Crowdsourcing map data processing method and device and related equipment

Abstract

The invention provides a crowdsourcing map data processing method, a crowdsourcing map data processing device and related equipment, and relates to map data processing technology applied to the field of vehicles, and the method comprises the following steps: in response to a vehicle entering a parking area, acquiring relative positioning information, absolute positioning information and geographic space data of the vehicle in a vehicle driving process; determining a plurality of key frames based on the relative positioning information; the difference values between the relative positioning of the vehicles corresponding to any two adjacent key frames are the same; extracting and corresponding the absolute positioning information and the geographic space data based on the timestamp information of the key frames, and determining the absolute positioning information and the geographic space data corresponding to each key frame; generating crowdsourcing map data according to the timestamp information of the key frame and the absolute positioning information and geographic space data corresponding to the key frame; and uploading the crowdsourcing map data to a cloud. According to the invention, it can be ensured that the vehicle uploads a small amount of data, does not occupy resources and traffic of the vehicle using end, and also ensures that the cloud can generate an accurate map.

Description

Technical Field

[0001] This disclosure relates to the application of map data processing technology in the vehicle field, and in particular to a crowdsourced map data processing method, apparatus and related equipment. Background Technology

[0002] With the gradual development of intelligent driving technology, people's requirements for parking are becoming increasingly higher. To realize valet parking functions in parking areas, it is essential to build accurate parking lot maps. Crowdsourcing can be used to collect crowdsourced data when users park their vehicles and upload it to the cloud, where the cloud can then create maps based on this large amount of data. However, the resource consumption of large amounts of crowdsourced data and upload bandwidth limitations have become bottlenecks restricting the large-scale construction of high-precision parking lot maps.

[0003] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0004] To overcome the problems existing in related technologies, this disclosure provides a crowdsourced map data processing method, apparatus, vehicle, electronic device, storage medium, and program product.

[0005] According to a first aspect of the present disclosure, a crowdsourced map data processing method is provided, comprising: In response to a vehicle entering a parking area, the relative positioning information, absolute positioning information, and geospatial data of the vehicle during its driving process are collected. Based on the relative positioning information, multiple key frames are determined; wherein, the difference between the relative positioning of the vehicles corresponding to any two adjacent key frames is the same. Based on the timestamp information of the key frame, the absolute positioning information and the geospatial data are extracted and correlated to determine the absolute positioning information and geospatial data corresponding to each key frame. Crowdsourced map data is generated based on the timestamp information of the keyframes, as well as the absolute positioning information and geospatial data corresponding to the keyframes. The crowdsourced map data is uploaded to the cloud.

[0006] In some possible embodiments of this disclosure, the crowdsourced map data processing method further includes: In response to determining that the vehicle is heading toward a target parking lot, the parking area is determined based on the location of the target parking lot; Uploading the crowdsourced map data to the cloud includes: Upon completion of parking in the target parking lot, the crowdsourced map data is uploaded to the cloud.

[0007] The above implementation can reduce communication frequency and bandwidth usage, and also ensure that the uploaded content fully covers the entire parking path, which facilitates structured parsing, fusion and version management in the cloud.

[0008] In some possible embodiments of this disclosure, the crowdsourced map data processing method further includes: In response to determining that the vehicle has left the target parking lot, the parking area is determined based on the location of the target parking lot; Uploading the crowdsourced map data to the cloud includes: In response to the vehicle leaving the parking area, the crowdsourced map data is uploaded to the cloud.

[0009] The above implementation can reduce communication frequency and bandwidth usage, and also ensure that the uploaded content fully covers the entire parking path, which facilitates structured parsing, fusion and version management in the cloud.

[0010] In some possible embodiments of this disclosure, the geospatial data includes one or any combination of point cloud data, topological data, and vector information.

[0011] In some possible embodiments of this disclosure, based on the timestamp information of the keyframes, the absolute positioning information and the geospatial data are extracted and correlated to determine the absolute positioning information and geospatial data corresponding to each keyframe, including: Based on the timestamp information of the key frame, extract the point cloud data within the time window from the previous key frame to the current key frame. The extracted point cloud data is spatially downsampled, accumulated, and fused to obtain the processed point cloud data. The processed point cloud data is compressed to obtain the point cloud data corresponding to the current keyframe.

[0012] The above implementation can effectively suppress the redundancy of point cloud data, significantly reduce the consumption of computing, storage and communication resources, and provide key technical support for efficient and scalable map generation and updating.

[0013] In some possible embodiments of this disclosure, based on the timestamp information of the keyframes, the absolute positioning information and the geospatial data are extracted and correlated to determine the absolute positioning information and geospatial data corresponding to each keyframe, including: Based on the timestamp information of the key frame, extract the vector information from the previous key frame to the current key frame; The extracted vector information is spatiotemporally aligned and geometrically fused to generate processed vector information; Redundancy is removed from the processed vector information to obtain the vector information corresponding to the current keyframe.

[0014] In the above implementation, temporal integration, spatial optimization and redundancy suppression of vector information can be achieved.

[0015] In some possible embodiments of this disclosure, the crowdsourced map data further includes: image data; Correspondingly, the provided crowdsourced map data processing methods also include: In response to the vehicle traveling within a preset image acquisition area, the vehicle's driving status is determined; The image acquisition angle is determined based on the vehicle's driving status, and image data is acquired according to the preset driving distance interval based on the image acquisition angle.

[0016] In the above implementation, the total amount of image data that needs to be uploaded can be significantly reduced while ensuring coverage of key scene images.

[0017] In some possible embodiments of this disclosure, determining the image acquisition perspective based on the vehicle's driving state includes: If the vehicle is in a reverse parking state, the image acquisition perspective is a rear-view acquisition perspective; If the vehicle is traveling in a forward direction, the image acquisition perspective is a forward-looking acquisition perspective.

[0018] According to a second aspect of the present disclosure, a crowdsourced map data processing method is provided, comprising: Receive crowdsourced map data uploaded by vehicles; The crowdsourced map data includes: multiple keyframes, and absolute positioning information and geospatial data corresponding to each keyframe; The crowdsourced map data is generated from the vehicle through the following steps: In response to a vehicle entering a parking area, the relative positioning information, absolute positioning information, and geospatial data of the vehicle during its driving process are collected. Based on the relative positioning information, multiple key frames are determined; wherein, the difference between the relative positioning of the vehicles corresponding to any two adjacent key frames is the same. Based on the timestamp information of the key frame, the absolute positioning information and the geospatial data are extracted and correlated to determine the absolute positioning information and geospatial data corresponding to each key frame. The crowdsourced map data is generated based on the timestamp information of the keyframes, as well as the absolute positioning information and geospatial data corresponding to the keyframes.

[0019] In some possible embodiments of this disclosure, the crowdsourced map data further includes: image data; The image data was collected by the vehicle through the following steps: In response to the vehicle traveling within a preset image acquisition area, the vehicle's driving status is determined; The image acquisition angle is determined based on the vehicle's driving state, and the image data is acquired according to the preset driving distance interval based on the image acquisition angle.

