy is generated to indicate the update strategy of at least one of the global scan map and the prior knowledge base. On the one hand, based on the analysis of the contour information within the global scan map of the target area, preset scene data matching the target area is obtained. Utilizing the prior knowledge provided by this preset scene data, the movement characteristics of objects corresponding to the target contours within the target area are determined, distinguishing between short-term fluctuations and long-term changes of different objects in the target area. This constructs a more targeted prior knowledge base, providing a reliable data foundation and decision-making basis for map updates. On the other hand, by integrating the descriptive information, movement attributes, and contour changes of object contours, key data that needs to be updated to the map can be selected, thereby improving the accuracy of map update strategy generation and enhancing the applicability of the updated map. Therefore, it can improve the accuracy of map update strategy generation.

[0065] Please see Figure 3 , Figure 3 This is a schematic diagram of the framework of an embodiment of the map knowledge base construction device of this application. The map knowledge base construction device 30 includes: a first identification module 31, a scene selection module 32, an attribute determination module 33, and a knowledge base construction module 34. The first identification module 31 is used to perform contour recognition based on a global scan map of the target area to obtain several target contours and their descriptive information within the global scan map. The scene selection module 32 is used to select existing scene data matching the target area as target scene data based on the relative distribution relationship between different target contours and their descriptive information. The existing scene data includes at least preset items within a preset scene, their contour description information, the relative distribution relationship between preset items, and their movement attributes. The movement attributes characterize the spatial position change characteristics of the items. The attribute determination module 33 is used to obtain the movement attributes of each target contour based on the preset scene data. The knowledge base construction module 34 is used to construct a priori knowledge base for the target area based on several target contours, their descriptive information, their movement attributes, the relative distribution relationship between them, and the preset scene data.

[0066] The above scheme involves a map knowledge base construction device 30 performing contour recognition based on a global scan map of the target area. This yields several target contours and their descriptive information within the global scan map. Based on the relative distribution relationships between different target contours and their descriptive information, existing scene data matching the target area is selected as the target scene data. This existing scene data includes at least preset items within a preset scene, their contour descriptions, the relative distribution relationships between preset items, and their movement attributes. Movement attributes characterize the spatial positional changes of the items. Based on the preset scene data, the movement attributes of each target contour are obtained. Finally, based on several target contours, their descriptive information, their movement attributes, the relative distribution relationships between target contours, and the preset scene data, a priori knowledge base for the target area is constructed. Therefore, by analyzing the contour information within the global scan map of the target area, and obtaining preset scene data matching the target area, the prior knowledge provided by the preset scene data is used to determine the movement characteristics of items corresponding to the target contours within the target area. This distinguishes between short-term fluctuations and long-term changes of different items within the target area, thus constructing a more targeted priori knowledge base and providing a reliable data foundation and decision-making basis for map updates. This improves the accuracy of map update strategy generation.

[0067] In some disclosed embodiments, the scene selection module 32 further includes a sub-region division module (not shown), used to divide the scene based on a global scan map to obtain several sub-regions; the scene selection module 32 further includes a first instruction construction module (not shown), used to construct a first prompt instruction based on the relative distribution relationship between different target contours within the sub-regions and the descriptive information of the target contours within the sub-regions; wherein, the first prompt instruction is used to instruct the first large model to select existing scene data that meets the similarity condition based on the matching similarity between existing scene data and sub-regions; the scene selection module 32 further includes a first output module (not shown), used to obtain the output data of the first large model in response to the first prompt instruction, as preset scene data.

[0068] In some disclosed embodiments, when the first large model detects that there is no existing scene data that meets the similarity conditions with the target sub-region, the map knowledge base construction device 30 further includes a description information acquisition module (not shown), used to acquire description information of the target contour in at least one candidate sub-region as first description information, and acquire description information of the target contour in the target sub-region as second description information; wherein, the target sub-region is one of several sub-regions, and the candidate sub-region is a sub-region in the globally scanned map that is adjacent to the target sub-region and has existing scene data that meets the similarity conditions with the candidate sub-region; the map knowledge base construction device 30 further includes an existing scene matching module (not shown), used to match movement attributes for the target contour in the target sub-region based on the similarity result between the first description information and the second description information and the existing scene data that matches the candidate sub-region.

