Map construction and vehicle control method, device, equipment, vehicle and medium

Abstract

The disclosure provides a map construction and vehicle control method, device, equipment, vehicle and medium, and relates to the field of map construction.The map construction method comprises the following steps: obtaining multi-dimensional features of each first track point in a plurality of vehicle driving tracks located in the same geographical range; wherein the multi-dimensional features at least include an elevation observation value and a distance along the road; performing outlier removal and down-sampling processing on the first track point according to the multi-dimensional features of each first track point and local density to obtain a second track point; generating a smooth elevation curve of the same road according to the distance along the road and the elevation observation value of the second track point on the same road in the same geographical range; and obtaining slope map data of the same geographical range in the slope map according to the smooth elevation curve of each road in the same geographical range. Therefore, the dependence on professional surveying and mapping equipment is significantly reduced, low-cost, wide-coverage and high-frequency-updating slope map production is realized, and the road surface elevation representation accuracy is improved.

Description

Technical Field

[0001] This disclosure relates to the field of map building technology, and in particular to a map building and vehicle control method, apparatus, equipment, vehicle and medium. Background Technology

[0002] With the rapid development of intelligent driving and advanced driver assistance systems (ADAS), the demand for high-precision road attribute data is becoming increasingly urgent. Road slope (longitudinal slope) is a key input parameter for longitudinal control, energy consumption estimation (especially electric vehicle range prediction), and safe braking distance calculation.

[0003] In related technologies, the construction of slope maps relies on professional surveying vehicles to collect road elevation data, which is costly (millions of dollars per vehicle), has limited coverage, and has a low update frequency. Summary of the Invention

[0004] This disclosure proposes a map building and vehicle control method, apparatus, device, vehicle, and medium to at least partially solve one of the technical problems in the related art.

[0005] One embodiment of this disclosure proposes a map construction method, comprising: acquiring multidimensional features of each first trajectory point in multiple vehicle driving trajectories located within the same geographical area; wherein the multidimensional features include at least the elevation observation value and road distance of the first trajectory point; the road distance is used to indicate the cumulative path distance from the starting point of the vehicle driving trajectory to the first trajectory point; performing outlier removal and downsampling processing on the first trajectory points based on the multidimensional features and local density of each first trajectory point to obtain second trajectory points; wherein the local density is used to characterize the density of surrounding trajectory points of the first trajectory point; generating a smoothed elevation curve of the same road based on the road distance and elevation observation value of the second trajectory points on the same road within the same geographical area; and obtaining slope map data of the same geographical area in a slope map based on the smoothed elevation curves of each road within the same geographical area.

[0006] Therefore, by acquiring the multidimensional features (including elevation observations and road distances) of each first trajectory point in multiple vehicle driving trajectories within the same geographical area, and combining local density to remove outliers and downsample each first trajectory point, second trajectory points are obtained. This effectively eliminates elevation anomalies caused by GPS signal interference, signal obstruction, or urban canyon effects, while avoiding the impact of excessive redundancy or uneven distribution of trajectory points on the accuracy of elevation curve generation. Based on this, smooth elevation curves are generated using the road distances and elevation observations of the second trajectory points, ensuring the continuity and stability of road elevation representation. Finally, slope map data is constructed based on the smooth elevation curves of each road, which not only significantly reduces the dependence on professional surveying equipment and achieves low-cost, wide-coverage, and high-frequency updateable slope map production, but also has higher road surface elevation representation accuracy in complex road sections such as bridges, tunnels, and overpasses compared to slope calculation methods based on coarse-resolution DEMs or simple elevation averages in related technologies, thus providing more reliable longitudinal environmental perception input for intelligent driving systems.

[0007] As one possible implementation, the multidimensional features also include geographical location. The step of performing outlier removal and downsampling on the first trajectory points based on their multidimensional features and local density to obtain second trajectory points includes: determining first trajectory points located on any road within the same geographical area; wherein each road is divided into multiple road segments; performing outlier removal on the first trajectory points on the same road segment based on their road-side distance to obtain candidate trajectory points; responding to the existence of a target road segment among the multiple road segments, performing downsampling on the candidate trajectory points based on their local density to obtain the second trajectory point; wherein the number of candidate trajectory points on the target road segment is greater than a first predetermined number.