[0020] In some possible embodiments of this disclosure, the crowdsourced map data processing method further includes: Based on the timestamp information in the image data, and combined with the timestamp information of the key frames, interpolation and coordinate transformation are performed to obtain the image recognition data associated with each key frame.

[0021] The above implementation can compensate for the information loss caused by sparse image sampling, enhance the performance of crowdsourced maps in terms of semantic integrity, geometric accuracy and topological consistency, and avoid the storage and bandwidth burden caused by high-frequency image acquisition by the vehicle in pursuit of strict synchronization.

[0022] In some possible embodiments of this disclosure, the provided crowdsourced map data processing method further includes: aligning, fusing, and structurally modeling the crowdsourced map data to generate a map of the target parking lot.

[0023] In the above implementation, a map of the target parking lot with high accuracy and rich semantics can be generated.

[0024] According to a third aspect of the present disclosure, a crowdsourced map data processing apparatus is provided, comprising: The data acquisition unit is used to collect the relative positioning information, absolute positioning information, and geospatial data of the vehicle during its driving process in response to the vehicle entering the parking area. A keyframe determination unit is used to determine multiple keyframes based on the relative positioning information; wherein the difference between the relative positioning of the vehicles corresponding to any two adjacent keyframes is the same. The data processing unit is used to extract and correlate the absolute positioning information and the geospatial data based on the timestamp information of the key frame, and to determine the absolute positioning information and geospatial data corresponding to each key frame. The map data generation unit is used to generate crowdsourced map data based on the timestamp information of the key frame, as well as the absolute positioning information and geospatial data corresponding to the key frame. The map data sending unit is used to upload the crowdsourced map data to the cloud.

[0025] In some possible embodiments of this disclosure, the crowdsourced map data further includes: image data; The crowdsourced map data processing device further includes: an image data generation unit, used for: In response to the vehicle traveling within a preset image acquisition area, the vehicle's driving status is determined; The image acquisition angle is determined based on the vehicle's driving status, and image data is acquired according to the preset driving distance interval based on the image acquisition angle.

[0026] According to a fourth aspect of the present disclosure, a crowdsourced map data processing apparatus is provided, comprising: The map data receiving unit is used to receive crowdsourced map data uploaded by vehicles. The crowdsourced map data includes: multiple keyframes, and absolute positioning information and geospatial data corresponding to each keyframe; The crowdsourced map data is generated from the vehicle through the following steps: In response to a vehicle entering a parking area, the relative positioning information, absolute positioning information, and geospatial data of the vehicle during its driving process are collected. Based on the relative positioning information, multiple key frames are determined; wherein, the difference between the relative positioning of the vehicles corresponding to any two adjacent key frames is the same. Based on the timestamp information of the key frame, the absolute positioning information and the geospatial data are extracted and correlated to determine the absolute positioning information and geospatial data corresponding to each key frame. The crowdsourced map data is generated based on the timestamp information of the keyframes, as well as the absolute positioning information and geospatial data corresponding to the keyframes.

[0027] In some possible embodiments of this disclosure, the crowdsourced map data further includes: image data; The image data was collected by the vehicle through the following steps: In response to the vehicle traveling within a preset image acquisition area, the vehicle's driving status is determined; The image acquisition angle is determined based on the vehicle's driving state, and the image data is acquired based on the image acquisition angle according to a preset driving distance interval. The crowdsourced map data processing device further includes: an image data processing unit, used for: Based on the timestamp information in the image data, and combined with the timestamp information of the key frames, interpolation and coordinate transformation are performed to obtain the image recognition data associated with each key frame.

[0028] According to a fifth aspect of the present disclosure, a vehicle is provided, comprising: processor; Memory used to store processor-executable instructions; The processor is configured to implement the steps of any of the crowdsourced map data processing methods described in the first aspect above.

[0029] According to a sixth aspect of the present disclosure, an electronic device is provided, comprising: Remote server and vehicle-side processor; The remote server is configured to implement the steps of any of the crowdsourced map data processing methods described in the second aspect above. The vehicle-mounted processor is configured to perform the steps of any of the crowdsourced map data processing methods described in the first aspect above.

[0030] According to a seventh aspect of the present disclosure, a non-transitory computer-readable storage medium is provided, wherein when instructions in the storage medium are executed by a processor of a terminal, the terminal is able to perform any of the crowdsourced map data processing methods described in the first or second aspect above.

[0031] According to an eighth aspect of the present disclosure, a computer program product is provided, including a computer program that, when executed by a processor, implements any of the crowdsourced map data processing methods described in the first or second aspect above.

[0032] The technical solutions provided by the embodiments of this disclosure may include the following beneficial effects: This embodiment of the disclosure collects the vehicle's relative positioning information, absolute positioning information, and geospatial data during its movement in response to a vehicle entering a parking area. Based on the relative positioning information, multiple keyframes are determined; wherein the difference between the relative positioning of any two adjacent keyframes is the same. Based on the timestamp information of the keyframes, the absolute positioning information and geospatial data are extracted and mapped to determine the absolute positioning information and geospatial data corresponding to each keyframe. Crowdsourced map data is generated based on the timestamp information of the keyframes and the absolute positioning information and geospatial data corresponding to the keyframes. The crowdsourced map data is then uploaded to the cloud. By using the various data collected from the vehicle side to determine keyframes based on relative positioning information and to determine the absolute positioning information and geospatial data corresponding to the keyframes, a large amount of collected data is reduced to obtain less but more effective crowdsourced map data. This ensures that the amount of data uploaded by the vehicle is small, does not occupy vehicle-side resources and traffic, and ensures that the cloud can generate accurate maps when using the generated crowdsourced map data for mapping.

[0033] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0034] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0035] Figure 1 This is a flowchart illustrating a crowdsourced map data processing method according to an exemplary embodiment of the present disclosure. Figure 1 .

[0036] Figure 2 This is a schematic flowchart illustrating the implementation process of step S130 according to an exemplary embodiment of the present disclosure. Figure 1 .

[0037] Figure 3 This is a schematic flowchart illustrating the implementation process of step S130 according to an exemplary embodiment of the present disclosure. Figure 2 .

[0038] Figure 4 This is a flowchart illustrating a crowdsourced map data processing method according to an exemplary embodiment of the present disclosure. Figure 2 .

[0039] Figure 5 This is an example schematic diagram illustrating crowdsourced map data generation and uploading in a specific instance according to an exemplary embodiment of this disclosure.

[0040] Figure 6 This is a structural block diagram of a crowdsourced map data processing apparatus according to an exemplary embodiment of the present disclosure.

[0041] Figure 7 This is a structural block diagram of another crowdsourced map data processing apparatus according to an exemplary embodiment of the present disclosure.

[0042] Figure 8 This is a block diagram illustrating a vehicle according to an exemplary embodiment of the present disclosure. Detailed Implementation

[0043] Exemplary embodiments of this disclosure will be described in detail herein, examples of which are illustrated in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings denote the same or similar elements unless otherwise indicated. Various changes, modifications, and equivalents of the methods, apparatus, and / or systems described herein will become apparent upon understanding this disclosure. For example, the order of operations described herein is merely illustrative and is not limited to those orders set forth herein, but can be changed as will become apparent upon understanding this disclosure, except for operations that must be performed in a particular order. Furthermore, for clarity and brevity, descriptions of features known in the art may be omitted.

