Map updating method and device, equipment, storage medium and program product

Abstract

The invention provides a map updating method and device, equipment, a storage medium and a program product, and relates to the technical field of electronic maps, the method comprises the following steps: obtaining a sensing area associated with construction record information, and setting a plurality of candidate sampling points in the sensing area; according to the first image data, the candidate sampling points and flight data of the low-altitude cruise equipment, a flight path of the low-altitude cruise equipment is determined, and the flight path comprises a plurality of target sampling points; the first image data is acquired by the low-altitude cruise equipment on candidate sampling points; the flight data are flight data of the low-altitude cruise equipment on the candidate sampling points; updating a map layer corresponding to the sensing area in the first map according to the second image data and inertial measurement unit (IMU) data when the low-altitude cruise equipment collects the second image data; and the second image data is acquired on the target sampling point by the low-altitude cruise equipment in the flight process according to the flight path, so that low-cost, real-time and accurate map updating is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electronic maps, and particularly to a map updating method and device, equipment, a storage medium and a program product. BACKGROUND

[0002] Under the background of rapid development of vehicle-road-cloud integration and intelligent auxiliary driving, temporary traffic events such as road construction have put forward higher real-time and accuracy requirements for high-precision map updating. The current intelligent driving is insufficient in identifying obstacles such as cones and barrels, and water barriers, and accidents are easily caused in highway scenes due to temporary road construction and detours. At the same time, the collection of road construction information relies on manual filing and manual modeling, and there are problems such as perception lag, insufficient accuracy, complicated issuing process, and untimely map updating. SUMMARY

[0003] Embodiments of the present application provide a map updating method, device, equipment, storage medium and program product, which solve the problems of untimely and insufficient accuracy of current map updating.

[0004] In a first aspect, to achieve the above object, the embodiments of the present application provide a map updating method, comprising:

[0005] obtaining a perception area associated with construction filing information, a plurality of candidate sampling points being arranged in the perception area;

[0006] determining a flight path of a low-altitude cruise device according to first image data, the candidate sampling points and flight data of the low-altitude cruise device, wherein the flight path comprises a plurality of target sampling points; the first image data is collected by the low-altitude cruise device at the candidate sampling points; and the flight data is flight data of the low-altitude cruise device at the candidate sampling points;

[0007] updating a map layer corresponding to the perception area in a first map according to second image data and inertial measurement unit (IMU) data of the low-altitude cruise device when collecting the second image data, wherein the second image data is collected by the low-altitude cruise device at the target sampling points in the process of flying according to the flight path.

[0008] wherein obtaining the perception area associated with the construction filing information comprises:

[0009] obtaining a construction center point and a construction direction according to the construction filing information;

[0010] performing semantic modeling and area division of the work area according to the construction center point and the construction direction to obtain a plurality of work sections, a semantic label of each work section and a sampling priority weight of each work section;

[0011] candidate sampling points are set in each of the work sections according to the sampling priority weight; wherein the interval of the candidate sampling points in different work sections is negatively correlated with the sampling priority weight;

[0012] The perception area is obtained according to each of the work sections and the candidate sampling points.

[0013] The flight path of the low-altitude cruise device is determined according to the first image data, the candidate sampling points and flight data of the low-altitude cruise device, comprising:

[0014] The structure perception score of each of the candidate sampling points is determined according to the flight data and the candidate sampling points;

[0015] The target candidate sampling point is screened from the candidate sampling points, wherein the structure perception score of the target candidate sampling point is greater than or equal to a first threshold value;

[0016] The target sampling point is selected from the target candidate sampling points according to the first image data corresponding to the target candidate sampling point and the structure perception score;

[0017] The flight path is determined according to the target sampling point.

[0018] The structure perception score of each of the candidate sampling points is determined according to the flight data and the candidate sampling points, comprising:

[0019] The relative height adaptation degree corresponding to the candidate sampling point is determined according to the flight height in the flight data and the recommended height corresponding to the work section where the candidate sampling point is located; wherein the perception area comprises a plurality of work sections;

[0020] The edge recognition accuracy ability corresponding to the candidate sampling point is determined according to the perspective direction of the candidate sampling point and the arrangement direction of the construction facility;

[0021] The occlusion probability corresponding to the candidate sampling point is determined according to the pre-established occlusion model, the field of view angle of the camera and the pose data in the flight data; wherein the field of view angle of the camera is the field of view angle of the camera on the low-altitude cruise device at the candidate sampling point;

[0022] The structure perception score is determined according to the relative height adaptation degree, the edge recognition accuracy ability and the occlusion probability.

[0023] The target sampling point is selected from the target candidate sampling points according to the first image data corresponding to the target candidate sampling point and the structure perception score, comprising:

[0024] For each of the target candidate sampling points, obtain a heat map of various construction facilities corresponding to the work area where the target candidate sampling point is located, obtain the facility weights corresponding to each type of construction facility, and obtain the field of view range of the low-altitude cruise equipment at the target candidate sampling point;

[0025] Based on the field of view, the heat map, and the facility weight, a first estimated value is determined for the target candidate sampling point, wherein the first estimated value represents the visibility of the construction facility at the target candidate sampling point;

[0026] Based on the field of view and the boundary information of the perception area in the first map, a second estimated value corresponding to the target candidate sampling point is determined, wherein the second estimated value represents the gain estimate of the target candidate sampling point to the boundary information;

[0027] The viewpoint score of the target sampling point is determined based on the structure perception score, the first estimated value, the second estimated value, and the sampling priority weight of the work area where the target sampling point is located.

[0028] The target sampling point is selected based on the first image data and the viewpoint scores of each of the target candidate sampling points.

[0029] The selection of the target sampling points based on the first image data and the viewpoint scores of each of the target candidate sampling points includes:

[0030] Based on the aforementioned perspective score, the target candidate sampling points are sorted in descending order;

[0031] The first image data corresponding to at least two consecutive target candidate sampling points in the sorting results are compared to obtain the comparison results;

[0032] Based on the comparison results, multiple first image data with overlapping viewpoints greater than the second threshold are obtained from the first image data;

[0033] Among the target candidate sampling points corresponding to the plurality of first image data, the target candidate sampling point with the highest viewpoint score is retained;

[0034] The first K target candidate sampling points from the remaining target candidate sampling points are taken as the target sampling points; where K is an integer.

[0035] Specifically, updating the map layer corresponding to the sensing area in the first map based on the second image data and the inertial measurement unit (IMU) data collected by the low-altitude cruise device during the acquisition of the second image data includes:

[0036] Identify construction facilities in the second image data;

[0037] Based on the IMU data, the identified construction facilities are projected onto a three-dimensional world coordinate system to obtain the location point of the construction facilities in the ground coordinate system;

[0038] The location points are projected onto a two-dimensional coordinate plane to obtain a two-dimensional point set, and the boundary of the sensing area is fitted in the two-dimensional coordinate plane to generate map data corresponding to the sensing area.

[0039] Update the map layer of the perceived area in the first map based on the map data.

[0040] The method further includes:

[0041] Acquire third image data, wherein the third image data is collected by the low-altitude cruise device at the target sampling point during flight along the flight path;

[0042] If construction is determined to be completed based on the third image data, the update of the map layer corresponding to the perceived area in the first map is cancelled.

[0043] Secondly, to achieve the above objectives, embodiments of this application provide a map updating apparatus, comprising:

[0044] The first acquisition module is used to acquire the sensing area associated with the construction filing information, and multiple candidate sampling points are set in the sensing area;

[0045] The determining module is used to determine the flight path of the low-altitude cruise device based on the first image data, the candidate sampling points, and the flight data of the low-altitude cruise device, wherein the flight path includes multiple target sampling points; the first image data is collected by the low-altitude cruise device at the candidate sampling points; and the flight data is the flight data of the low-altitude cruise device at the candidate sampling points.

[0046] The first update module is used to update the map layer corresponding to the perception area in the first map based on the second image data and the inertial measurement unit (IMU) data of the low-altitude cruise device when collecting the second image data; wherein the second image data is collected by the low-altitude cruise device at the target sampling point during flight along the flight path.