[0008] Therefore, by removing outliers and downsampling at the road segment level, elevation misjudgment caused by the mixing of trajectory points across roads or road segments is avoided. At the same time, downsampling is only performed on target road segments where the number of candidate trajectory points exceeds a first set number, which not only ensures the data integrity of sparse road segments, but also effectively controls the data redundancy of high-density road segments. This ensures the accuracy of elevation curve generation while improving computational efficiency and reducing storage overhead.

[0009] As one possible implementation, the step of removing outliers from the first trajectory points on the same road segment based on their road-side distances to obtain candidate trajectory points includes: sorting the first trajectory points on the same road segment in ascending order based on their road-side distances to obtain a first sorting sequence; determining an upper elevation limit and a lower elevation limit based on the elevation observation values ​​of each first trajectory point in the first sorting sequence; and filtering the first trajectory points on the same road segment based on the upper elevation limit and the lower elevation limit to obtain the candidate trajectory points; wherein the elevation observation values ​​of the candidate trajectory points are located between the upper elevation limit and the lower elevation limit.

[0010] In summary, by sorting the first trajectory points in ascending order based on their distance along the road within the same road segment, and dynamically determining the upper and lower elevation limits based on the distribution characteristics of the elevation observations of the sorted first trajectory points, adaptive and unsupervised removal of abnormal elevation observations is achieved. Therefore, without the need for a preset fixed threshold, it can effectively identify and filter out outlier elevation observations caused by GPS multipath effects, signal jumps, or urban canyons, while retaining normal elevation observations that conform to the actual longitudinal slope trend of the road segment. Compared to global thresholding or simple mean filtering, this disclosure achieves stronger robustness and adaptability to different slope characteristics (such as gentle and steep road segments), significantly improving the quality of candidate trajectory points.

[0011] As one possible implementation, the step of downsampling the candidate trajectory points based on the local density of the candidate trajectory points on the target road segment to obtain the second trajectory point includes: determining the k nearest neighbors of each candidate trajectory point based on the elevation observation values ​​of the candidate trajectory points on the target road segment; where k is a positive integer greater than 1; determining the local density of each candidate trajectory point based on the average distance from each candidate trajectory point to the k nearest neighbors; where the local density is negatively correlated with the average; sorting the candidate trajectory points on the target road segment in descending order based on the local density to obtain a second sorting sequence; and taking the second predetermined number of candidate trajectory points at the top of the second sorting sequence as the second trajectory point.

[0012] Therefore, by defining local density inversely based on the mean of k-nearest neighbor distances, and sorting and truncating downsampling candidate trajectory points accordingly, it is possible to prioritize the retention of spatially more centrally distributed trajectory points (i.e., those with closer and higher density neighbors) in dense trajectory point areas, while eliminating marginal or isolated trajectory points. Under the premise of ensuring that key elevation information is not lost, data redundancy is effectively suppressed, the computational complexity of subsequent elevation curve fitting is reduced, and the reliability and representativeness of subsequent smoothed elevation curves are enhanced.

[0013] As one possible implementation, before performing outlier removal and downsampling on the first trajectory points based on their multidimensional features and local density, the method further includes: resampling the road geometry of any road to obtain multiple first sampling points; wherein the first sampling points are located within the same geographical area; dividing the road based on the multiple first sampling points to obtain multiple intervals; determining a target interval covering the first trajectory points from the multiple intervals based on their multidimensional features; determining the coverage rate of the elevation observation value of any road based on the proportion of the target interval; and determining whether to generate and output the slope data corresponding to any road based on the relationship between the coverage rate and a set probability.

[0014] Therefore, by resampling the road geometry to generate multiple first sampling points before outlier removal and downsampling, the road is divided into multiple intervals. Then, based on the multidimensional features of the first trajectory points, the target intervals they cover are determined, and the proportion of the target intervals is calculated as the coverage rate of elevation observations. This allows for a quantitative assessment of the spatial coverage integrity of the driving trajectories of crowdsourced vehicles, objectively reflecting whether a road has received sufficient elevation observations from trajectory points. This avoids forcibly outputting road slope data when elevation observations are sparse or locally missing, thus ensuring the reliability and representativeness of subsequent slope map generation. Optionally, when the coverage rate is lower than a set probability, the generation and output of slope data for that road can be actively rejected. This effectively prevents slope estimation distortion caused by insufficient trajectory point coverage (such as only some road sections having elevation observations, or the existence of large blind spots), significantly improving the quality controllability and reliability of the slope map, ensuring that the slope data of each output road is based on sufficient observations.