[0044] The embodiments described below, which are examples of some of the embodiments of this disclosure, do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure as detailed in the appended claims.

[0045] It should be noted that the crowdsourced data involved in this disclosure is anonymized data, does not involve the disclosure of user privacy, and can be uploaded only after obtaining user authorization.

[0046] The specific implementation methods of the embodiments of this disclosure will now be described in detail with reference to the accompanying drawings.

[0047] Figure 1 This is a flowchart illustrating a crowdsourced map data processing method according to an exemplary embodiment of the present disclosure. Figure 1 It can be used in vehicles, including new energy vehicles.

[0048] In some embodiments of this disclosure, the crowdsourced map data processing method, applied to the vehicle side, includes the following steps: In step S110, in response to the vehicle entering the parking area, the relative positioning information, absolute positioning information, and geospatial data of the vehicle during its driving process are collected.

[0049] It should be noted that the cloud generates high-precision maps based on multi-source environmental perception data collected by vehicles. Vehicles can collect the following types of data: semantic point cloud data, which is acquired through onboard image sensors and includes road marking information with semantic meaning, such as ground lane lines and stop lines; laser point cloud data, obtained by LiDAR scanning, used to accurately describe the geometric structure of the three-dimensional environment around the vehicle; vector information, including but not limited to static road elements with clear spatial locations and geometric features, such as parking spaces, stop lines, and posts; and topological information, such as road centerlines, used to characterize the connectivity and driving logic of the road network.

[0050] In some embodiments of this disclosure, geospatial data includes one or any combination of point cloud data, topological data, and vector information. Point cloud data may include semantic point cloud data and laser point cloud data; topological data may include the aforementioned topological information. The geospatial data composed of the above multimodalities, after cloud-based fusion processing, can jointly construct a structured, semantically rich, high-precision map, providing reliable prior environmental information for intelligent driving systems.

[0051] It should be noted that relative positioning information represents the vehicle's relative position with respect to the previous data collection time, and may include distance and attitude, thus representing the vehicle's movement. Absolute positioning information represents the vehicle's precise geographical location in the global coordinate system, which can be determined by GNSS (Global Navigation Satellite System) combined with high-precision positioning algorithms.

[0052] In some embodiments of this disclosure, the parking area is pre-defined and can be dynamically adjusted based on the vehicle's driving route, the user's driving intention, and environmental context information. For example, when the vehicle senses its destination is a parking space in a parking lot, the parking area can be defined as a continuous segment along the driving path from 200 meters away from the parking lot entrance until the vehicle finally parks in the target space. Correspondingly, when the vehicle is sensed to be leaving the underground parking garage, the parking area extends from the currently parked space along the exit path to a position 200 meters away from the parking lot exit. This range can be adaptively scaled or offset according to the actual road structure, traffic efficiency, or historical parking trajectories. It is understood that most parking lots are underground, and their enclosed or semi-enclosed structures often result in limited satellite signals. The absolute positioning information collected by the vehicle within the parking lot may be inaccurate and may contain deviations. The cloud-based system needs to refer to the absolute positioning of the parking lot when generating maps to achieve seamless integration with external road maps or other regional maps.

[0053] To ensure the accuracy of map fusion processing in the cloud, this embodiment defines a parking area and combines high-reliability absolute positioning data acquired within a 200-meter radius of the parking lot entrance / exit point, such as transitional sections during vehicle entry / exit, to calibrate or anchor the positioning information within the parking lot. Since vehicles typically receive relatively stable GNSS signals and have high absolute positioning accuracy near entrances / exits, this area can be used as a geographic reference benchmark to help improve the global consistency and spatial accuracy of map features throughout the parking area, thereby effectively ensuring the reliability and processing accuracy of subsequent map generation.

[0054] In step S120, multiple keyframes are determined based on relative positioning information.

[0055] It should be noted that the difference between the relative vehicle positions corresponding to any two adjacent keyframes is the same.

[0056] In some embodiments of this disclosure, the difference between the relative vehicle positioning, including position and / or orientation, of any two adjacent keyframes remains substantially consistent. In other words, if the displacement in the relative coordinate system (such as the distance traveled along the direction of travel) or the change in pose (such as a combination of translation and rotation) is used as the sampling basis, then the metric value between adjacent keyframes is approximately equal. It is understood that a keyframe is determined when the vehicle's motion changes by a certain distance or angle. This avoids the collection of a large amount of redundant data due to the vehicle stopping or moving slowly at a certain position, thus avoiding a waste of computational resources. It also prevents the loss of important environmental structures due to sparse sampling. Furthermore, it ensures that the constructed local map or semantic model is spatially uniformly distributed, which is beneficial for subsequent registration, stitching, and topology consistency maintenance.

[0057] In step S130, based on the timestamp information of the key frames, the absolute positioning information and geospatial data are extracted and matched to determine the absolute positioning information and geospatial data corresponding to each key frame.

[0058] It should be noted that the relative positioning information, absolute positioning information, and geospatial data collected by the vehicle all contain corresponding time information, that is, the correspondence between the sampling time and the sampling data. Based on the timestamp information of the key frame, the corresponding absolute positioning information and geospatial data at the same sampling time can be determined, extracted, and matched, and compressed into the absolute positioning information and geospatial data corresponding to the key frame, so as to align data from various data sources and reduce the amount of data.

[0059] In step S140, crowdsourced map data is generated based on the timestamp information of the key frame, as well as the absolute positioning information and geospatial data corresponding to the key frame.

[0060] It should be noted that the timestamp information records the precise moment corresponding to the keyframe, which can be used to establish the time series relationship of the data and support alignment and synchronization with other vehicle or cloud data in the time dimension. Through the timestamp information of the keyframe, absolute positioning information, geospatial data, and relative positioning information carried by the keyframe can be spatiotemporally aligned and fused to aggregate and generate crowdsourced map data with wide coverage and rich content.

[0061] In step S150, the crowdsourced map data is uploaded to the cloud.

[0062] It should be noted that the generated crowdsourced map data can be uploaded to the cloud for the construction, updating or distribution of high-precision maps, providing reliable and dynamically evolving environmental prior information for subsequent vehicle assisted parking, route planning and positioning.

[0063] As analyzed above, this embodiment of the present disclosure, in response to a vehicle entering a parking area, collects the vehicle's relative positioning information, absolute positioning information, and geospatial data during the vehicle's movement; based on the relative positioning information, it determines multiple keyframes; wherein the difference between the relative positioning of any two adjacent keyframes is the same; based on the timestamp information of the keyframes, it extracts and correlates the absolute positioning information and geospatial data to determine the absolute positioning information and geospatial data corresponding to each keyframe; based on the timestamp information of the keyframes, and the absolute positioning information and geospatial data corresponding to the keyframes, it generates crowdsourced map data; and it uploads the crowdsourced map data to the cloud. By using the various data collected from the vehicle side, determining keyframes based on relative positioning information, and determining the absolute positioning information and geospatial data corresponding to the keyframes, a large amount of collected data is simplified to obtain less but more effective crowdsourced map data. This ensures that the amount of data uploaded by the vehicle is small, does not occupy vehicle-side resources and traffic, and ensures that the cloud can generate accurate maps when using the generated crowdsourced map data for mapping.