[0047] Thirdly, to achieve the above objectives, embodiments of this application provide a map updating device, including a transceiver, a processor, a memory, and a program stored in the memory and executable on the processor; the transceiver transmits and receives data under the control of the processor, and the processor executes the program to implement the map updating method as described in the first aspect.

[0048] Fourthly, to achieve the above objectives, embodiments of this application provide a readable storage medium having a program or instructions stored thereon, which, when executed by a processor, implement the map update method as described in the first aspect.

[0049] Fifthly, to achieve the above objectives, embodiments of this application provide a computer program product including computer instructions that, when executed by a processor, implement the map update method as described in the first aspect.

[0050] The beneficial effects of the above technical solution in this application are as follows:

[0051] In the embodiments of this application, firstly, a sensing area associated with construction registration information is acquired, and multiple candidate sampling points are set within the sensing area; secondly, based on first image data, the candidate sampling points, and flight data of the low-altitude cruise device, the flight path of the low-altitude cruise device is determined, wherein the flight path includes multiple target sampling points; the first image data is collected by the low-altitude cruise device at the candidate sampling points; the flight data is the flight data of the low-altitude cruise device at the candidate sampling points; thirdly, based on second image data and the inertial measurement unit (IMU) data of the low-altitude cruise device when collecting the second image data, the map layer corresponding to the sensing area in the first map is updated; wherein the second image data is collected by the low-altitude cruise device at the target sampling points during flight along the flight path. Thus, a fast, accurate, low-cost, and universally applicable update of the map for temporary traffic events related to construction is achieved, ensuring that the updated content is compatible with the construction map. Attached Figure Description

[0052] Figure 1 This is one of the flowcharts illustrating the map update method according to an embodiment of this application;

[0053] Figure 2 This is a schematic diagram of the deployment of construction facilities in an embodiment of this application;

[0054] Figure 3 This is a second schematic flowchart of the map update method according to an embodiment of this application;

[0055] Figure 4 The structure of the map updating device according to an embodiment of this application;

[0056] Figure 5 This is a schematic diagram of the structure of a map update device according to an embodiment of this application. Detailed Implementation

[0057] To make the technical problems, technical solutions and advantages of this application clearer, a detailed description will be provided below in conjunction with the accompanying drawings and specific embodiments.

[0058] It should be understood that the phrase "one embodiment" or "an embodiment" throughout the specification means that a specific feature, structure, or characteristic related to the embodiment is included in at least one embodiment of this application. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments.

[0059] In the various embodiments of this application, it should be understood that the sequence number of each process described below does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0060] In addition, the terms "system" and "network" are often used interchangeably in this article.

[0061] In the embodiments provided in this application, it should be understood that "B corresponding to A" means that B is associated with A, and B can be determined based on A. However, it should also be understood that determining B based on A does not mean determining B solely based on A; B can also be determined based on A and / or other information.

[0062] Before describing the embodiments of this application, the relevant technical points will be explained first:

[0063] With the continuous development of intelligent connected vehicles and vehicle-road-cloud integration, high-precision maps, as a crucial support for perception, decision-making, and path planning, are gradually evolving towards real-time and dynamic capabilities. Current autonomous driving solutions rely on high-precision maps to obtain current road and environmental information. If the map does not match the actual road conditions, it can be extremely dangerous for autonomous driving. Therefore, to create a safe driving environment, real-time, accurate, and rapid updates to the maps relied upon by autonomous vehicles are essential. Especially in scenarios involving highways and urban arterial roads, temporary traffic events such as construction and closures pose new challenges to the accuracy and update frequency of map information. If the map does not match the actual situation, it can easily lead to a downgrade in intelligent driving, a lack of vehicle-side warnings, or even traffic accidents.

[0064] On highways, road maintenance work often involves temporary road closures and detours. According to relevant operating procedures, lane channelization facilities typically consist of traffic cones, crash barriers, and water-filled barriers. However, current intelligent driving systems often lack sufficient recognition of these obstacles, potentially leading to safety hazards. Therefore, timely scanning and mapping of highway maintenance work areas is essential for ensuring driving safety. Currently, map updates for road construction information still rely primarily on the traditional process of "manual filing + back-end mapping + manual review." This method is lengthy, slow, and particularly ineffective for responding to short-term construction events.

[0065] Currently, various related technologies have proposed solutions to the problem of temporary map changes, but these solutions still have certain limitations in practical applications.

[0066] 1) When abnormal scenarios occur, various types of abnormal scenario sensing data are acquired, and the minimum safety boundary is calculated based on this data. Finally, the map is updated based on the calculated minimum safety boundary. This solution relies on the vehicle's own sensor capabilities and multimodal algorithms, and the newly generated map data is difficult to share with other brands of vehicles, thus failing to guarantee the safety of all passing vehicles.

[0067] 2) Acquire roadside sensing data for the current road segment and track and detect road targets based on this data. Determine the movement trajectory of the road targets and whether it matches a preset trajectory based on the tracking and detection results. If it does, determine the road target behavior data based on the movement trajectory and send it to the cloud, enabling the cloud to update the high-precision map of the current road segment based on this data. This solution only supports map updates in areas where sensing devices are deployed on the roadside and cannot be adapted to other road segments.

[0068] It is evident that the relevant technologies have significant limitations in terms of the sources of perception and the scope of application for map updates, making it difficult to meet the requirements of "fast, accurate, low-cost, and universal" map updates for temporary traffic events such as construction. Specifically, the above-mentioned scheme 1) relies on the vehicle's own sensor system (such as cameras, millimeter-wave radar, and lidar) to perceive abnormal scenes and calculates the minimum safety boundary through multimodal sensing algorithms to update the map. This method requires vehicles to have high local perception and computing capabilities, suitable for high-level autonomous vehicles, but lacks universality for mid-to-low-end models or different brands of vehicles. Furthermore, the newly generated map data is difficult to standardize and share, failing to ensure that all passing vehicles have a consistent map understanding of the same construction event, resulting in information silos. Scheme 2) utilizes roadside perception devices (such as cameras and radar) to continuously acquire the movement trajectory of traffic targets and extract target behavior data through trajectory stability analysis, thereby inferring scene changes and updating the map. This scheme is highly dependent on the perception area and is only applicable to road sections where high-performance roadside perception devices have been deployed, unable to cover the vast areas where such devices have not yet been deployed. Meanwhile, this type of method is based on the perception of the historical movement trajectory of vehicles by roadside equipment, rather than directly modeling the boundary information of static spatial structures such as construction areas, resulting in limited content updates and difficulty in adapting to the needs of construction maps.

[0069] To address the aforementioned issues, this application proposes a UAV perception module based on the fusion of a monocular camera and an inertial measurement unit (IMU), combined with a gain-driven path sampling algorithm featuring a "locally dense + globally sparse" structure, to construct a closed-loop processing system for road construction information, applicable to various open-space road construction scenarios. Upon receiving construction registration information, the system automatically dispatches flight missions, scheduling a lightweight UAV equipped with a monocular camera and IMU to the construction area. Based on multi-view sampling and information gain evaluation, key shooting points are selected to achieve spatial positioning and boundary fitting of construction facilities (such as cones, water-filled barriers, and fences). The generated construction boundaries are encapsulated in a structured high-precision map format, which can be distributed to RSU broadcasts, displayed on guidance screens, or synchronized with navigation map providers. After construction is completed, the system conducts a secondary verification flight, automatically revoking the map through target recognition and boundary comparison, achieving a closed-loop update of the construction status.

[0070] The embodiments of this application utilize a low-cost perception module to achieve accurate perception and efficient mapping of construction areas, improving the real-time performance, accuracy, and automation of construction map generation. Combined with an information gain-driven flight path planning algorithm, scanning efficiency and boundary coverage are significantly improved, avoiding invalid paths and omissions of critical areas. Using drones for scanning and mapping saves labor costs and allows for timely updates to temporary routes in construction areas. This method is applicable to various open-space construction scenarios, including highways and urban arterial roads, and effectively supports integrated vehicle-road-cloud systems and intelligent driving platforms in rapidly perceiving and making decisions regarding temporary traffic conditions, thereby improving overall road operation safety and efficiency.

[0071] The implementation process of the embodiments of this application will be described in detail below.