[0069] In some disclosed embodiments, the first recognition module 31 further includes a contour segmentation module (not shown), used to acquire image data about the target area and perform contour segmentation based on a global scan map to obtain target contours belonging to each item; the first recognition module 31 further includes a feature extraction module (not shown), used to extract item features about the target contour based on the image data, and extract geometric features of the target contour based on the target contour; the first recognition module 31 further includes a feature fusion module (not shown), used to obtain descriptive information of the target contour based on the item features and geometric features of the target contour.

[0070] In some disclosed embodiments, the prior knowledge base in the map knowledge base construction apparatus 30 further includes: a reference item type used for area positioning within the target area during map updating; and / or, the map knowledge base construction apparatus 30 further includes a prompt information generation module (not shown) for generating prompt information to prompt the target object to correct or determine the prior knowledge base.

[0071] Please see Figure 4 , Figure 4This is a schematic diagram of the framework of an embodiment of the map updating device of this application. The map updating device 40 includes: a map acquisition module 41, a second recognition module 42, a contour change module 43, and a strategy generation module 44. The map acquisition module 41 is used to acquire the current scan map collected by the target device when it is running in the target area; the second recognition module 42 is used to perform contour recognition on the current scan map to obtain the contours of several items in the target scan map and the descriptive information of the item contours; the contour change module 43 is used to obtain the movement attributes of the item contours based on the prior knowledge base of the target area and the descriptive information of the item contours, and to obtain the contour change between the current scan map and the global scan map of the target area; wherein, the prior knowledge base of the target area is obtained based on the map knowledge base construction device 30 in the aforementioned embodiment; the strategy generation module 44 is used to generate a target strategy based on the descriptive information of the item contours, the movement attributes of the item contours, the contour change, and the prior knowledge base; wherein, the target strategy is used to indicate the update strategy of at least one of the global scan map and the prior knowledge base.

[0072] In the above scheme, during the map update process, the map update device 40 acquires the current scan map collected by the target device while it is running in the target area, performs contour recognition on the current scan map to obtain the contours of several items and their descriptions within the target scan map, obtains the movement attributes of the item contours based on the prior knowledge base of the target area and the descriptions of the item contours, and acquires the contour changes between the current scan map and the global scan map of the target area. Based on the descriptions of the item contours, their movement attributes, the contour changes, and the prior knowledge base, a target strategy is generated to indicate the update strategy of at least one of the global scan map and the prior knowledge base. On the one hand, based on the analysis of the contour information within the global scan map of the target area, preset scene data matching the target area is obtained. Utilizing the prior knowledge provided by this preset scene data, the movement characteristics of objects corresponding to the target contours within the target area are determined, distinguishing between short-term fluctuations and long-term changes of different objects in the target area. This constructs a more targeted prior knowledge base, providing a reliable data foundation and decision-making basis for map updates. On the other hand, by integrating the descriptive information, movement attributes, and contour changes of object contours, key data that needs to be updated to the map can be selected, thereby improving the accuracy of map update strategy generation and enhancing the applicability of the updated map. Therefore, it can improve the accuracy of map update strategy generation.

[0073] In some disclosed embodiments, the strategy generation module 44 further includes a second instruction construction module (not shown), used to construct a second prompt instruction based on the description information of the item outline, the movement attribute of the item outline, the outline change status, and the prior knowledge base; wherein, the second prompt instruction is used to instruct the second large model to determine the data whose movement attribute is long-term static in the global scan map and the prior knowledge base to be updated; the strategy generation module 44 further includes a second output module (not shown), used to obtain the output data of the second large model in response to the second prompt instruction, as the target strategy.

[0074] In some disclosed embodiments, when the target device collects a sub-region of the target area corresponding to the current scan map, the contour change module 43 further includes a spatial range determination module (not shown), used to determine a local scan map in the global scan map that is consistent with the current scan map area, and to determine the spatial range of the item contour based on the circumscribed graphic of the item contour; the contour change module 43 further includes a corresponding region determination module (not shown), used to determine the corresponding region of the item contour in the local scan map based on the spatial range of the item contour and the target contour in the local scan map; the contour change module 43 further includes a change amount calculation module (not shown), used to calculate the contour change amount between the item contour and the target contour based on the corresponding region and the spatial range of the target contour in the local scan map; the contour change module 43 further includes a change determination module (not shown), used to determine the contour change between the current scan map and the global scan map based on the contour change amount.