[0064] In some embodiments of this disclosure, a crowdsourced map data processing method is provided, further comprising: in response to determining that a vehicle is heading towards a target parking lot, determining a parking area based on the location of the target parking lot. That is, the vehicle's heading towards the target parking lot is determined through vehicle perception, and the parking area is delineated based on the location of the target parking lot, which may be its entrance. Accordingly, the implementation of step S140 may include: in response to completing a parking operation in the target parking lot, uploading the crowdsourced map data to the cloud. Specifically, after the vehicle drives to the target parking space in the target parking lot and parking is completed, the crowdsourced map data generated from the process of entering the parking area to parking is uniformly uploaded to the cloud, thereby reducing communication frequency and bandwidth consumption, and ensuring that the uploaded content completely covers the entire parking path, facilitating structured parsing, fusion, and version management in the cloud.

[0065] In some embodiments of this disclosure, a crowdsourced map data processing method is provided, further comprising: in response to determining that a vehicle has left a target parking lot, determining a parking area based on the location of the target parking lot. That is, the vehicle's departure from the target parking lot is determined through vehicle perception, and the parking area is delineated based on the location of the target parking lot, which could be its exit location. Accordingly, the implementation of step S140 may include: in response to the vehicle leaving the parking area, uploading the crowdsourced map data to the cloud. Specifically, from the vehicle's departure from the target parking space in the target parking lot until it exits the boundary of the parking area, the crowdsourced map data generated during the journey is uniformly uploaded to the cloud, thereby reducing communication frequency and bandwidth consumption, and ensuring that the uploaded content completely covers the entire parking path, facilitating structured parsing, fusion, and version management in the cloud.

[0066] In some embodiments of this disclosure, the implementation process of step S130 is as follows: Figure 2 As shown, the steps include the following.

[0067] In step S210, point cloud data within the time window from the previous key frame to the current key frame is extracted based on the timestamp information of the key frame.

[0068] In step S220, the extracted point cloud data is spatially downsampled, accumulated, and fused to obtain processed point cloud data.

[0069] In step S230, the processed point cloud data is compressed to obtain the point cloud data corresponding to the current keyframe.

[0070] In some embodiments of this disclosure, a time window is defined based on the timestamp information of the determined key frames: the timestamp corresponding to the previous key frame is taken as the starting point and the timestamp corresponding to the current key frame is taken as the ending point. All point cloud data continuously collected by LiDAR or other 3D sensors within the time window are extracted. The time window covers the complete perception process of the vehicle from the previous representative pose to the current pose, ensuring the continuity and integrity of environmental information.

[0071] In practice, spatial dimensionality reduction algorithms can be used to downsample high-density point clouds to eliminate local redundancy. The downsampled point clouds are then uniformly transformed to the local coordinate system of the current keyframe based on the relative poses of the vehicles at each acquisition time, and then accumulated and fused to form an aggregated point cloud that covers the entire time window, is non-repeating, and geometrically consistent. This significantly reduces the number of points and enhances the spatial coherence and structural clarity of the point cloud.

[0072] In practice, a general point cloud compression standard or a custom binary format can be used to efficiently encode attributes such as point coordinates, intensity, and semantic labels, thereby generating compressed point cloud data that is smaller in size and easier to store or upload. The compression result is then used as the final point cloud representation corresponding to the current keyframe for subsequent map building, feature extraction, or cloud crowdsourcing uploads.

[0073] It should be noted that when a vehicle is driving in a parking lot, the speed is relatively low, and there may be a lot of overlap and redundancy between point cloud data collected at adjacent times. The above steps can effectively extract representative environmental information, remove redundancy, and generate a compact and structured point cloud representation as the point cloud data corresponding to the key frame.

[0074] This disclosure, through the processing of point cloud data, effectively suppresses the redundancy of point cloud data while ensuring the accuracy of environmental perception, and significantly reduces the consumption of computing, storage, and communication resources, providing key technical support for efficient and scalable map generation and updating.

[0075] In other embodiments of this disclosure, the implementation process of step S130 is as follows: Figure 3 As shown, the steps include the following.

[0076] In step S310, vector information from the previous key frame to the current key frame is extracted based on the timestamp information of the key frame.

[0077] In step S320, the extracted vector information is spatiotemporally aligned and geometrically fused to generate processed vector information.

[0078] In step S330, redundancy is removed from the processed vector information to obtain the vector information corresponding to the current keyframe.

[0079] In some embodiments of this disclosure, the timestamp of the previous keyframe is used as the starting point and the timestamp of the current keyframe is used as the ending point. All raw vector information output by the vehicle-mounted perception system is extracted from this time window. This information may come from image semantic segmentation, laser point cloud fitting, or high-precision positioning fusion modules, and is usually represented as structured elements in the environment in the form of line segments, polygons, point sets, or parametric curves.

[0080] In practice, based on the vehicle pose at the corresponding acquisition time, including relative and / or absolute positioning, all vector elements are uniformly transformed to the reference coordinate system of the current keyframe to achieve spatial alignment. For the same type of geographic feature, such as a lane line or a parking space, the results of multiple observations within the time window are geometrically fused, for example through least-squares fitting, spline interpolation, or topological consistency verification, to generate a smoother, more accurate, and geometrically continuous fused vector representation. This effectively eliminates vector jitter or breakage caused by sensor noise, local occlusion, or transient false detections.

[0081] In practice, vector objects with highly overlapping spatial locations or the same semantics are identified and merged, such as the same pillar or the same stop line detected multiple times. Invalid, isolated or confidence-lower-threshold vector segments are removed. Based on preset topology rules, such as parking spaces must be closed and lane lines should be parallel, legality verification and simplification are performed. The processed vector information output is the deredundant and structured vector information corresponding to the current keyframe.

[0082] It should be noted that since vector information is typically continuously output by the perception module in units of events or frames, during low-speed vehicle movement or parking, vector features extracted at adjacent moments may exhibit issues such as duplication, slight offsets, or inconsistent representation. Therefore, steps S310 to S330 can be executed to achieve temporal integration, spatial optimization, and redundancy suppression of the vector information.

[0083] This embodiment of the disclosure achieves efficient aggregation and optimization of dynamically acquired vector information through processes such as redundancy removal of vector information. This not only improves the geometric accuracy and semantic integrity of map elements during a single parking process, but also provides high-quality and standardized input for subsequent cloud-based fusion of multi-vehicle crowdsourced data, thereby better supporting the automatic construction and continuous evolution of high-precision parking lot maps.

[0084] It should be noted that image data can also be collected while the vehicle is in motion and integrated into crowdsourced data transmitted to the cloud. This allows the cloud to better perform semantic understanding, geometric verification, and dynamic updates of map elements. Understandably, image data can include high-resolution environmental images captured by front-view, surround-view, or rear-view cameras. These images contain rich texture, color, and semantic information, such as ground marking types, parking space numbers, no-parking signs, traffic signs, and parking lot elevator entrances. This information may be difficult to fully represent or may be ambiguous in point cloud data or vector information.