[0072] like Figure 1 As shown, an embodiment of this application provides a map updating method, the method comprising:

[0073] Step 101: Obtain the sensing area associated with the construction filing information. Multiple candidate sampling points are set within the sensing area. For example, the construction filing information includes the location coordinates of the highway maintenance work area, the construction scope, the expected start and end times, the construction unit, etc. The sensing area is related to the highway maintenance work area. For example, the sensing area includes the highway maintenance work area and the surrounding area. The candidate sampling points are non-uniform grid sampling points. For example, the closer the area is to the highway maintenance work area, the higher the sampling point density.

[0074] Step 102: Based on the first image data, the candidate sampling points, and the flight data of the low-altitude cruise device, determine the flight path of the low-altitude cruise device, wherein the flight path includes multiple target sampling points; the first image data is acquired by the low-altitude cruise device at the candidate sampling points; the flight data is the flight data of the low-altitude cruise device at the candidate sampling points, for example, the flight altitude of the low-altitude cruise device at the candidate sampling points, the angle between the shooting direction of the low-altitude cruise device at the candidate sampling points and the arrangement direction of the construction facilities, etc. Figure 2 Taking a construction scenario as an example, the construction facilities set up in a temporary road diversion construction scenario can include: temporary markings, fences, signs, etc.

[0075] For example, the low-altitude cruise device is a drone; the target sampling point is at least some of the candidate sampling points; and the flight path is the path through which the low-altitude cruise device collects image data of the perceived area for updating the map.

[0076] Step 102 above is the execution subject of this application embodiment (e.g., a cloud platform, edge computing device, low-altitude cruise device, low-altitude cruise device nest, specifically, a flight path planning module deployed on the aforementioned device or platform). Based on the first image data collected by the low-altitude cruise device at each candidate sampling point and the flight data of the low-altitude cruise device at each candidate sampling point, the target sampling point is selected from the candidate sampling points to generate a flight path based on the target sampling point.

[0077] Step 103: Update the map layer corresponding to the perception area in the first map based on the second image data and the inertial measurement unit (IMU) data of the low-altitude cruise device when collecting the second image data; wherein, the second image data is collected by the low-altitude cruise device at the target sampling point during flight along the flight path.

[0078] Specifically, step 103 above involves generating image data corresponding to the sensing area based on the second image data and IMU data, and updating the map layer in the first map corresponding to the sensing area based on this data. For example, the second image data is overlaid on the layer in the first map corresponding to the sensing area.

[0079] In the map update method of this application embodiment, firstly, a perception area associated with construction filing information is obtained, and multiple candidate sampling points are set within the perception area; secondly, the flight path of the low-altitude cruise device is determined based on first image data, the candidate sampling points, and flight data of the low-altitude cruise device, wherein the flight path includes multiple target sampling points; the first image data is collected by the low-altitude cruise device at the candidate sampling points; the flight data is the flight data of the low-altitude cruise device at the candidate sampling points; thus, the low-altitude cruise device can subsequently collect image data at the target sampling points according to the flight path, without relying on the vehicle's own sensor capabilities and multimodal algorithms, nor needing to deploy perception devices on the roadside to perceive the road scene, solving the problem that the perception sources and applicable scope of map updates in related technologies are quite limited; finally, the map layer corresponding to the perception area in the first map is updated based on second image data and the inertial measurement unit (IMU) data of the low-altitude cruise device when collecting the second image data; wherein the second image data is collected by the low-altitude cruise device at the target sampling points during flight according to the flight path. In this way, it is possible to update the map quickly, accurately, at low cost, and universally in response to temporary traffic incidents caused by construction, so that the updated content can be adapted to the construction map.

[0080] As an optional implementation, step 101 above, obtaining the sensing area associated with the construction filing information, includes:

[0081] Step 1: Obtain the construction center point and construction direction based on the construction filing information.

[0082] Step 2: Perform semantic modeling and region division of the work area based on the construction center point and construction direction to obtain multiple work segments, semantic labels for each work segment, and sampling priority weights for each work segment. For example, this step can be: based on the construction center point and construction direction in the construction filing information, divide the work area into six segments according to the specified distance. These six segments are, in order: Warning Area → Upstream Transition Area → Buffer Zone (Horizontal, Vertical) → Work Area → Downstream Transition Area → Termination Area. Each segment can be defined as a spatial polygon S. j And with the semantic tag ZONEtype(S) j ) and sampling priority weight W j The weights are set as follows: the working area has the highest weight, followed by the buffer, transition area, warning area, and termination area, which serve as the basis for subsequent view sampling and gain scoring.

[0083] Step 3: Based on the sampling priority weight, set candidate sampling points in each of the work sections; wherein, the spacing between the candidate sampling points in different work sections is negatively correlated with the sampling priority weight; that is, the higher the sampling priority weight, the smaller the spacing between candidate sampling points, i.e., the greater the density of candidate sampling points. Specifically: in each work section S... j Within this area, different sampling intervals are set according to their priority. For example, denser sampling intervals are set in the working area and buffer zone, while sparser sampling intervals are allocated in the transition zone and warning zone. In other words, the candidate sampling points in this step are sampled using a non-uniform grid. This results in a high sampling point density in important work areas and a low sampling density in non-important work areas, which enables the planning results for the sensing area to have significant content relevance and task adaptability.

[0084] For example, the relevant information of the candidate sampling point includes the spatial coordinates of the candidate sampling point, the viewing direction of the image data collected at the candidate sampling point, and the label of the work segment to which the candidate sampling point belongs. The viewing direction, by default, points to the direction of maximum target density.

[0085] Step 4: Obtain the sensing area based on each of the work sections and the candidate sampling points; for example, the sensing area consists of multiple work sections, and each work section is provided with candidate sampling points set according to Step 3 above.

[0086] It should be noted that existing technologies typically rely on regular grids, preset waypoints, or navigation path generation methods for UAV flight. Path generation is independent of the mapping content, depending only on the scene shape or mission start and end points, and the generated grid is also fixed. However, among the aforementioned optional implementations, for the specific scenario of road construction, regional division is combined with six semantic segments of highway construction. The sampling density and priority are automatically adjusted based on the semantics, resulting in denser candidate sampling points in key areas and sparser candidate sampling points in secondary areas. This ensures that the target sampling points selected from the candidate sampling points in key areas are dense, while those selected from the candidate sampling points in secondary areas are sparse. This approach achieves relevance and task adaptability to the specific scenario of road construction, improving the accuracy of the updated map.

[0087] As an optional implementation, step 102, determining the flight path of the low-altitude cruise device based on the first image data, the candidate sampling points, and the flight data of the low-altitude cruise device, includes:

[0088] Step 1: Based on the flight data and the candidate sampling points, determine the structure-aware score for each candidate sampling point. The structure-aware score represents the ability of a candidate sampling point to express image quality. By designing the structure-aware score, the adaptability of the shooting perspective to map building can be assessed in advance at the sampling point level. For example, the flight data refers to the flight data of the low-altitude cruise equipment at the candidate sampling points, including, for example, flight altitude, pose data, and viewing direction.

[0089] Step 2: Among the candidate sampling points, target candidate sampling points are selected, wherein the structure-aware score of the target candidate sampling points is greater than or equal to a first threshold; that is, this sub-step is used to select candidate sampling points with relatively good expressive ability of imaging quality from the candidate sampling points. The first threshold can be set by the actual deployment density of construction facilities and can be fine-tuned based on scenario experience. For example, the first threshold is 0.3.

[0090] Step 3: Based on the first image data corresponding to the target candidate sampling point and the structure-aware score, select a target sampling point from the target candidate sampling points; wherein, the target sampling point is a high-value sampling point and the overlap of the viewing angle range of different target sampling points should be less than a second threshold, so as to avoid repeatedly collecting images of the same location.

[0091] Step 4: Determine the flight path based on the target sampling points.

[0092] In the above-mentioned optional implementation methods, firstly, candidate sampling points with good image quality expression ability are selected based on the structure perception score of the ability to express the image quality. Then, target sampling points with small viewpoint overlap and relatively high quality are selected from the candidate sampling points. Flight paths are generated based on the target sampling points, so that the subsequent low-altitude cruise equipment flies along the flight path and collects image data at the target sampling points, generating map data of the construction area in real time, thereby realizing map updates. This can improve the timeliness and accuracy of map updates.