[0015] As one possible implementation, generating a smooth elevation curve for the same road based on the road-side distance and elevation observations of the second trajectory points on the same road within the same geographical area includes: determining hyperparameters in a Gaussian process regression model, using the road-side distance of the second trajectory points on the same road as the input variable and the elevation observations of the second trajectory points on the same road as the output variable; determining the mean elevation prediction of multiple prediction points within the road-side distance range based on the hyperparameters, the road-side distance, and the elevation observations of the second trajectory points on the same road; wherein the road-side distance range is the minimum to maximum value among the road-side distances of the second trajectory points on the same road; and generating a smooth elevation curve for the same road based on the mean elevation prediction of the multiple prediction points.

[0016] Therefore, by adopting a Gaussian process regression model, using the road distance of the second trajectory point as the input variable and the elevation observation value as the output variable, and optimizing the model hyperparameters, the spatial correlation between the second trajectory points can be fully utilized to perform probabilistic and non-parametric smooth interpolation of the road elevation. Compared with the linear interpolation or spline fitting methods in related technologies, this disclosure effectively suppresses the oscillation or overfitting phenomenon caused by local noise or sparse distribution of the second trajectory points while maintaining the overall smoothness of the elevation curve, thereby generating an elevation curve that is more consistent with the actual longitudinal profile of the road and significantly improving the geometric quality of the slope calculation.

[0017] As one possible implementation, the Gaussian process regression model also outputs the prediction variance of the multiple prediction points. The method further includes: determining the confidence level of the smoothed elevation curve of the same road based on the prediction variance of the multiple prediction points; and determining the reliability of the slope data of the same road in the slope map data based on the confidence level of the smoothed elevation curve of the same road.

[0018] Therefore, by utilizing the prediction variance output by the Gaussian process regression model, the uncertainty of the elevation estimation at each prediction point can be quantified, and the confidence level of the entire smooth elevation curve can be constructed accordingly. This confidence level can then be transferred to the reliability assessment of the slope data of the corresponding road segment in the slope map. This enables a closed-loop transfer of uncertainty from data modeling to downstream applications, allowing the intelligent driving system to dynamically adjust its control strategy based on the reliability of the slope data (such as conservatively estimating energy consumption or braking distance in low-confidence areas). This significantly enhances the safety and robustness of the intelligent driving system in complex or insufficient elevation observation scenarios.

[0019] As one possible implementation, the input variable is the normalized distance along the road, the output variable is the normalized elevation observation value, and the step of generating a smooth elevation curve for the same road based on the average elevation prediction value of the multiple prediction points includes: performing inverse normalization on the average elevation prediction value of the multiple prediction points to obtain the elevation prediction value of the multiple prediction points; and generating a smooth elevation curve for the same road based on the elevation prediction value of the multiple prediction points.

[0020] Therefore, by normalizing the input variables (distance along the road) and the output variables (elevation observations), the impact of dimensional differences and numerical scale inconsistencies on the training and prediction stability of the Gaussian process regression model is effectively eliminated, improving the convergence speed of hyperparameter optimization and the model's generalization ability. After the prediction is completed, the mean of the elevation prediction is denormalized to restore the real physical units, ensuring the engineering usability of the final smooth elevation curve.

[0021] As one possible implementation, the slope map includes multiple tiles, each tile indicating a corresponding geographical area; the method further includes: determining neighboring tiles of the target tile belonging to the same geographical area from the multiple tiles based on the tile identifiers of the multiple tiles in the slope map; determining the common boundary road between the target tile and the neighboring tiles; and smoothing the slope data corresponding to the boundary road in the slope map data of the target tile and the slope map data of the neighboring tiles.

[0022] In summary, by actively identifying neighboring tiles and the common boundary roads between them after generating the slope map data of the target tile, and then performing cross-tile smoothing on the slope data on the boundary roads, the problem of boundary discontinuity caused by independent calculation during the tile-based construction of large-scale slope maps is effectively solved. This ensures that adjacent tiles have consistent slope representations on shared road geometry, avoiding slope jumps or breaks caused by tile cutting, thereby improving the spatial continuity and topological integrity of the overall slope map and providing a seamless and consistent foundation for longitudinal environmental perception in intelligent driving systems.