[0085] Accordingly, in some embodiments of this disclosure, the crowdsourced map data processing method and the crowdsourced map data further include image data. By incorporating image data into the crowdsourced map data, not only is the semantic layer of the map enriched, but a more comprehensive data foundation is also provided for realizing the "perception-mapping-service" closed loop, significantly improving the construction quality and update efficiency of cloud-based high-precision maps in complex indoor parking environments.

[0086] like Figure 4 The diagram shown is a flowchart illustrating a crowdsourced map data processing method according to an exemplary embodiment of this disclosure. Figure 2 Steps S410 to S430 and step S470 in the diagram are... Figure 1 Steps S110 to S130 and step S150 correspond to each other and will not be repeated here. Figure 1 In addition to the implementation process shown, the following steps are also included.

[0087] In step S440, the vehicle's driving state is determined in response to the vehicle driving in a preset image acquisition area.

[0088] In step S450, the image acquisition angle is determined based on the vehicle's driving status, and image data is acquired based on the image acquisition angle according to a preset driving distance interval.

[0089] In step S460, crowdsourced map data is generated based on the timestamp information of the key frame, the absolute positioning information and geospatial data corresponding to the key frame, and the image data.

[0090] It should be noted that although image data has rich texture and semantic information, the amount of image data is large. A portion of the parking area can be selected as the image acquisition area, that is, key areas can be selected for image acquisition. The selection rules can be adjusted according to actual needs, and this embodiment does not limit them here.

[0091] In some embodiments of this disclosure, in response to the vehicle sensing that it has traveled within a preset image acquisition area, the current driving state of the vehicle is determined. The driving state includes at least a forward driving state, i.e., a normal parking driving state, such as the vehicle driving forward along the lane towards the target parking space; and a reverse parking state, i.e., a reverse parking maneuver, such as the vehicle performing a reverse or manual reverse operation. The driving state can be comprehensively determined through vehicle control signals, such as gear information, steering angle, speed change trend, and the output results of the vehicle perception module, such as the relative position of obstacles and trajectory curvature.

[0092] In some embodiments of this disclosure, the implementation of step S450 may include: if the vehicle is in a reversing parking state, the image acquisition view is a rear-view acquisition view; if the vehicle is in a forward driving state, the image acquisition view is a forward-view acquisition view.

[0093] In some embodiments of this disclosure, when the vehicle is traveling in a forward direction, the passageway within the parking lot can be divided into multiple segments. An intersection refers to a crossroads, a turn, or a key structural node, such as the start or end point of a ramp, or the location of a gate. The passageway between two adjacent intersections is defined as a segment. For each segment, if its length does not exceed a preset maximum threshold, such as 50 meters, a forward view image is collected and retained only at the starting point of the segment, i.e., the previous intersection. If the segment length exceeds the preset maximum threshold, such as 50 meters, a forward view image is collected every fixed distance within the segment, such as 10 meters, 20 meters, or 25 meters, to ensure that key structural features are effectively covered.

[0094] When the vehicle is in reverse parking mode, since the vehicle mainly relies on the rear view to complete the parking operation, rear view images can be continuously collected to refine the details of the parking space. A small number of high-value rear view images can also be collected during the reversing process, such as key moments such as starting to reverse, aligning with the parking space, and before coming to a complete stop, to avoid redundancy caused by continuous high-frequency collection.

[0095] In practice, the collected image data is fused with the aforementioned multi-source information to generate complete crowdsourced map data. Specifically, this crowdsourced map data includes: timestamp information of each keyframe, corresponding absolute positioning information of the keyframe, geospatial data, and multiple collected image data.

[0096] This embodiment of the disclosure selectively acquires images from specific areas and perspectives, significantly reducing the total amount of image data that needs to be uploaded while ensuring coverage of key scenes. This greatly reduces the storage pressure on the vehicle and the uplink communication load, while also ensuring that the image data uploaded to the cloud has high information density and high task relevance, providing high-quality input for subsequent cloud-based tasks such as map feature recognition, semantic annotation, change detection, and visual relocation.

[0097] like Figure 5 The image shown is a specific example of crowdsourced map data generation and uploading based on the crowdsourced map data processing method provided in the above embodiments, according to an embodiment of this disclosure.

[0098] As shown in the image, when a vehicle enters or exits the target parking lot, the parking area is defined as the space between the exit / entrance and the designated parking space. Relative location information, absolute location information, and geospatial data are collected from the moment the vehicle reaches 200 meters from the parking lot entrance until it parks in its designated space. This process is shown in the image as the start of data collection and the data upload process after the vehicle reaches its parking space. Since the image data is used to understand the internal conditions of the parking lot, image data collection can begin after the vehicle enters the parking lot. After the vehicle parks in its designated space, the generated crowdsourced map data is uploaded to the cloud. For example, Figure 5 When a vehicle enters the entrance, image data recording begins and continues until the image data is uploaded at the designated storage location.

[0099] As the vehicle departs, after the vehicle is powered on and started, it begins collecting relative positioning information, absolute positioning information, and geospatial data until it has traveled 200 meters from the parking lot exit. Figure 5 As shown, data collection begins at the parking space and continues until the vehicle is 200 meters away from the parking lot exit, at which point the data is uploaded. Image data is collected until the vehicle exits the parking lot. Since it's unnecessary to collect image data for the final 200 meters, the image data can be transmitted to the cloud as soon as the vehicle leaves the parking lot exit. In other words, image data recording begins at the parking space, and the image data is uploaded after the vehicle exits the parking lot. After the relative positioning information, absolute positioning information, and geospatial data are collected, the remaining parts of the crowdsourced map data are uploaded to the cloud. The aforementioned 200 meters is an example; the specific value can be adjusted according to actual needs, and this disclosure does not impose any limitations on it.

[0100] This disclosure also provides a crowdsourced map data processing method that can be applied to the cloud, including: receiving crowdsourced map data uploaded by vehicles.

[0101] It should be noted that the cloud can receive crowdsourced map data uploaded by at least one vehicle. The crowdsourced map data uploaded by each vehicle includes: multiple keyframes, as well as the absolute positioning information and geospatial data corresponding to each keyframe.

[0102] Crowdsourced map data is generated from vehicles through the following steps: In response to a vehicle entering a parking area, the system collects the vehicle's relative positioning information, absolute positioning information, and geospatial data during the vehicle's movement. Based on relative positioning information, multiple key frames are determined; among them, the difference between the relative positioning of the vehicles corresponding to any two adjacent key frames is the same. Based on the timestamp information of key frames, absolute positioning information and geospatial data are extracted and correlated to determine the absolute positioning information and geospatial data corresponding to each key frame. Crowdsourced map data is generated based on the timestamp information of the keyframes, as well as the absolute positioning information and geospatial data corresponding to the keyframes.