[0093] As a specific implementation, step one above, which determines the structure-aware score of each candidate sampling point based on the flight data and the candidate sampling points, includes the following sub-steps:

[0094] Sub-step 1: Based on the flight altitude in the flight data and the recommended altitude corresponding to the work segment where the candidate sampling point is located, determine the relative altitude fit of the candidate sampling point; wherein, the sensing area includes multiple work segments; wherein, the relative altitude fit is used to evaluate the closeness between the current flight altitude and the recommended altitude of the work segment, for example: the relative altitude fit is related to the ratio of the height difference between the flight altitude and the recommended altitude to the recommended altitude.

[0095] For example, the above sub-step one can be represented by the following formula:

[0096]

[0097] Among them, H(V) i Z represents the relative high degree of fit. i Z represents flight altitude. opt Indicates the recommended height.

[0098] Sub-step 2: Determine the edge recognition accuracy of the candidate sampling point based on the viewing direction of the candidate sampling point and the arrangement direction of the construction facilities; wherein, the edge recognition accuracy is related to the angle between the viewing direction and the arrangement direction.

[0099] For example, the above sub-step two can be represented by the following formula:

[0100]

[0101] in, V represents the angle between the viewing direction and the arrangement direction. edge (V i () indicates the accuracy of edge recognition.

[0102] Sub-step 3: Based on the pre-established occlusion model, the camera's field of view, and the pose data in the flight data, determine the occlusion probability corresponding to the candidate sampling point; wherein, the camera's field of view is the field of view of the camera on the low-altitude cruise device at the candidate sampling point; for example, the occlusion probability is used to evaluate the viewpoint V. i The risk of obstruction that may be encountered during shooting. This value is between 0 and 1, with a higher value indicating that the viewpoint is more easily obstructed.

[0103] The obstruction model in sub-step three above can be established based on prior geographic information data and specific targets (such as buildings, trees, etc.) in the construction area. This obstruction model reflects targets that may obstruct the field of vision (such as buildings or large construction equipment).

[0104] The above sub-step three can be to calculate the occlusion area within the field of view based on the relative position between the current pose of the UAV and the target object and the field of view of the camera.

[0105] Sub-step four: Determine the structure perception score based on the relative height adaptation, the edge recognition accuracy, and the occlusion probability; wherein the structure perception score can be determined based on the smaller of the relative height adaptation and the edge recognition accuracy, and the occlusion probability.

[0106] For example, the above sub-step four can be represented by the following formula:

[0107] Struct(V i )=(min{H(V i ),V edge (V i )})×(1-P occ (V i ))

[0108] Among them, P occ (V i ) indicates that at candidate sampling point V i The occlusion probability, Struct(V) i ) represents the structure perception score.

[0109] As a specific implementation, step three of the above optional implementation, selecting a target sampling point from the target candidate sampling points based on the first image data corresponding to the target candidate sampling point and the structure-aware score, includes:

[0110] Sub-step 1: For each of the target candidate sampling points, obtain the heat distribution map of various construction facilities corresponding to the work section where the target candidate sampling point is located, obtain the facility weights corresponding to each type of construction facility, and obtain the field of view range of the low-altitude cruise equipment at the target candidate sampling point;

[0111] It's important to note that due to the varying distribution and importance of targets in different work areas, the heatmaps for each type of construction facility will differ across these areas. Specifically, in the work area, construction facilities are densely distributed and typically of high importance, such as water-filled barriers, temporary road markings, and traffic cones. In the warning area, the number of construction facilities is minimal, primarily consisting of warning signs or notices, resulting in a more dispersed distribution on the heatmap. Furthermore, heatmaps for various types of construction facilities can be generated through statistical analysis and prior rules, specifically based on modeling the possible distribution of construction facilities within the work area.

[0112] It's also important to note that the weight of each type of construction facility depends on its function and task requirements in different work areas. For example, traffic cones are used to mark construction boundaries in all areas of a construction site, so their weight can be set to the same in all areas. Similarly, in warning zones, warning signs are used to issue important warnings to drivers, therefore their weight is higher; while in termination zones, the main function of warning signs is to remind drivers to resume normal traffic after construction is completed, therefore their weight is lower.

[0113] In addition, during the generation of high-precision maps, or in other words, during the updating of high-precision maps (i.e., the aforementioned first map), the weight of construction facilities is used to reflect the degree of contribution of various construction facilities to map modeling in the generation of high-precision map data corresponding to the area to be updated.

[0114] Sub-step two: Based on the field of view, the heat map, and the facility weights, determine a first estimated value for the target candidate sampling point. This first estimated value represents the visibility of the construction facilities at the target candidate sampling point. Specifically, the first estimated value is used to assess the density of construction facilities that can be captured within the shooting range of the candidate sampling point. For example, it can be obtained by integrating the probability of construction facilities within the field of view of each candidate sampling point.

[0115] For example, the above sub-step two can be represented by the following formula:

[0116]

[0117] Among them, P obj (V i F(V) represents the first estimate. i ) represents candidate sampling point V iThe corresponding field of view, that is, the area that can be captured from this perspective. Specifically, this area is determined based on the drone's position, flight altitude, and the camera's field of view; W e Denotes the facility weight of construction facility e; E is the set of highway maintenance safety facilities, such as temporary markings, cones, fences, water-filled barriers, signs, etc.; D e (x,y) represents the probability distribution of category e in the construction facility heatmap. The construction target heatmap can be used to infer the density of construction facilities based on different work section types and typical layout habits. Cones are densely packed in buffer zones and work areas, while warning areas have many signs. The section type is mapped to the target probability map using a template library.

[0118] Sub-step 3: Based on the field of view and the boundary information of the perception area in the first map, determine the second estimated value corresponding to the target candidate sampling point, wherein the second estimated value represents the gain estimate of the target candidate sampling point to the boundary information; that is, the second estimated value is used to estimate how many feasible area boundary points can be brought to the map under the current view, that is, to evaluate whether the target candidate sampling point can complete or enrich the construction boundary lines.

[0119] It should be noted here that the area boundary is usually obtained through prior planning, i.e., it can be based on historical data, construction application information, or a simplified physical model. For each target candidate sampling point, the amount of boundary information that the viewpoint can capture can be estimated by evaluating the degree of overlap between its visible range and the boundary of the construction area. That is, this second estimate is calculated based on previous map information, target location, and geometry.

[0120] Sub-step four: Determine the viewpoint score of the target sampling point based on the structure-aware score, the first estimated value, the second estimated value, and the sampling priority weight of the work area where the target sampling point is located; for example, the target candidate sampling point V i The perspective rating can be expressed by the following formula:

[0121] G(V i )=(P obj (V i )+P edge (V i ))×Struct(V i )×W j

[0122] Among them, G(V) i (P) is a rating based on perspective. obj (V i P is the first estimate. edge (V i ) is the second estimate, Struct(V)i W is the score for structure perception. j This represents the sampling priority weight.

[0123] Sub-step five: Select the target sampling point based on the first image data and the viewpoint scores of each of the target candidate sampling points. The target sampling point is, for example, a candidate target sampling point with a high viewpoint score and minimal overlapping area in the acquired image.

[0124] In the specific implementation method described above, information gain evaluation that integrates target coverage, boundary gain, and segment priority is used to screen high-value points in the mapping area from the perspective of the UAV, perform path planning tasks, and achieve efficient data collection under different density sampling points, which not only ensures the integrity of mission spatial information but also improves the flight efficiency of the UAV.

[0125] As a more specific implementation, sub-step five of the above specific implementation, selecting the target sampling point based on the first image data and the viewpoint scores of each of the target candidate sampling points, includes:

[0126] Based on the aforementioned perspective score, the target candidate sampling points are sorted in descending order;

[0127] The first image data corresponding to at least two consecutive target candidate sampling points in the sorting results are compared to obtain the comparison results; here, the comparison results are, for example, the proportion of overlapping viewpoints in the field of view.

[0128] Based on the comparison results, multiple first image data with overlapping viewpoints greater than the second threshold are obtained from the first image data;

[0129] Among the target candidate sampling points corresponding to the multiple first image data, the target candidate sampling point with the highest viewpoint score is retained; this step is used to remove low-value sampling points and retain high-value sampling points from target candidate sampling points with many overlapping viewpoints.