[0023] As one possible implementation, the smoothing process for the slope data corresponding to the boundary road in the slope map data of the target tile and the slope map data of the neighboring tiles includes: obtaining a first slope value of at least one second sampling point on the boundary road from the slope map data of the target tile; obtaining a second slope value of at least one third sampling point on the boundary road from the slope map data of the neighboring tiles; wherein the distance between the second sampling point and the third sampling point is less than a set distance; fusing the first slope value of the second sampling point and the second slope value of the third sampling point to obtain a fused slope value; and updating the first slope value in the slope map data of the target tile and the second slope value in the slope map data of the neighboring tiles according to the fused slope value.

[0024] This not only eliminates slope inconsistencies caused by minor geometric deviations or estimation differences resulting from independent tile modeling, but also enhances the robustness and accuracy of slope estimation for boundary roads through data fusion. Especially in cases of uneven coverage of multi-source trajectories or the presence of local noise, the cross-tile slope fusion mechanism can effectively suppress random errors in single-tile estimation, ensuring the smoothness and reliability of the slope map at a global scale.

[0025] As one possible implementation, the method further includes: responding to the map slope data indicating the existence of a fourth sampling point without a slope value on the boundary road, and the proportion of the fourth sampling point being greater than a set proportion, then interpolating and filling the slope value of the fourth sampling point to obtain the interpolated slope value of the fourth sampling point; and updating the slope data of the boundary road in the map slope data according to the interpolated slope value of the fourth sampling point.

[0026] In summary, the interpolation and filling mechanism effectively addresses the issue of missing slope data caused by gaps in trajectory coverage, data processing interruptions, or tile splicing. It avoids breaks in road longitudinal profile information due to the absence of slope values ​​at local sampling points, ensuring that the slope map maintains complete spatial coverage and continuous representation even at boundary roads. Furthermore, interpolation and filling based on nearby valid slope values ​​enables reasonable estimation of the slope of missing sampling points even in the absence of direct elevation observations, maintaining the consistency of road geometry. This improves the completeness of the slope map, providing usable elevation-derived information for downstream intelligent driving functions (such as longitudinal control and energy consumption prediction), significantly enhancing the robustness and reliability of intelligent driving systems in edge or weak observation scenarios.

[0027] As one possible implementation, the method for determining the same geographical area includes: determining an offset based on the number of binary bits of the tile identifier of the target tile to which the same geographical area belongs, and determining a map zoom level based on the difference between the number of binary bits and the offset; determining a target code for indicating the geographical location of the target tile based on the tile identifier, the map zoom level, and the offset; performing a bit interleaving inverse operation on the target code to decode the target code into two-dimensional grid coordinates; and determining the same geographical area based on the map zoom level and the two-dimensional grid coordinates.

[0028] In summary, by determining the offset based on the number of binary bits in the target tile's identifier, and then determining the map zoom level based on the difference between the number of binary bits and the offset, the hierarchical information of the tile can be directly parsed from its identifier itself, without relying on additional metadata. The target code is jointly determined using the tile identifier, map zoom level, and offset. By performing a bit-interleaving inverse operation on the target code, it is decoded into two-dimensional grid coordinates, achieving a deterministic mapping from tile identifiers to standardized spatial coordinates. Finally, the geographic extent of the target tile can be uniquely determined by combining the map zoom level and the two-dimensional grid coordinates. Therefore, the spatial location information of the target tile can be embedded within its encoding structure, making the geographic extent determination process completely self-contained, reversible, and computationally efficient. This avoids dependence on external indexes or configuration files, significantly improving the deployment flexibility and processing consistency of the slope map construction system in distributed or edge computing environments.

[0029] As one possible implementation, obtaining the multidimensional features of each first trajectory point in multiple vehicle driving trajectories located within the same geographical area includes: determining target road network data located within the same geographical area from map road network data, and constructing a road network topology map based on the target road network data; wherein, nodes in the road network topology map are used to indicate connection points between connected roads, and edges between nodes are used to indicate topological relationships between connected roads; obtaining multiple vehicle driving trajectories located within the same geographical area and containing elevation observation values, and sampling key points for each vehicle driving trajectory to obtain each first trajectory point in each vehicle driving trajectory; matching each first trajectory point to the road network topology map to obtain map matching results for each first trajectory point; wherein, the map matching results are used to indicate the road identifier of the road matched by the first trajectory point, the geographical location of the first trajectory point after projection onto the matched road, and the road distance of the first trajectory point; generating multidimensional features for each first trajectory point based on the map matching results for each first trajectory point.