[0103] In some embodiments of this disclosure, the crowdsourced map data further includes image data. The image data is collected by the vehicle through the following steps: The vehicle's driving status is determined in response to the vehicle driving within the preset image acquisition area; The image acquisition angle is determined based on the vehicle's driving status, and image data is acquired according to the preset driving distance interval based on the image acquisition angle.

[0104] In some embodiments of this disclosure, the crowdsourced map data processing method further includes: performing interpolation and coordinate transformation based on the timestamp information in the image data and the timestamp information of the key frames to obtain image recognition data associated with each key frame.

[0105] It is understandable that the frequency of image acquisition may be lower than the frequency of keyframes. In other words, the time of image acquisition may not correspond one-to-one with the time of keyframes. For example, keyframes may be generated at intervals of 1 meter or 0.5 seconds, while images may be acquired only on demand in specific road sections or under specific conditions, with a large sampling interval. At the time corresponding to most keyframes, there is no directly synchronized image data.

[0106] To ensure spatiotemporal alignment, a time-space interpolation and coordinate transformation mechanism can be introduced. Based on the vehicle's pose trajectory over continuous time, including relative and absolute positioning, the vehicle's motion state between two adjacent acquired images can be modeled. If any image is located between the timestamps of two keyframes, the vehicle's pose is estimated through time interpolation. By determining the poses of objects or the environment identified in the image, the coordinates are transformed from the camera coordinate system at the original acquisition time, through the vehicle coordinate system, to the global or local reference coordinate system corresponding to the two keyframes, thereby generating image recognition data virtually aligned with the keyframes.

[0107] In some embodiments of this disclosure, the image recognition data may include: the projection position of the original image pixels in the keyframe coordinate system for multi-view fusion; and the extracted structured semantic elements, such as the detected stop line endpoints, parking space corners, and pillar centers, which are then embedded as vector elements in the geospatial data of the keyframe after coordinate transformation.

[0108] As can be seen from the above process, even if the images collected by the vehicle are not directly captured at the key frame moment, the cloud can still associate high-confidence, geometrically consistent image recognition results with each key frame, achieving deep fusion of multimodal data. This not only compensates for the information loss caused by sparse image sampling, but also enhances the performance of crowdsourced maps in terms of semantic integrity, geometric accuracy, and topological consistency, while avoiding the storage and bandwidth burden caused by high-frequency image collection by the vehicle in pursuit of strict synchronization.

[0109] In some embodiments of this disclosure, the provided crowdsourced map data processing method further includes: aligning, fusing, and structurally modeling the crowdsourced map data to generate a map of the target parking lot.

[0110] It should be noted that the cloud receives crowdsourced map data uploaded from different vehicles. Each data set contains multimodal information, including keyframe timestamps, absolute positioning information, relative pose, point clouds, vector features, and associated image recognition results. Since the time, path, and sensor status of each vehicle entering the same parking lot may differ, there may be slight deviations in the data within the global coordinate system. High-precision, high-confidence absolute positioning points near parking lot entrances and exits with good GNSS signals can be used as geographic anchors to uniformly calibrate all trajectories to a unified geographic coordinate system. The aligned multi-source data undergoes spatiotemporal aggregation and consistency optimization: repeatedly observed environmental elements have their geometric parameters fused to generate a more robust and smoother representation; conflicting observations, such as abnormal vectors caused by occlusion or false detections, are automatically identified and removed based on confidence scores, frequency of occurrence, and topological rationality; and image recognition results, such as ground text "B2-035" and no-parking signs, are semantically bound to the fused geometric structure to enhance the map's information dimension. During the structured modeling phase, the cloud transforms the fused raw perception data into a structured map model with clear semantics and topological relationships.

[0111] This enables the cloud to generate a map of the target parking lot with high accuracy and rich semantics, which can be distributed to subsequent vehicles entering the parking lot through the vehicle-cloud collaboration mechanism for high-precision positioning, parking path planning, or scene understanding.

[0112] Regarding the methods in the above embodiments, the specific manner of execution has been described in detail in the embodiments of the crowdsourced map data processing method applied to vehicles, and will not be elaborated here.

[0113] It should be noted that the acquisition, storage, use, and processing of information or data in this disclosed technical solution comply with the relevant provisions of national laws and regulations.

[0114] This disclosure also provides a crowdsourced map data processing device. Figure 6 This is a structural block diagram of a crowdsourced map data processing apparatus according to some embodiments of the present disclosure. Figure 6 As shown, the crowdsourced map data processing device 600 applied to vehicles includes: a data acquisition unit 601, a keyframe determination unit 602, a data processing unit 603, a map data generation unit 604, and a map data sending unit 605.

[0115] Among them, the data acquisition unit 601 is used to collect the relative positioning information, absolute positioning information and geospatial data of the vehicle during the vehicle's driving process in response to the vehicle entering the parking area. The keyframe determination unit 602 is used to determine multiple keyframes based on relative positioning information; wherein, the difference between the relative positioning of the vehicles corresponding to any two adjacent keyframes is the same. The data processing unit 603 is used to extract and correspond absolute positioning information and geospatial data based on the timestamp information of key frames, and to determine the absolute positioning information and geospatial data corresponding to each key frame. The map data generation unit 604 is used to generate crowdsourced map data based on the timestamp information of the key frame, as well as the absolute positioning information and geospatial data corresponding to the key frame. The map data sending unit 605 is used to upload crowdsourced map data to the cloud.

[0116] In some exemplary embodiments of this disclosure, the provided crowdsourced map data processing apparatus further includes: a parking area determination unit, configured to: determine a parking area based on the location of the target parking lot in response to determining that a vehicle is heading towards a target parking lot. Accordingly, the map data sending unit 605 is configured to: upload the crowdsourced map data to the cloud in response to completing a parking operation in the target parking lot.

[0117] In some exemplary embodiments of this disclosure, the provided crowdsourced map data processing apparatus further includes: a parking area determination unit, configured to: determine a parking area based on the location of the target parking lot in response to determining that a vehicle has left the target parking lot. Accordingly, the map data sending unit 605 is configured to: upload the crowdsourced map data to the cloud in response to the vehicle leaving the parking area.

[0118] In some exemplary embodiments of this disclosure, geospatial data includes one or any combination of point cloud data, topological data, and vector information.

[0119] In some exemplary embodiments of this disclosure, the data processing unit 603 is configured to: Based on the timestamp information of keyframes, extract point cloud data within the time window from the previous keyframe to the current keyframe. The extracted point cloud data is spatially downsampled, accumulated, and fused to obtain the processed point cloud data. The processed point cloud data is compressed to obtain the point cloud data corresponding to the current keyframe.

[0120] In some exemplary embodiments of this disclosure, the data processing unit 603 is configured to: Based on the timestamp information of the keyframes, extract the vector information from the previous keyframe to the current keyframe; The extracted vector information is spatiotemporally aligned and geometrically fused to generate processed vector information; Redundancy is removed from the processed vector information to obtain the vector information corresponding to the current keyframe.