[0130] The first K target candidate sampling points from the remaining target candidate sampling points are taken as the target sampling points; where K is an integer.

[0131] In the more specific implementation described above, the target viewpoint is selected based on the viewpoint score and the overlapping viewpoints within the field of view of different target candidate sampling points. This enables the device to automatically select the fewest but most effective flight points, greatly improving flight efficiency and mapping efficiency.

[0132] Among the above implementation methods, the effectiveness of each candidate viewpoint is comprehensively evaluated through a structure-aware scoring function (considering height adaptability, shooting direction, occlusion probability, etc.), and the sampling density is dynamically controlled based on the results. The sampling density is dense in key sections and sparse in secondary areas, and the planning results have significant content relevance and task adaptability.

[0133] As an optional implementation, step 103, based on the second image data and the inertial measurement unit (IMU) data collected by the low-altitude cruise device when acquiring the second image data, updates the map layer corresponding to the sensing area in the first map, including:

[0134] Identify the construction facilities in the second image data; for example, this step can be: identifying the construction facilities (such as the aforementioned cones, fences, signs, etc.) through an image target detection model.

[0135] Based on the IMU data, the identified construction facilities are projected onto a three-dimensional world coordinate system to obtain the location point of the construction facilities in the ground coordinate system;

[0136] The above two steps are implemented as follows: the second image data and IMU data are time-synchronized, and the Simultaneous Localization and Mapping (SLAM) algorithm is used to recover the UAV trajectory and the pose of each image; the construction facilities (such as cones, fences, signs, etc.) are identified through the image target detection model, and the targets in the identification box are back-projected into the three-dimensional world coordinate system in combination with the camera extrinsic parameters to obtain the position point of the target in the ground coordinate system.

[0137] The location points are projected onto a two-dimensional coordinate plane to obtain a two-dimensional point set. The boundary of the sensing area is then fitted within the two-dimensional coordinate plane to generate map data corresponding to the sensing area. For example, this step involves projecting the spatial location points of all construction facilities onto the XY plane to form a two-dimensional point set. The boundary of the construction area is fitted, and a closed polygonal region is generated after fitting, and then encapsulated into a high-precision map format.

[0138] Update the map layer of the perceived area in the first map based on the map data.

[0139] It's important to note that after updating the map layer corresponding to the perception area in the first map based on the map data, the map data can be reviewed. For example, staff can view the fitted boundaries, target recognition results, etc., on the platform map and make visual corrections or confirmations. After approval, the updated map is automatically distributed, for example, to: RSU broadcast systems to send MAP information to vehicle OBUs; navigation map provider platforms for navigation systems to adjust routes; and roadside guidance screen systems to display route guidance for construction areas.

[0140] Furthermore, as an optional implementation, the method also includes:

[0141] Acquire third image data, wherein the third image data is collected by the low-altitude cruise device at the target sampling point during flight along the flight path.

[0142] For example, the above steps may involve obtaining third image data when construction is determined to be completed based on construction information, or when construction completion information is obtained from the construction unit.

[0143] If construction is determined to be completed based on the third image data, the update of the map layer corresponding to the perceived area in the first map is cancelled.

[0144] In other words, after the construction is completed, the construction unit submits a "completion declaration" in the system. Then, the execution subject of this application embodiment automatically dispatches the drone to perform a second flight mission. By comparing the identification results with the original construction target data, if it is detected that all construction targets have been removed or the boundary difference is less than the set threshold, the system automatically removes the relevant map layers and issues instructions to each terminal. Each terminal rolls back the updated map to the version before the update, realizing the closed-loop update of the map.

[0145] It should be noted that the map update method of the embodiments of this application can be executed by a cloud platform, an edge computing device, a drone, or a drone nest. When executed by a cloud platform, the cloud platform can obtain relevant data of the drone through interaction with the drone. When executed by a drone, the drone can obtain construction filing information and other content from the cloud platform. When executed by a drone nest or an edge computing device, these devices can interact with the drone and the cloud platform respectively to obtain corresponding data.

[0146] The system components related to the map update method described in the embodiments of this application may include the following devices or modules:

[0147] The cloud control platform receives construction registration information, manages information such as construction areas, construction units, and time periods, and generates flight operation requests based on task type. It provides a manual review interface for verifying boundary fitting and target recognition results. After approval, the map is simultaneously distributed to RSU broadcasting equipment, guidance screen display systems, map service provider platforms, etc. After construction completion and declaration, the drone mission is triggered again to complete re-scanning and mapping. The platform determines whether construction is complete by comparing targets and calculating boundary differences, automatically cancels map information, updates the system database, and completes closed-loop map lifecycle management.

[0148] Unmanned Aerial Vehicle (UAV): Equipped with a monocular camera and IMU (Induction Unit), it possesses autonomous flight and image acquisition capabilities. This unit can receive flight missions and paths from the platform, complete shooting tasks, and upload sensor data.

[0149] Drone Nest: Supports automatic drone exit and return to home triggered by cloud commands, and has capabilities such as drone charging, cloud command issuance, cloud communication, and data transmission.

[0150] Flight path planning and mission issuance module: Deployed on a cloud control platform or edge computing terminal, it defines a local perception area based on the construction area, generates candidate shooting points within it, performs information gain scoring, and outputs the optimal flight path. The path planning results are then converted into mission control commands and issued to the UAV.

[0151] Visual-inertial mapping module: Deployed on cloud control platform or edge computing terminal; processes image sequences and IMU data collected by UAV to restore camera flight trajectory and image shooting posture.

[0152] Below, in conjunction with Figure 3 The implementation process of the map update method according to the embodiments of this application will be described in detail below, along with the aforementioned system. This example includes the following steps:

[0153] Step 1: Triggering Construction Tasks and Generating Flight Tasks:

[0154] This step includes Figure 3 The process involves steps such as construction information reporting, task triggering, drone nest receiving the task, and drone flying to the task area. Specifically, the cloud control platform receives construction registration information from the construction party, including the location coordinates of the highway maintenance work area, the construction scope, the estimated start and end times, and the construction unit. Based on this registration information, the system automatically generates a perception task, defines a local scanning area, selects nearby drone take-off and landing points, and sends the task coordinate information to the flight path planning module for route and shooting point planning.

[0155] Step 2: Path Sampling and Gain-Driven Route Planning:

[0156] This step includesFigure 3 The path sampling algorithm and flight path planning steps are described below. Specifically, after the UAV flies to the target mission area, the system performs viewpoint sampling within the perception area, calculating the information gain value for each candidate shooting point. The information gain considers factors such as the visibility of the construction target, boundary coverage, and occlusion risk. The system selects several high-gain points as key shooting points and uses a path optimization algorithm to generate the optimal flight route, which is then sent to the UAV.

[0157] In the detailed implementation of step 2 above, a UAV flight path planning mechanism combining semantic work area division, multi-view sampling, multi-factor information gain evaluation, and path optimization is proposed for perceptual mapping tasks of road construction areas and boundaries. Its core objective is to optimize flight paths and improve image effectiveness and mapping efficiency while ensuring complete coverage of construction boundaries. This method aims to maximize viewpoint value, comprehensively considering the layout of construction structures, target coverage density, boundary extraction potential, and occlusion risk, and preferentially selecting a limited number of high-value flight shooting points to improve map generation efficiency. Specifically, it includes:

[0158] I. Semantic Modeling and Regional Division of the Work Area: Based on the center point and construction direction in the construction record information, the system divides the work area into six segments according to the standard distance: Warning Area → Upstream Transition Area → Buffer Zone (Horizontal and Vertical) → Work Area → Downstream Transition Area → Termination Area. Each segment S j Defined as a spatial polygon, and accompanied by the semantic label ZONEtype(S j ) and sampling priority weight W j The weights are set to be highest in the working area, and then decrease in the buffer, transition area, alarm area and termination area, which serve as the basis for subsequent view sampling and gain scoring.