[0030] Therefore, by extracting target road network data from map road network data and constructing a road network topology map, the vehicle driving trajectory containing elevation observations is sampled for key points and matched to the road network topology map to obtain the map matching results of each sampled trajectory point. Based on this, structured multidimensional features are generated, which can realize the accurate mapping of trajectory points from the original GPS coordinates to the road semantic space. This effectively solves the problem of road classification ambiguity caused by positioning drift or signal noise of the original trajectory points, ensuring that subsequent elevation modeling is strictly based on the actual driving road, avoiding invalid observation interference across roads or non-roads, and significantly improving the road-level accuracy and semantic consistency of slope map construction.

[0031] Another embodiment of this disclosure proposes a vehicle control method, comprising: obtaining slope map data of at least one planned route of the vehicle from a slope map; wherein the slope map is constructed using the map construction method described in the foregoing embodiment; and performing driving control on the vehicle based on the slope map data of the at least one planned route.

[0032] Therefore, by obtaining the slope data of the blocks covered by the vehicle's planned route from the slope map and directly using this slope data for vehicle driving control, a deep integration of refined road longitudinal slope prior information with vehicle dynamic control strategy is achieved, providing reliable input for energy management, power distribution and braking strategies, thereby significantly improving the predictability, energy efficiency and safety of vehicle control.

[0033] As one possible implementation, the step of controlling the vehicle's driving based on the slope map data of the tiles covered by the at least one planned route includes: determining the energy consumption information required for the vehicle to drive along the planned route based on the slope map data of the tiles covered by each planned route; determining a navigation route from each planned route based on the energy consumption information of each planned route; determining the slope data of at least one segment of the navigation route based on the slope map data of the tiles covered by the navigation route; and controlling the vehicle's driving based on the slope data of the at least one segment.

[0034] Therefore, by using slope map data based on the blocks covered by each planned route, the energy consumption information of each planned route is quantitatively calculated, and the navigation route with the lowest energy consumption is selected accordingly, realizing intelligent path decision-making oriented towards energy saving goals. Furthermore, for the selected navigation route, the fine slope data of at least one section is extracted, and the vehicle's driving control parameters (such as motor torque, braking pressure, cruising speed, etc.) are dynamically adjusted accordingly, so that the vehicle can execute the optimal control strategy under different working conditions such as uphill, downhill, or flat roads. This not only effectively extends the vehicle's driving range, but also improves driving comfort and active safety levels.

[0035] Another embodiment of this disclosure proposes a map building apparatus, comprising: a first acquisition module, configured to acquire multidimensional features of each first trajectory point in multiple vehicle driving trajectories located within the same geographical area; wherein the multidimensional features include at least the elevation observation value and road distance of the first trajectory point; the road distance is used to indicate the cumulative path distance from the starting point of the vehicle driving trajectory to the first trajectory point; a processing module, configured to perform outlier removal and downsampling processing on the first trajectory points based on the multidimensional features and local density of each first trajectory point to obtain second trajectory points; wherein the local density is used to characterize the density of surrounding trajectory points of the first trajectory point; a first generation module, configured to generate a smoothed elevation curve of the same road based on the road distance and elevation observation value of the second trajectory points on the same road within the same geographical area; and a second generation module, configured to obtain slope map data of the same geographical area in a slope map based on the smoothed elevation curves of each road within the same geographical area.

[0036] Therefore, by acquiring the multidimensional features (including elevation observations and road distances) of each first trajectory point in multiple vehicle driving trajectories within the same geographical area, and combining local density to remove outliers and downsample each first trajectory point, second trajectory points are obtained. This effectively eliminates elevation anomalies caused by GPS signal interference, signal obstruction, or urban canyon effects, while avoiding the impact of excessive redundancy or uneven distribution of trajectory points on the accuracy of elevation curve generation. Based on this, smooth elevation curves are generated using the road distances and elevation observations of the second trajectory points, ensuring the continuity and stability of road elevation representation. Finally, slope map data is constructed based on the smooth elevation curves of each road, which not only significantly reduces the dependence on professional surveying equipment and achieves low-cost, wide-coverage, and high-frequency updateable slope map production, but also has higher road surface elevation representation accuracy in complex road sections such as bridges, tunnels, and overpasses compared to slope calculation methods based on coarse-resolution DEMs or simple elevation averages in related technologies, thus providing more reliable longitudinal environmental perception input for intelligent driving systems.