[0075] Please see Figure 5 , Figure 5 This is a schematic diagram of a framework of an embodiment of the electronic device of this application. The electronic device 50 includes at least a memory 51 and a processor 52 coupled to each other. The memory 51 stores at least program instructions, and the processor 52 is used to execute the program instructions to implement any of the map knowledge base construction methods or map update method embodiments described above. For details, please refer to the foregoing disclosed embodiments, which will not be repeated here. Furthermore, as a possible example, the electronic device 50 may also include a display screen (not shown), which can be used to input configuration instructions, etc. Of course, the electronic device 50 may also include other electronic components, which are not limited here.

[0076] Specifically, processor 52 controls itself and memory 51 to implement any of the steps in the map knowledge base construction method or map update method embodiments described above. Processor 52 can also be called a CPU (Central Processing Unit). Processor 52 may be an integrated circuit chip with signal processing capabilities. Processor 52 can also be a general-purpose processor, digital signal processor (DSP), application-specific integrated circuit (ASIC), field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. A general-purpose processor can be a microprocessor or any conventional processor. Furthermore, processor 52 can be implemented using integrated circuit chips.

[0077] In the above scheme, electronic device 50 performs contour recognition based on a global scan map of the target area, obtaining several target contours and their descriptive information within the global scan map. Based on the relative distribution relationship between different target contours and their descriptive information, it selects existing scene data matching the target area as target scene data. The existing scene data includes at least preset items within a preset scene, their contour descriptions, the relative distribution relationship between preset items, and their movement attributes. Movement attributes characterize the spatial positional changes of the items. Based on the preset scene data, the movement attributes of each target contour are obtained. Based on several target contours, their descriptive information, their movement attributes, and the target... Based on the relative distribution relationships between contours and preset scene data, a prior knowledge base for the target area is constructed. During map updates, the current scan map collected by the target device while running in the target area is acquired. Contour recognition is performed on the current scan map to obtain the contours of several items and their descriptions within the target scan map. Based on the prior knowledge base of the target area and the descriptions of the item contours, the movement attributes of the item contours are obtained, and the contour changes between the current scan map and the global scan map of the target area are acquired. Based on the descriptions of the item contours, their movement attributes, the contour changes, and the prior knowledge base, a target strategy is generated to indicate the update strategy of at least one of the global scan map and the prior knowledge base. On the one hand, based on the analysis of the contour information within the global scan map of the target area, preset scene data matching the target area is obtained. Utilizing the prior knowledge provided by this preset scene data, the movement characteristics of objects corresponding to the target contours within the target area are determined, distinguishing between short-term fluctuations and long-term changes of different objects in the target area. This constructs a more targeted prior knowledge base, providing a reliable data foundation and decision-making basis for map updates. On the other hand, by integrating the descriptive information, movement attributes, and contour changes of object contours, key data that needs to be updated to the map can be selected, thereby improving the accuracy of map update strategy generation and enhancing the applicability of the updated map. Therefore, it can improve the accuracy of map update strategy generation.

[0078] Please see Figure 6 , Figure 6 This is a schematic diagram of a framework of an embodiment of the computer-readable storage medium of this application. The computer-readable storage medium 60 stores program instructions 61 that can be executed by a processor. The program instructions 61 are used to implement the steps in any of the above-described map knowledge base construction methods or map update method embodiments.

[0079] In the above scheme, the computer-readable storage medium 60 performs contour recognition based on a global scan map of the target area, obtaining several target contours and their descriptive information within the global scan map. Based on the relative distribution relationship between different target contours and their descriptive information, existing scene data matching the target area is selected as the target scene data. The existing scene data includes at least preset items within a preset scene, their contour descriptions, the relative distribution relationship between the preset items, and their movement attributes. Movement attributes characterize the spatial positional changes of the items. Based on the preset scene data, the movement attributes of each target contour are obtained. Based on the several target contours, their descriptive information, and their movement attributes... Based on the relative distribution relationships between target contours and preset scene data, a prior knowledge base for the target area is constructed. During map updates, the current scan map collected by the target device while running in the target area is acquired. Contour recognition is performed on the current scan map to obtain the contours of several items and their descriptions within the target scan map. Based on the prior knowledge base of the target area and the descriptions of the item contours, the movement attributes of the item contours are obtained, and the contour changes between the current scan map and the global scan map of the target area are acquired. Based on the descriptions of the item contours, their movement attributes, the contour changes, and the prior knowledge base, a target strategy is generated to indicate the update strategy of at least one of the global scan map and the prior knowledge base. On the one hand, based on the analysis of the contour information within the global scan map of the target area, preset scene data matching the target area is obtained. Utilizing the prior knowledge provided by this preset scene data, the movement characteristics of objects corresponding to the target contours within the target area are determined, distinguishing between short-term fluctuations and long-term changes of different objects in the target area. This constructs a more targeted prior knowledge base, providing a reliable data foundation and decision-making basis for map updates. On the other hand, by integrating the descriptive information, movement attributes, and contour changes of object contours, key data that needs to be updated to the map can be selected, thereby improving the accuracy of map update strategy generation and enhancing the applicability of the updated map. Therefore, it can improve the accuracy of map update strategy generation.