[0159] II. Viewpoint Sampling: Responsible for generating a global viewpoint candidate point set V = {V1, V2, ..., V} within the perception area. n The aforementioned candidate sampling points serve as the initial input for path optimization. The specific process is as follows:

[0160] 2.1 Non-uniform grid sampling: In each working segment S j Within the sampling area, different sampling intervals are set according to their priority. Each sampling point includes: spatial coordinates, view direction, and the label of its segment. More dense sampling intervals are set in the working area and buffer zone, while sparser sampling intervals are allocated in the transition zone and warning zone; sampling points point by default to the direction of maximum target density.

[0161] 2.2 Structure-aware prediction mechanism (this step is to prevent low-quality image locations from entering the mapping process):

[0162] To enhance the ability of sampling points to represent image quality, a structure-aware prediction function Struct(V) was designed. i This is used to assess the suitability of the shooting perspective for map building at the sampling point level in advance. It is defined as:

[0163] Struct(V i )=(min{H(V i ),V edge (V i )})×(1-P occ (V i ))

[0164] in:

[0165] H(V i () is the relative altitude fit evaluation function, which evaluates the current flight altitude Z. i Recommended height Z of the target section opt Proximity:

[0166]

[0167] V edge (V i The angle () indicates the edge recognition accuracy capability (angle between the shooting direction and the direction of the construction facilities arrangement).

[0168]

[0169] P occ (V i ) represents the occlusion probability, used to evaluate the viewpoint V. i The risk of obstruction that may be encountered during shooting. This value is between 0 and 1, with a higher value indicating that the viewpoint is more easily obstructed.

[0170] The structure-aware score ranges from (0, 1), reflecting the direct contribution of this perspective to boundary mapping. The system sets a filtering threshold; when Struct(V) = ... i If V < threshold, then V i Sampling points identified as low-value will be removed and excluded from path planning. The filtering threshold is set based on the actual deployment density of construction facilities and can be fine-tuned based on scenario experience.

[0171] Information gain function construction and high-value point screening:

[0172] Information gain function construction and high-value point screening:

[0173] The remaining available viewpoints (corresponding to the aforementioned target candidate sampling points) will participate in the information gain function calculation for subsequent sorting and selection:

[0174] G(V i)=(P obj (V i )+P edge (V i ))×Struct(V i )×W j

[0175] Among them, P obj (V i The visibility prediction of the construction target (i.e., the aforementioned construction facilities) is used to assess the viewpoint V. i The density of construction targets that can be captured within the shooting range. The probability of targets within the field of view of each viewpoint is integrated.

[0176]

[0177] E represents the set of highway maintenance safety facilities, such as temporary markings, cones, fences, water-filled barriers, and signs; D e (x,y) is a heatmap of construction targets, corresponding to the probability distribution of category e. The probability distribution of category e is used to infer target density based on different section types and typical layout habits. Cones are densely packed in buffer zones and work areas, while warning areas have many signs. A template library is used to map section types to target probability maps. W e Weights for various objectives.

[0178] P edge (V i This is for boundary information gain estimation, estimating how many feasible boundary points can be brought to the map from the current viewpoint, and evaluating the viewpoint V. i Can the construction boundary lines be completed or enriched? j The importance weight of the sampling point to which it belongs is the aforementioned sampling priority weight, with the highest weight in the working area and decreasing weight towards the termination area.

[0179] Calculate the score G(V) for all viewpoints. i The list is obtained by sorting the scores in descending order:

[0180]

[0181] For highly overlapping viewpoints, a filtering process is performed, retaining high-scoring points and selecting the top K target candidate sampling points for UAV path planning:

[0182]

[0183] Step 3: Unmanned Aerial Vehicle (UAV) Flight and Sensing Data Acquisition:

[0184] The drone flies to the mission area according to platform instructions and sequentially reaches the shooting points (i.e., the aforementioned target sampling points) along the planned path. Images are acquired at each shooting point, and IMU data is recorded for subsequent pose recovery. The flight process supports automatic obstacle avoidance and hovering shooting, with data transmitted back in real time.

[0185] Step 4: Visual-inertial mapping and target localization:

[0186] The algorithm performs time synchronization processing on the acquired images and IMU data, and uses the SLAM algorithm to recover the UAV trajectory and the pose of each image capture. The system identifies construction facilities (such as cones, fences, signs, etc.) through an image target detection model, and backprojects the targets in the identified boxes into a three-dimensional world coordinate system based on the camera's extrinsic parameters to obtain the location points of the construction facilities in the ground coordinate system.

[0187] Step 5: Map encapsulation. This step specifically includes boundary extraction and map encapsulation.

[0188] All spatial locations of construction facilities are projected onto the XY plane to form a two-dimensional point set. The boundary of the construction area is fitted, and a closed polygonal region is generated after fitting. This region is then encapsulated in a high-precision map format for subsequent use in roadside unit (RSU) broadcasts, guidance screen displays, and map service platform information updates.

[0189] Step 6: Manual review and issuance:

[0190] The packaged map data first enters the review process. Staff can view the fitted boundaries, target recognition results, and other information on the platform map and make visual corrections or confirmations. After the review is approved, the map will be automatically distributed to: the RSU broadcast system, used to send map (MAP) information to the On-Board Unit (OBU); the navigation map provider platform, for the navigation system to adjust the route; and the roadside guidance screen system, displaying route guidance for construction areas.

[0191] Step 7: Completion inspection and map rollback, i.e., construction completion inspection and map closed-loop update:

[0192] After construction is completed, the construction unit submits a "completion declaration" in the system. The system automatically dispatches drones to perform a second flight mission. By comparing the recognition results with the original construction target data, if it detects that all construction targets have been removed or the boundary differences are less than the set threshold, the system automatically removes the relevant map layers and sends instructions to each terminal. Each terminal restores the original map (i.e., the aforementioned first map), thus achieving a closed-loop update of the map.

[0193] In the above embodiments of this application, firstly, a semantic model of road construction operation sections is introduced to achieve differentiated control of sparse and dense sampling, and a view gain-driven evaluation system for high-precision map construction is established, including multi-dimensional factors such as regional priority, target density, and structural perception. Secondly, an information gain model and a high-value point screening algorithm are introduced: an information gain evaluation model integrating target coverage, boundary gain, and section priority is constructed to screen high-value points in the mapping area from the perspective of the UAV, perform path planning tasks, and achieve efficient data collection under different density sampling points, ensuring the integrity of task spatial information and improving UAV flight efficiency. Thirdly, a lightweight automatic generation mechanism for high-precision maps of construction areas is introduced: using low-cost sensors such as monocular cameras and inertial navigation equipment mounted on the UAV, high-precision maps of highway construction areas are automatically generated, and combined with the information gain model for sparse and dense sampling of construction areas, the high-precision maps of construction areas are updated in a low-cost manner without relying on on-site manual operations or modifying construction machinery. Fourthly, a closed-loop management process for construction area information based on drones is introduced, from reporting to distribution: Utilizing the automated flight capabilities of drones, combined with the mapping model algorithms and information management system mounted on the cloud control platform, high-precision maps of temporary construction areas are collected and updated in a timely manner, and the maps are distributed in conjunction with roadside infrastructure; after construction is completed, the drones trigger a re-flight mission and withdraw the map, completing the closed loop of road construction information management. This achieves map updates with strong real-time performance and fast response speed.

[0194] This application proposes a high-precision map automatic update platform application for road construction scenarios. The core idea is to use construction registration information as a trigger source, automatically operate low-altitude inspection equipment deployed along the highway, and use a drone equipped with a monocular camera and IMU to achieve aerial perception and mapping of the highway maintenance work area. Combined with an information gain-driven flight path planning algorithm, the platform accurately identifies the boundaries of the construction area, generates a temporary high-precision map conforming to a standard format, and publishes and updates the multi-source map through a cloud control platform. Specifically, after receiving road construction registration information, the platform automatically dispatches a drone to the highway maintenance work area. During flight, the drone collects image data and IMU pose data, and uses inertial navigation to recover the flight trajectory and camera attitude. Simultaneously, it performs information gain scoring through a preset set of candidate viewpoints (i.e., the aforementioned candidate sampling points), selecting key shooting points with advantages in construction target visibility and boundary coverage, planning an efficient flight path, and executing the task. After data acquisition, the system performs 3D localization of the construction targets identified in the images and projects them onto the ground coordinate system. Using a fitting algorithm, it generates the spatial boundaries of the construction area and finally encapsulates the data into a formatted map for manual review before distributing it to RSUs, guidance terminals, or navigation platforms. Furthermore, an automatic construction status verification mechanism is introduced. After construction is completed, the system again dispatches drones to perform a "re-scanning task." By comparing the perception results with the original construction boundary data, it determines whether the construction area has been cleared. Upon successful verification, it automatically cancels or updates the map information, achieving a complete closed-loop update of map data from application, generation, publication to clearing.