[0037] As one possible implementation, the multidimensional feature further includes geographic location. The processing module is configured to: determine a first trajectory point located on any road within the same geographic area; wherein the any road is divided into multiple road segments; perform outlier removal processing on the first trajectory points on the same road segment based on the road distance of the first trajectory points on the same road segment to obtain candidate trajectory points; in response to the existence of a target road segment among the multiple road segments, perform downsampling processing on the candidate trajectory points based on the local density of the candidate trajectory points on the target road segment to obtain a second trajectory point; wherein the number of candidate trajectory points on the target road segment is greater than a first predetermined number.

[0038] Therefore, by removing outliers and downsampling at the road segment level, elevation misjudgment caused by the mixing of trajectory points across roads or road segments is avoided. At the same time, downsampling is only performed on target road segments where the number of candidate trajectory points exceeds a first set number, which not only ensures the data integrity of sparse road segments, but also effectively controls the data redundancy of high-density road segments. This ensures the accuracy of elevation curve generation while improving computational efficiency and reducing storage overhead.

[0039] As one possible implementation, the processing module is configured to: sort the first trajectory points on the same road segment in ascending order according to their distance along the road, to obtain a first sorting sequence; determine an upper elevation limit and a lower elevation limit based on the elevation observation values ​​of each first trajectory point in the first sorting sequence; and filter the first trajectory points on the same road segment according to the upper elevation limit and the lower elevation limit to obtain candidate trajectory points; wherein the elevation observation values ​​of the candidate trajectory points are located between the upper elevation limit and the lower elevation limit.

[0040] In summary, by sorting the first trajectory points in ascending order based on their distance along the road within the same road segment, and dynamically determining the upper and lower elevation limits based on the distribution characteristics of the elevation observations of the sorted first trajectory points, adaptive and unsupervised removal of abnormal elevation observations is achieved. Therefore, without the need for a preset fixed threshold, it can effectively identify and filter out outlier elevation observations caused by GPS multipath effects, signal jumps, or urban canyons, while retaining normal elevation observations that conform to the actual longitudinal slope trend of the road segment. Compared to global thresholding or simple mean filtering, this disclosure achieves stronger robustness and adaptability to different slope characteristics (such as gentle and steep road segments), significantly improving the quality of candidate trajectory points.

[0041] In another aspect of this disclosure, a vehicle control device is provided, comprising: a second acquisition module, configured to acquire slope map data of at least one planned route of the vehicle from a slope map; wherein the slope map is constructed using the map construction device described in the preceding embodiment; and a control module, configured to perform driving control on the vehicle based on the slope map data of the at least one planned route.

[0042] Therefore, by obtaining the slope data of the blocks covered by the vehicle's planned route from the slope map and directly using this slope data for vehicle driving control, a deep integration of refined road longitudinal slope prior information with vehicle dynamic control strategy is achieved, providing reliable input for energy management, power distribution and braking strategies, thereby significantly improving the predictability, energy efficiency and safety of vehicle control.

[0043] As one possible implementation, the control module is configured to: determine the energy consumption information required for the vehicle to travel along the planned route based on the slope map data of the tiles covered by each planned route; determine a navigation route from each planned route based on the energy consumption information of each planned route; determine the slope data of at least one segment of the navigation route based on the slope map data of the tiles covered by the navigation route; and perform driving control on the vehicle based on the slope data of the at least one segment.

[0044] Therefore, by using slope map data based on the blocks covered by each planned route, the energy consumption information of each planned route is quantitatively calculated, and the navigation route with the lowest energy consumption is selected accordingly, realizing intelligent path decision-making oriented towards energy saving goals. Furthermore, for the selected navigation route, the fine slope data of at least one section is extracted, and the vehicle's driving control parameters (such as motor torque, braking pressure, cruising speed, etc.) are dynamically adjusted accordingly, so that the vehicle can execute the optimal control strategy under different working conditions such as uphill, downhill, or flat roads. This not only effectively extends the vehicle's driving range, but also improves driving comfort and active safety levels.