[0080] In some embodiments, the functions or modules of the apparatus provided in this disclosure can be used to perform the methods described in the above method embodiments. The specific implementation can be referred to the description of the above method embodiments, and for the sake of brevity, it will not be repeated here.

[0081] The description of the various embodiments above tends to emphasize the differences between the various embodiments. The similarities or similarities between them can be referred to, and for the sake of brevity, they will not be repeated here.

[0082] In the several embodiments provided in this application, it should be understood that the disclosed methods and apparatus can be implemented in other ways. For example, the apparatus implementations described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0083] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0084] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0085] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods of various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0086] If the technical solution of this application involves personal information, the product using this technical solution has clearly informed the user of the personal information processing rules and obtained the user's voluntary consent before processing the personal information. If the technical solution of this application involves sensitive personal information, the product using this technical solution has obtained the user's separate consent before processing the sensitive personal information, and also meets the requirement of "express consent". For example, at personal information collection devices such as cameras, clear and prominent signs are set up to inform users that they have entered the scope of personal information collection and that personal information will be collected. If an individual voluntarily enters the collection scope, it is deemed that they have agreed to the collection of their personal information; or on the personal information processing device, with clear signs / information informing users of the personal information processing rules, authorization is obtained from the individual through pop-up information or by asking the individual to upload their personal information; wherein, the personal information processing rules may include information such as the personal information processor, the purpose of personal information processing, the processing method, and the types of personal information processed.

Claims

1. A method of constructing a map knowledge base, characterized by, The method comprises: performing contour recognition on a global scanning map of a target region to obtain a plurality of target contours in the global scanning map and description information of the target contours; selecting existing scene data matching the target region as target scene data based on relative distribution relationships between different target contours and the description information of the target contours; wherein the existing scene data at least comprises preset articles in a preset scene, contour description information of the preset articles, relative distribution relationships between the preset articles, and movement attributes of the preset articles, the movement attributes representing variation characteristics of spatial positions of the articles; obtaining movement attributes of each target contour based on the preset scene data; constructing a prior knowledge base of the target region based on the plurality of target contours, the description information of the target contours, the movement attributes of the target contours, the relative distribution relationships between the target contours, and the preset scene data.

2. The method of claim 1, wherein, The method of selecting existing scene data matching the target region as target scene data based on relative distribution relationships between different target contours and the description information of the target contours comprises: dividing the global scanning map to obtain a plurality of sub-regions; constructing a first prompt instruction based on relative distribution relationships between different target contours in the sub-regions and the description information of the target contours in the sub-regions; wherein the first prompt instruction is used to instruct a first large model to select existing scene data satisfying a similarity condition based on a matching similarity between the existing scene data and the sub-regions; obtaining output data of the first large model in response to the first prompt instruction as the preset scene data.

3. The method of claim 2, wherein, In a case where the first large model detects that there is no existing scene data satisfying the similarity condition with a target sub-region, the method further comprises: obtaining description information of the target contours in at least one candidate sub-region as first description information, and obtaining description information of target contours in the target sub-region as second description information; wherein the target sub-region is one of the plurality of sub-regions, the candidate sub-region is a sub-region adjacent to the target sub-region in the global scanning map, and there is existing scene data satisfying the similarity condition with the candidate sub-region; matching movement attributes for the target contours in the target sub-region based on a similarity result between the first description information and the second description information and the existing scene data matching the candidate sub-region.

4. The method of claim 1, wherein, The method of performing contour recognition on a global scanning map of a target region to obtain a plurality of target contours in the global scanning map and description information of the target contours comprises: obtaining image data about the target region, and performing contour segmentation based on the global scanning map to obtain the target contours respectively belonging to each article; extracting article features about the target contours based on the image data, and extracting geometric features of the target contours based on the target contours; Obtaining description information of the target contour based on the object feature and the geometric feature of the target contour.