[0195] like Figure 4 As shown, embodiments of this application also provide a map updating device, such as... Figure 4 As shown, the device includes:

[0196] The first acquisition module 401 is used to acquire a sensing area associated with construction filing information, wherein multiple candidate sampling points are set in the sensing area;

[0197] The determining module 402 is used to determine the flight path of the low-altitude cruise device based on the first image data, the candidate sampling points, and the flight data of the low-altitude cruise device, wherein the flight path includes multiple target sampling points; the first image data is collected by the low-altitude cruise device at the candidate sampling points; and the flight data is the flight data of the low-altitude cruise device at the candidate sampling points.

[0198] The first update module 403 is used to update the map layer corresponding to the perception area in the first map based on the second image data and the inertial measurement unit (IMU) data of the low-altitude cruise device when collecting the second image data; wherein, the second image data is collected by the low-altitude cruise device at the target sampling point during flight along the flight path.

[0199] The first acquisition module 401 includes:

[0200] The first acquisition submodule is used to acquire the construction center point and construction direction based on the construction filing information;

[0201] The region segmentation submodule is used to perform semantic modeling and region segmentation of the work area based on the construction center point and the construction direction, and to obtain multiple work segments, semantic labels of each work segment, and sampling priority weights of each work segment.

[0202] The sampling point setting submodule is used to set candidate sampling points in each of the work sections according to the sampling priority weight; wherein, the spacing between the candidate sampling points in different work sections is negatively correlated with the sampling priority weight;

[0203] The second acquisition submodule is used to obtain the sensing area based on each of the work segments and the candidate sampling points.

[0204] The determining module 402 includes:

[0205] The first determining submodule is used to determine the structure perception score of each of the candidate sampling points based on the flight data and the candidate sampling points;

[0206] A filtering submodule is used to filter target candidate sampling points from the candidate sampling points, wherein the structure-aware score of the target candidate sampling points is greater than or equal to a first threshold.

[0207] The selection submodule is used to select a target sampling point from the target candidate sampling points based on the first image data corresponding to the target candidate sampling point and the structure-aware score.

[0208] The second determining submodule is used to determine the flight path based on the target sampling points.

[0209] The first determining submodule includes:

[0210] The first determining unit is configured to determine the relative altitude fit of the candidate sampling point based on the flight altitude in the flight data and the recommended altitude corresponding to the work section where the candidate sampling point is located; wherein, the sensing area includes multiple work sections;

[0211] The second determining unit is used to determine the edge recognition accuracy capability corresponding to the candidate sampling point based on the viewing direction of the candidate sampling point and the arrangement direction of the construction facilities.

[0212] The third determining unit is used to determine the occlusion probability corresponding to the candidate sampling point based on the pre-established occlusion model, the field of view of the camera, and the pose data in the flight data; wherein, the field of view of the camera is the field of view of the camera on the low-altitude cruise equipment at the candidate sampling point;

[0213] The fourth determining unit is used to determine the structure perception score based on the relative height adaptation, the edge recognition accuracy, and the occlusion probability.

[0214] The selection submodule includes:

[0215] The first acquisition unit is used to acquire, for each of the target candidate sampling points, a heat map of various construction facilities corresponding to the work section where the target candidate sampling point is located, acquire the facility weights corresponding to each of the construction facilities, and acquire the field of view range of the low-altitude cruise equipment at the target candidate sampling point.

[0216] The fifth determining unit is used to determine a first estimated value of the target candidate sampling point based on the field of view, the heat distribution map, and the facility weight, wherein the first estimated value represents the visibility of the construction facility at the target candidate sampling point;

[0217] The sixth determining unit is used to determine a second estimated value corresponding to the target candidate sampling point based on the field of view and the boundary information of the perception area in the first map, wherein the second estimated value represents the gain estimate of the target candidate sampling point to the boundary information;

[0218] The seventh determining unit is used to determine the viewpoint score of the target sampling point based on the structure perception score, the first estimated value, the second estimated value, and the sampling priority weight of the work area where the target sampling point is located;

[0219] The selection unit is used to select the target sampling point based on the first image data and the viewpoint scores of each of the target candidate sampling points.

[0220] The selection unit includes:

[0221] The sorting subunit is used to sort the target candidate sampling points in descending order according to the viewpoint score;

[0222] The comparison subunit is used to compare the first image data corresponding to at least two consecutive target candidate sampling points in the sorting result to obtain the comparison result;

[0223] The acquisition subunit is used to acquire multiple first image data with an overlapping viewpoint greater than a second threshold from the first image data based on the comparison result.

[0224] The elimination subunit is used to retain the target candidate sampling point with the highest viewpoint score among the target candidate sampling points corresponding to the plurality of first image data;

[0225] A subunit is defined to select the top K target candidate sampling points from the remaining target candidate sampling points as the target sampling points; where K is an integer.

[0226] The first update module 403 includes:

[0227] The identification submodule is used to identify construction facilities in the second image data;

[0228] The projection submodule is used to project the identified construction facility onto a three-dimensional world coordinate system based on the IMU data, so as to obtain the position point of the construction facility in the ground coordinate system;

[0229] The processing submodule is used to project the location points onto a two-dimensional coordinate plane to obtain a two-dimensional point set, and to fit the boundary of the sensing area in the two-dimensional coordinate plane to generate map data corresponding to the sensing area.

[0230] The update submodule is used to update the map layer of the perceived area in the first map based on the map data.

[0231] The device further includes:

[0232] The second acquisition module is used to acquire third image data, wherein the third image data is collected by the low-altitude cruise device at the target sampling point during the flight of the flight path;

[0233] The second update module is used to cancel the update of the map layer corresponding to the perception area in the first map when the construction is determined to be completed based on the third image data.

[0234] It should be noted that the map update apparatus provided in this application embodiment can implement all the method steps implemented in the above map update method embodiment and can achieve the same technical effect. Here, the parts that are the same as those in the method embodiment and the beneficial effects will not be described in detail.

[0235] An embodiment of this application also provides a map updating device, including a transceiver 510, a processor 500, a memory 520, and a program stored in the memory 520 and executable on the processor 500; wherein, when the processor 500 executes the program, it implements the map updating method as described above.

[0236] The transceiver 510 is used to receive and send data under the control of the processor 500.

[0237] Among them, Figure 5 In this context, the bus architecture can include any number of interconnected buses and bridges, specifically linking various circuits together, represented by one or more processors (processor 500) and memory (memory 520). The bus architecture can also link various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. The bus interface provides an interface. The transceiver 510 can be multiple elements, including a transmitter and a receiver, providing a unit for communicating with various other devices over a transmission medium.

[0238] The processor 500 is responsible for managing the bus architecture and general processing, while the memory 520 can store the data used by the processor 500 when performing operations.

[0239] This application also provides a readable storage medium storing a program. When executed by a processor, this program implements the map update method described above and achieves the same technical effect. To avoid repetition, it will not be described again here. The readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, etc.

[0240] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, disk, optical disk) and includes several instructions for executing the methods described in the various embodiments of this application.

[0241] Therefore, embodiments of this application also provide a computer program product, including computer instructions, which, when executed by a processor, implement the map update method described above and achieve the same technical effect. To avoid repetition, further details are omitted here.

[0242] In this embodiment, the module can be implemented in software so that it can be executed by various types of processors. For example, an identified executable code module may include one or more physical or logical blocks of computer instructions, which may be constructed as objects, procedures, or functions. Nevertheless, the executable code of the identified module does not need to be physically located together, but may include different instructions stored in different bits, which, when logically combined, constitute the module and achieve the module's intended purpose.

[0243] In practice, an executable code module can be a single instruction or many instructions, and can even be distributed across multiple different code segments, different programs, and across multiple memory devices. Similarly, operational data can be identified within the module and can be implemented in any suitable form and organized within any suitable type of data structure. This operational data can be collected as a single dataset or distributed across different locations (including different storage devices), and can exist, at least in part, solely as electronic signals within the system or network.