[0045] This disclosure also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the map building method as described in one aspect above, or the vehicle control method as described in another aspect above.

[0046] In another aspect of this disclosure, a vehicle is provided, comprising: a processor; and a memory for storing processor-executable instructions; wherein the processor is configured to implement the vehicle control method as described in the foregoing other aspect.

[0047] Another aspect of this disclosure provides a chip including an interface circuit and a processing circuit coupled to each other. The interface circuit is used to input or output signals, and the processing circuit is configured to perform a map building method as described in one aspect above, or to perform a vehicle control method as described in another aspect above.

[0048] In another aspect of this disclosure, a non-transitory computer-readable storage medium is provided, having stored thereon computer program instructions that, when executed by a processor, implement the map construction method as described in the foregoing aspect, or, when executed, implement the vehicle control method as described in the foregoing aspect.

[0049] Another aspect of this disclosure provides a computer program product having a computer program stored thereon, which, when executed by a processor, implements the map construction method as described in the foregoing aspect, or, when executed, implements the vehicle control method as described in the foregoing aspect.

[0050] Additional aspects and advantages of this disclosure will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this disclosure. Attached Figure Description

[0051] The above and / or additional aspects and advantages of this disclosure will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, in which: Figure 1A schematic flowchart of a map construction method provided for an exemplary embodiment of this disclosure; Figure 2 A schematic flowchart of another map construction method provided for an exemplary embodiment of this disclosure; Figure 3 A schematic flowchart of yet another map construction method provided as an exemplary embodiment of this disclosure; Figure 4 A schematic flowchart of another map construction method provided as an exemplary embodiment of the present disclosure; Figure 5 A schematic flowchart of another map construction method provided for an exemplary embodiment of this disclosure; Figure 6 A schematic flowchart of another map construction method provided for an exemplary embodiment of this disclosure; Figure 7 A schematic flowchart of yet another map construction method provided as an exemplary embodiment of this disclosure; Figure 8 A schematic diagram illustrating a map construction principle provided for an exemplary embodiment of this disclosure; Figure 9 A schematic diagram of a process for calculating slope segment by segment provided for an exemplary embodiment of this disclosure; Figure 10 A schematic flowchart of a vehicle control method provided for an exemplary embodiment of the present disclosure; Figure 11 A schematic diagram of the structure of a map building apparatus provided for an exemplary embodiment of the present disclosure; Figure 12 A schematic diagram of the structure of a vehicle control device provided for an exemplary embodiment of the present disclosure; Figure 13 A schematic diagram of the structure of an electronic device provided for an exemplary embodiment of the present disclosure; Figure 14 A block diagram illustrating a vehicle according to an exemplary embodiment; Figure 15 This is a schematic diagram of the structure of a chip proposed as an exemplary embodiment of the present disclosure. Detailed Implementation

[0052] Embodiments of this disclosure are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this disclosure, and should not be construed as limiting this disclosure.

[0053] It should be noted that the acquisition, storage, use, and processing of data in this disclosed technical solution comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0054] It should also be noted that all data processed in this disclosure is data that has been explicitly authorized by users or relevant parties, and has been de-identified or anonymized before collection and use, and does not contain any personally identifiable information or user privacy content; all data is used only for map construction and vehicle control purposes, ensuring that data security and user privacy rights are fully protected while achieving technical effects.

[0055] In related technologies, the following two methods are mainly used to construct slope maps: The first method, based on a Digital Elevation Model (DEM), involves overlaying the road centerline onto the DEM raster data. Elevation observations are extracted from the corresponding raster values ​​at the road's sampling points. The slope is then calculated by the ratio of the difference between elevation observations from adjacent sampling points to the horizontal distance. A typical workflow includes: obtaining the road centerline coordinates → sampling elevation observations from the DEM → calculating the slope using the finite difference method.

[0056] However, this method has obvious limitations: on the one hand, the spatial resolution of DEM is usually 30 meters or coarser, which cannot reflect the fine slope changes at the road level (50-meter interval); on the other hand, DEM represents the natural surface elevation, not the actual road surface elevation, which will produce systematic bias in areas with artificial structures such as bridges, tunnels, and overpasses, resulting in distorted slope estimation.