[0121] In some exemplary embodiments of this disclosure, the crowdsourced map data further includes: image data. Accordingly, the crowdsourced map data processing apparatus further includes: an image data generation unit, used for: The vehicle's driving status is determined in response to the vehicle driving within the preset image acquisition area; The image acquisition angle is determined based on the vehicle's driving status, and image data is acquired according to the preset driving distance interval based on the image acquisition angle.

[0122] In some exemplary embodiments of this disclosure, the image data generation unit is configured as follows: If the vehicle is in a reverse parking state, the image acquisition perspective is the rear view acquisition perspective. If the vehicle is traveling in a forward direction, the image acquisition perspective is the forward-looking perspective.

[0123] This disclosure also provides a crowdsourced map data processing device. Figure 7 This is a structural block diagram of a crowdsourced map data processing apparatus according to some embodiments of the present disclosure. Figure 7 As shown, the crowdsourced map data processing device 700 applied to the cloud includes: a map data receiving unit 701.

[0124] The map data receiving unit 701 is used to receive crowdsourced map data uploaded by vehicles. The crowdsourced map data includes: multiple keyframes, as well as the absolute positioning information and geospatial data corresponding to each keyframe; Crowdsourced map data is generated from vehicles through the following steps: In response to a vehicle entering a parking area, the system collects the vehicle's relative positioning information, absolute positioning information, and geospatial data during the vehicle's movement. Based on relative positioning information, multiple key frames are determined; among them, the difference between the relative positioning of the vehicles corresponding to any two adjacent key frames is the same. Based on the timestamp information of key frames, absolute positioning information and geospatial data are extracted and correlated to determine the absolute positioning information and geospatial data corresponding to each key frame. Crowdsourced map data is generated based on the timestamp information of the keyframes, as well as the absolute positioning information and geospatial data corresponding to the keyframes.

[0125] In some exemplary embodiments of this disclosure, the crowdsourced map data further includes: image data. The image data is collected by the vehicle through the following steps: The vehicle's driving status is determined in response to the vehicle driving within the preset image acquisition area; The image acquisition angle is determined based on the vehicle's driving status, and image data is acquired according to the preset driving distance interval based on the image acquisition angle. In some exemplary embodiments of this disclosure, the provided crowdsourced map data processing apparatus further includes: an image data processing unit, configured to: Based on the timestamp information in the image data, and combined with the timestamp information of the key frames, interpolation and coordinate transformation are performed to obtain the image recognition data associated with each key frame.

[0126] In some exemplary embodiments of this disclosure, the provided crowdsourced map data processing apparatus further includes: a map generation unit, used to: align, fuse, and structure the crowdsourced map data to generate a map of the target parking lot.

[0127] Figure 8 This is a block diagram illustrating a vehicle 800 according to an exemplary embodiment. For example, vehicle 800 can be a hybrid vehicle, a non-hybrid vehicle, an electric vehicle, a fuel cell vehicle, or other types of vehicles. Vehicle 800 can be an intelligent driving vehicle, a semi-intelligent driving vehicle, or a non-intelligent driving vehicle.

[0128] Reference Figure 8 The vehicle 800 may include various subsystems, such as an infotainment system 810, a perception system 820, a decision control system 830, a drive system 840, and a computing platform 850. The vehicle 800 may also include more or fewer subsystems, and each subsystem may include multiple components. Furthermore, each subsystem and each component of the vehicle 800 can be interconnected via wired or wireless means.

[0129] In some embodiments, the infotainment system 810 may include a communication system, an entertainment system, and a navigation system, etc.

[0130] The perception system 820 may include several sensors for sensing information about the environment surrounding the vehicle 800. For example, the perception system 820 may include a global positioning system (which may be GPS, BeiDou, or other positioning systems), an inertial measurement unit (IMU), lidar, millimeter-wave radar, ultrasonic radar, and a camera device.

[0131] The decision control system 830 may include a computing system, a vehicle controller, a steering system, a throttle, and a braking system.

[0132] The drive system 840 may include components that provide powered motion to the vehicle 800. In one embodiment, the drive system 840 may include an engine, an energy source, a transmission system, and wheels. The engine may be one or a combination of internal combustion engines, electric motors, and compressed air engines. The engine is capable of converting energy provided by the energy source into mechanical energy.

[0133] Some or all of the functions of the vehicle 800 are controlled by a computing platform 850. The computing platform 850 may include at least one processor 851 and a memory 852, the processor 851 being able to execute instructions 853 stored in the memory 852.

[0134] Processor 851 can be any conventional processor. Processors may also include, for example, a Graphic Processing Unit (GPU), a Field Programmable Gate Array (FPGA), a System on Chip (SOC), an Application Specific Integrated Circuit (ASIC), or a combination thereof.

[0135] The memory 852 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0136] In addition to instruction set 853, memory 852 can also store data, such as road maps, route information, vehicle position, direction, speed, and other data. The data stored in memory 852 can be used by computing platform 850.

[0137] In this embodiment of the disclosure, processor 851 may execute instruction 853 to complete all or part of the steps of the crowdsourced map data processing method described above.

[0138] In some embodiments of this disclosure, an electronic device is provided, including a remote server and a vehicle-side processor. The remote server is located in the cloud and can be configured to implement all or part of the steps of the crowdsourced map data processing method applied to the cloud. The vehicle-side processor is located in a vehicle and can be configured to implement all or part of the steps of the crowdsourced map data processing method applied to the vehicle.

[0139] In some embodiments of this disclosure, a non-transitory computer-readable storage medium is provided, which, when the instructions in the storage medium are executed by a vehicle's processor, enables the vehicle to perform the crowdsourced map data processing method described above.

[0140] In some embodiments of this disclosure, a computer program product is provided, including a computer program that, when executed by a processor, implements the crowdsourced map data processing method described above.

[0141] The above description is merely a specific embodiment of this disclosure, enabling those skilled in the art to understand or implement it. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this disclosure. Therefore, this disclosure is not to be limited to the embodiments described herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A crowdsourced map data processing method, characterized in that, include: In response to a vehicle entering a parking area, the relative positioning information, absolute positioning information, and geospatial data of the vehicle during its driving process are collected. Based on the relative positioning information, multiple key frames are determined; wherein, the difference between the relative positioning of the vehicles corresponding to any two adjacent key frames is the same. Based on the timestamp information of the key frame, the absolute positioning information and the geospatial data are extracted and correlated to determine the absolute positioning information and geospatial data corresponding to each key frame. Crowdsourced map data is generated based on the timestamp information of the keyframes, as well as the absolute positioning information and geospatial data corresponding to the keyframes. The crowdsourced map data is uploaded to the cloud.

2. The crowdsourced map data processing method according to claim 1, characterized in that, Also includes: In response to determining that the vehicle is heading toward a target parking lot, the parking area is determined based on the location of the target parking lot; Uploading the crowdsourced map data to the cloud includes: Upon completion of parking in the target parking lot, the crowdsourced map data is uploaded to the cloud.

3. The crowdsourced map data processing method according to claim 1, characterized in that, Also includes: In response to determining that the vehicle has left the target parking lot, the parking area is determined based on the location of the target parking lot; Uploading the crowdsourced map data to the cloud includes: In response to the vehicle leaving the parking area, the crowdsourced map data is uploaded to the cloud.