[0244] When a module can be implemented using software, considering the current level of hardware technology, modules that can be implemented in software can be implemented using hardware circuits by those skilled in the art to achieve the corresponding functions, without considering cost. These hardware circuits include conventional very-large-scale integrated circuits (VLSI) or gate arrays, as well as existing semiconductors such as logic chips and transistors, or other discrete components. Modules can also be implemented using programmable hardware devices, such as field-programmable gate arrays, programmable array logic, and programmable logic devices.

[0245] The exemplary embodiments described above are with reference to the accompanying drawings. Many different forms and embodiments are feasible without departing from the spirit and teachings of this application. Therefore, this application should not be construed as limiting the exemplary embodiments set forth herein. Rather, these exemplary embodiments are provided to make this application complete and convey the scope of this application to those skilled in the art. In these drawings, component dimensions and relative dimensions may be exaggerated for clarity. The terminology used herein is for the purpose of describing particular exemplary embodiments only and is not intended to be limiting. As used herein, unless clearly indicated otherwise, the singular forms “a,” “an,” and “the” are intended to include all such forms. It will be further understood that the terms “comprising” and / or “including”, when used in this specification, indicate the presence of the stated features, integers, steps, operations, components, and / or elements, but do not exclude the presence or addition of one or more other features, integers, steps, operations, components, and / or groups thereof. Unless otherwise indicated, when stated, a range of values ​​includes the upper and lower limits of the range and any subranges in between.

[0246] The above description is the preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principles described in this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A map updating method, characterized in that, include: Obtain the sensing area associated with the construction filing information, wherein multiple candidate sampling points are set within the sensing area; Based on the first image data, the candidate sampling points, and the flight data of the low-altitude cruise device, the flight path of the low-altitude cruise device is determined, wherein the flight path includes multiple target sampling points; the first image data is collected by the low-altitude cruise device at the candidate sampling points; the flight data is the flight data of the low-altitude cruise device at the candidate sampling points. Based on the second image data and the inertial measurement unit (IMU) data collected by the low-altitude cruise device when acquiring the second image data, the map layer corresponding to the perception area in the first map is updated; wherein, the second image data is acquired by the low-altitude cruise device at the target sampling point during flight along the flight path.

2. The method according to claim 1, characterized in that, The sensing area associated with construction filing information is obtained, including: The construction center point and construction direction are obtained based on the construction filing information. Based on the construction center point and the construction direction, semantic modeling and regional division of the work area are performed to obtain multiple work segments, semantic labels of each work segment, and sampling priority weights of each work segment; Candidate sampling points are set in each of the work sections according to the sampling priority weight; wherein the spacing between the candidate sampling points in different work sections is negatively correlated with the sampling priority weight. The sensing area is obtained based on each of the described working sections and the candidate sampling points.

3. The method according to claim 1, characterized in that, Based on the first image data, the candidate sampling points, and the flight data of the low-altitude cruise device, the flight path of the low-altitude cruise device is determined, including: Based on the flight data and the candidate sampling points, a structure perception score is determined for each of the candidate sampling points; Among the candidate sampling points, target candidate sampling points are selected, wherein the structure-aware score of the target candidate sampling points is greater than or equal to a first threshold. Based on the first image data corresponding to the target candidate sampling point and the structure-aware score, a target sampling point is selected from the target candidate sampling points; The flight path is determined based on the target sampling points.

4. The method according to claim 3, characterized in that, Based on the flight data and the candidate sampling points, a structure-aware score is determined for each candidate sampling point, including: Based on the flight altitude in the flight data and the recommended altitude corresponding to the work section where the candidate sampling point is located, the relative altitude fit of the candidate sampling point is determined; wherein, the sensing area includes multiple work sections; The edge recognition accuracy of the candidate sampling points is determined based on the viewing direction of the candidate sampling points and the arrangement direction of the construction facilities. Based on the pre-established occlusion model, the camera's field of view, and the pose data in the flight data, the occlusion probability corresponding to the candidate sampling point is determined; wherein, the camera's field of view is the field of view of the camera on the low-altitude cruise equipment at the candidate sampling point; The structure perception score is determined based on the relative height adaptation, the edge recognition accuracy, and the occlusion probability.

5. The method according to claim 3, characterized in that, Based on the first image data corresponding to the target candidate sampling point and the structure-aware score, selecting a target sampling point from the target candidate sampling points includes: For each of the target candidate sampling points, obtain a heat map of various construction facilities corresponding to the work area where the target candidate sampling point is located, obtain the facility weights corresponding to each type of construction facility, and obtain the field of view range of the low-altitude cruise equipment at the target candidate sampling point; Based on the field of view, the heat map, and the facility weight, a first estimated value is determined for the target candidate sampling point, wherein the first estimated value represents the visibility of the construction facility at the target candidate sampling point; Based on the field of view and the boundary information of the perception area in the first map, a second estimated value corresponding to the target candidate sampling point is determined, wherein the second estimated value represents the gain estimate of the target candidate sampling point to the boundary information; The viewpoint score of the target sampling point is determined based on the structure perception score, the first estimated value, the second estimated value, and the sampling priority weight of the work area where the target sampling point is located. The target sampling point is selected based on the first image data and the viewpoint scores of each of the target candidate sampling points.

6. The method according to claim 5, characterized in that, Selecting the target sampling point based on the first image data and the viewpoint scores of each of the target candidate sampling points includes: Based on the aforementioned perspective score, the target candidate sampling points are sorted in descending order; The first image data corresponding to at least two consecutive target candidate sampling points in the sorting results are compared to obtain the comparison results; Based on the comparison results, multiple first image data with overlapping viewpoints greater than the second threshold are obtained from the first image data; Among the target candidate sampling points corresponding to the plurality of first image data, the target candidate sampling point with the highest viewpoint score is retained; The first K target candidate sampling points from the remaining target candidate sampling points are taken as the target sampling points; where K is an integer.

7. The method according to claim 1, characterized in that, Based on the second image data and the inertial measurement unit (IMU) data collected by the low-altitude cruise device when acquiring the second image data, the map layer corresponding to the sensing area in the first map is updated, including: Identify construction facilities in the second image data; Based on the IMU data, the identified construction facilities are projected onto a three-dimensional world coordinate system to obtain the location point of the construction facilities in the ground coordinate system; The location points are projected onto a two-dimensional coordinate plane to obtain a two-dimensional point set, and the boundary of the sensing area is fitted in the two-dimensional coordinate plane to generate map data corresponding to the sensing area. Update the map layer of the perceived area in the first map based on the map data.

8. The method according to claim 1, characterized in that, The method further includes: Acquire third image data, wherein the third image data is collected by the low-altitude cruise device at the target sampling point during flight along the flight path; If construction is determined to be completed based on the third image data, the update of the map layer corresponding to the perceived area in the first map is cancelled.

9. A map updating device, characterized in that, include: The first acquisition module is used to acquire the sensing area associated with the construction filing information, and multiple candidate sampling points are set in the sensing area; The determining module is used to determine the flight path of the low-altitude cruise device based on the first image data, the candidate sampling points, and the flight data of the low-altitude cruise device, wherein the flight path includes multiple target sampling points; the first image data is collected by the low-altitude cruise device at the candidate sampling points; and the flight data is the flight data of the low-altitude cruise device at the candidate sampling points. The first update module is used to update the map layer corresponding to the perception area in the first map based on the second image data and the inertial measurement unit (IMU) data of the low-altitude cruise device when collecting the second image data; wherein the second image data is collected by the low-altitude cruise device at the target sampling point during flight along the flight path.

10. A map updating device, comprising a transceiver, a processor, a memory, and a program stored in the memory and executable on the processor; characterized in that, The transceiver transmits and receives data under the control of the processor, and the processor implements the map update method as described in any one of claims 1 to 8 when executing the program.

11. A readable storage medium having a program or instructions stored thereon, characterized in that, When the program or instructions are executed by the processor, they implement the map update method as described in any one of claims 1 to 8.

12. A computer program product, characterized in that, It includes computer instructions that, when executed by a processor, implement the map update method as described in any one of claims 1 to 8.

Images