5. The method of claim 1, wherein, The prior knowledge base further comprises: reference object types used for area positioning in the target area in a map updating process; And / or, the method further comprises: generating prompt information for prompting a target object to correct or determine the prior knowledge base.

6. A map update method characterized by comprising: Comprising: Obtaining a current scan map collected when a target device runs in a target area; Performing contour recognition on the current scan map to obtain a plurality of object contours in the target scan map and description information of the object contours; Based on the prior knowledge base of the target area and the description information of the object contours, obtaining the movement attribute of the object contour, and obtaining the contour change between the current scan map and a global scan map of the target area; wherein the prior knowledge base of the target area is obtained based on the map knowledge base construction method of any one of claims 1 to 5; Based on the description information of the object contour, the movement attribute of the object contour, the contour change and the prior knowledge base, a target strategy is generated; wherein the target strategy is used to indicate the update strategy of at least one of the global scan map and the prior knowledge base.

7. The method of claim 6, wherein, The target strategy is generated based on the description information of the object contour, the movement attribute of the object contour, the contour change and the prior knowledge base, comprising: Based on the description information of the object contour, the movement attribute of the object contour, the contour change and the prior knowledge base, a second prompt instruction is constructed; wherein the second prompt instruction is used to instruct a second large model to determine data with a long-term stationary movement attribute to be updated to the global scan map and the prior knowledge base; The output data of the second large model in response to the second prompt instruction is obtained as the target strategy.

8. The method of claim 6, wherein, In the case that the current scan map collected by the target device corresponds to a sub-area of the target area, the contour change between the current scan map and the global scan map of the target area is obtained, comprising: Determining a local scan map consistent with the region of the current scan map in the global scan map, and determining the spatial range of the object contour based on the circumscribed figure of the object contour; Based on the spatial range of the object contour and the target contour in the local scan map, determining the corresponding region of the object contour in the local scan map; Based on the corresponding region and the spatial range of the target contour in the local scan map, the contour change amount between the object contour and the target contour is calculated; Based on the contour change amount, the contour change between the current scan map and the global scan map is determined.

9. A map knowledge base construction apparatus characterized by comprising: Comprising: A first recognition module is configured to perform contour recognition based on a global scan map of a target area to obtain a plurality of target contours in the global scan map and description information of the target contours; The scene selection module is configured to select existing scene data matching the target region as target scene data based on the relative distribution relationship between different target contours and the description information of the target contours; wherein the existing scene data at least includes preset articles in a preset scene, contour description information of the preset articles, a relative distribution relationship between the preset articles, and movement attributes of the preset articles, the movement attributes representing variation characteristics of spatial positions of the articles; The attribute determination module is configured to obtain movement attributes of each target contour based on the preset scene data; The knowledge base construction module is configured to construct a prior knowledge base of the target region based on the target contours, the description information of the target contours, the movement attributes of the target contours, the relative distribution relationship between the target contours, and the preset scene data.

10. A map update device characterized by comprising: The map acquisition module is configured to acquire a current scanning map collected when a target device runs in a target region; The second recognition module is configured to perform contour recognition on the current scanning map to obtain a plurality of article contours in the target scanning map and description information of the article contours; The contour change module is configured to obtain movement attributes of the article contours based on the prior knowledge base of the target region and the description information of the article contours, and to acquire a contour change condition between the current scanning map and a global scanning map of the target region; wherein the prior knowledge base of the target region is obtained based on the map knowledge base construction apparatus of claim 9; The strategy generation module is configured to generate a target strategy based on the description information of the article contours, the movement attributes of the article contours, the contour change condition, and the prior knowledge base; wherein the target strategy is used to indicate an update strategy of at least one of the global scanning map and the prior knowledge base. The memory at least stores program instructions, and the processor is configured to execute the program instructions to implement the map knowledge base construction method of any one of claims 1 to 5, or to implement the map update method of any one of claims 6 to 8.

11. An electronic device, comprising: The memory stores program instructions capable of being executed by the processor, and the program instructions are used to implement the map knowledge base construction method of any one of claims 1 to 5, or to implement the map update method of any one of claims 6 to 8.

12. A computer-readable storage medium, characterized in that, ​

Images