4. The crowdsourced map data processing method according to claim 1, characterized in that, The geospatial data includes one or any combination of point cloud data, topological data, and vector information.

5. The crowdsourced map data processing method according to claim 4, characterized in that, Based on the timestamp information of the keyframes, the absolute positioning information and the geospatial data are extracted and correlated to determine the absolute positioning information and geospatial data corresponding to each keyframe, including: Based on the timestamp information of the key frame, extract the point cloud data within the time window from the previous key frame to the current key frame. The extracted point cloud data is spatially downsampled, accumulated, and fused to obtain the processed point cloud data. The processed point cloud data is compressed to obtain the point cloud data corresponding to the current keyframe.

6. The crowdsourced map data processing method according to claim 4, characterized in that, Based on the timestamp information of the keyframes, the absolute positioning information and the geospatial data are extracted and correlated to determine the absolute positioning information and geospatial data corresponding to each keyframe, including: Based on the timestamp information of the key frame, extract the vector information from the previous key frame to the current key frame; The extracted vector information is spatiotemporally aligned and geometrically fused to generate processed vector information; Redundancy is removed from the processed vector information to obtain the vector information corresponding to the current keyframe.

7. The crowdsourced map data processing method according to claim 1 or 4, characterized in that, The crowdsourced map data also includes: image data; The crowdsourced map data processing method further includes: In response to the vehicle traveling within a preset image acquisition area, the vehicle's driving status is determined; The image acquisition angle is determined based on the vehicle's driving state, and image data is acquired according to the preset driving distance interval based on the image acquisition angle.

8. The crowdsourced map data processing method according to claim 7, characterized in that, Determining the image acquisition perspective based on the vehicle's driving state includes: If the vehicle is in a reverse parking state, the image acquisition perspective is a rear-view acquisition perspective; If the vehicle is traveling in a forward direction, the image acquisition perspective is a forward-looking acquisition perspective.

9. A crowdsourced map data processing method, characterized in that, include: Receive crowdsourced map data uploaded by vehicles; The crowdsourced map data includes: multiple keyframes, and absolute positioning information and geospatial data corresponding to each keyframe; The crowdsourced map data is generated from the vehicle through the following steps: In response to a vehicle entering a parking area, the relative positioning information, absolute positioning information, and geospatial data of the vehicle during its driving process are collected. Based on the relative positioning information, multiple key frames are determined; wherein, the difference between the relative positioning of the vehicles corresponding to any two adjacent key frames is the same. Based on the timestamp information of the key frame, the absolute positioning information and the geospatial data are extracted and correlated to determine the absolute positioning information and geospatial data corresponding to each key frame. The crowdsourced map data is generated based on the timestamp information of the keyframes, as well as the absolute positioning information and geospatial data corresponding to the keyframes.

10. The crowdsourced map data processing method according to claim 9, characterized in that, The crowdsourced map data also includes: image data; The image data was collected by the vehicle through the following steps: In response to the vehicle traveling within a preset image acquisition area, the vehicle's driving status is determined; The image acquisition angle is determined based on the vehicle's driving state, and the image data is acquired according to the preset driving distance interval based on the image acquisition angle.

11. The crowdsourced map data processing method according to claim 10, characterized in that, Also includes: Based on the timestamp information in the image data, and combined with the timestamp information of the key frames, interpolation and coordinate transformation are performed to obtain the image recognition data associated with each key frame.

12. The crowdsourced map data processing method according to any one of claims 9 to 11, characterized in that, Also includes: The crowdsourced map data is aligned, merged, and structured to generate a map of the target parking lot.

13. A crowdsourced map data processing device, characterized in that, include: The data acquisition unit is used to collect the relative positioning information, absolute positioning information, and geospatial data of the vehicle during its driving process in response to the vehicle entering the parking area. A keyframe determination unit is used to determine multiple keyframes based on the relative positioning information; wherein the difference between the relative positioning of the vehicles corresponding to any two adjacent keyframes is the same. The data processing unit is used to extract and correlate the absolute positioning information and the geospatial data based on the timestamp information of the key frame, and to determine the absolute positioning information and geospatial data corresponding to each key frame. The map data generation unit is used to generate crowdsourced map data based on the timestamp information of the key frame, as well as the absolute positioning information and geospatial data corresponding to the key frame. The map data sending unit is used to upload the crowdsourced map data to the cloud.

14. The crowdsourced map data processing device according to claim 13, characterized in that, The crowdsourced map data also includes: image data; The crowdsourced map data processing device further includes: an image data generation unit, used for: In response to the vehicle traveling within a preset image acquisition area, the vehicle's driving status is determined; The image acquisition angle is determined based on the vehicle's driving state, and image data is acquired according to the preset driving distance interval based on the image acquisition angle.

15. A crowdsourced map data processing device, characterized in that, include: The map data receiving unit is used to receive crowdsourced map data uploaded by vehicles. The crowdsourced map data includes: multiple keyframes, and absolute positioning information and geospatial data corresponding to each keyframe; The crowdsourced map data is generated from the vehicle through the following steps: In response to a vehicle entering a parking area, the relative positioning information, absolute positioning information, and geospatial data of the vehicle during its driving process are collected. Based on the relative positioning information, multiple key frames are determined; wherein, the difference between the relative positioning of the vehicles corresponding to any two adjacent key frames is the same. Based on the timestamp information of the key frame, the absolute positioning information and the geospatial data are extracted and correlated to determine the absolute positioning information and geospatial data corresponding to each key frame. The crowdsourced map data is generated based on the timestamp information of the keyframes, as well as the absolute positioning information and geospatial data corresponding to the keyframes.

16. The crowdsourced map data processing apparatus according to claim 15, characterized in that, The crowdsourced map data also includes: image data; The image data was collected by the vehicle through the following steps: In response to the vehicle traveling within a preset image acquisition area, the vehicle's driving status is determined; The image acquisition angle is determined based on the vehicle's driving state, and the image data is acquired based on the image acquisition angle according to a preset driving distance interval. The crowdsourced map data processing device further includes: an image data processing unit, used for: Based on the timestamp information in the image data, and combined with the timestamp information of the key frames, interpolation and coordinate transformation are performed to obtain the image recognition data associated with each key frame.

17. A vehicle, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to implement the crowdsourced map data processing method according to any one of claims 1 to 8.

18. An electronic device, characterized in that, include: Remote server and vehicle-side processor; The remote server is configured to implement the steps of the crowdsourced map data processing method according to any one of claims 9 to 12. The vehicle-mounted processor is configured to implement the steps of the crowdsourced map data processing method according to any one of claims 1 to 8.

19. A non-transitory computer-readable storage medium, wherein instructions in the storage medium, when executed by a processor of a terminal, enable the terminal to perform the steps of a crowdsourced map data processing method according to any one of claims 1 to 8 or any one of claims 9 to 12.

20. A computer program product, characterized in that, It includes a computer program that, when executed by a processor, implements the crowdsourced map data processing method as described in any one of claims 1 to 8 or the crowdsourced map data processing method as described in any one of claims 9 to 12.

Images