[0057] The second method is a simple statistical mean method based on mass-source trajectories: collect multiple Global Positioning System (GPS) trajectories that pass through the same road segment, use map matching technology to attach the trajectory points in the GPS trajectories to the road network topology, set sampling points at fixed intervals on the road segment, take the arithmetic mean of the elevation observations of all trajectory points falling into the same sampling interval, and then calculate the slope through the average elevation difference between adjacent sampling points.

[0058] However, this method has at least the following drawbacks: it does not effectively filter out abnormal elevation observations from GPS trajectories. Due to factors such as signal multipath and urban canyon effects, the elevation data of GPS trajectories often contains significant noise, and these abnormal elevation observations will seriously pollute the mean estimation; the density of trajectory points varies greatly in different sampling intervals, and the mean estimation in sparse areas has a large variance and low reliability; it lacks uncertainty quantification capabilities and cannot provide confidence or error range for slope estimation for downstream applications (such as vehicle control and energy consumption prediction); tiled processing leads to discontinuous boundaries. When constructing large-scale slope maps in parallel, each tile is calculated independently, and the slope values ​​at the boundaries of adjacent tiles may jump, destroying the spatial consistency of the map.

[0059] Therefore, in view of at least one of the problems existing in the above-mentioned related technologies, this disclosure proposes a map generation and vehicle control method, apparatus, device, vehicle and medium.

[0060] The following description, with reference to the accompanying drawings, outlines a map construction and vehicle control method, apparatus, device, vehicle, and medium according to embodiments of the present disclosure.

[0061] Figure 1 This is a schematic flowchart of a map construction method provided for an exemplary embodiment of the present disclosure.

[0062] It should be noted that the map building method of this disclosure can be applied to a map building apparatus. In some possible embodiments, the map building apparatus can be configured in an electronic device or chip so that the electronic device or chip can perform map building functions. Additionally, in some possible embodiments, the map building apparatus can also be software within an electronic device.

[0063] Among them, electronic devices include, but are not limited to: terminals, servers (or cloud, servers, etc.). Terminals can be cars with communication functions, smart cars, mobile phones, wearable devices, computers, tablets (Pads), computers with wireless transceiver functions, supercomputers, etc.

[0064] like Figure 1 As shown, the map construction method may include the following steps S101 to S104: Step S101: Obtain multidimensional features of each first trajectory point in multiple vehicle driving trajectories located within the same geographical area; wherein, the multidimensional features include at least the elevation observation value and the distance along the road of the first trajectory point.

[0065] The aforementioned same geographical range can be the geographical range of any tile in the slope map, or it can be the geographical range of a tile specified by the user. This disclosure does not limit the way tiles are divided in the slope map. For example, tiles include, but are not limited to, tiles, grids, raster, geohash units, etc.

[0066] The multidimensional features include, but are not limited to: longitude, latitude, elevation observation value, timestamp, confidence level, distance along the road, and road signs of the first trajectory point. The distance along the road indicates the cumulative path distance from the starting point of the vehicle's travel trajectory to the first trajectory point. The elevation observation value can be a relative elevation or a non-relative elevation; this disclosure does not impose any limitations on this.

[0067] In this embodiment of the disclosure, multiple vehicle driving trajectories located within the same geographical area can be determined from a massive amount of vehicle driving trajectories, and multidimensional features of each first trajectory point in the multiple vehicle driving trajectories can be obtained.

[0068] Step S102: Based on the multidimensional features and local density of each first trajectory point, outlier removal and downsampling are performed on each first trajectory point to obtain the corresponding second trajectory point; wherein, local density is used to characterize the density of the surrounding trajectory points of the first trajectory point.

[0069] In this embodiment of the disclosure, the local density of each first trajectory point can be calculated, and outlier removal and downsampling processing can be performed on the first trajectory points based on the multidimensional features and local density of each first trajectory point to obtain the retained second trajectory points.

[0070] As an example, based on multidimensional features, first trajectory points with excessively small or large elevation observations can be removed, and first trajectory points with relatively low local density can be filtered out to obtain the retained second trajectory points.

[0071] Step S103: Generate a smooth elevation curve for the same road based on the road distance and elevation observation values ​​of the second trajectory point on the same road within the same geographical area.

[0072] As an example, the elevation observations of the second trajectory point on the same road within the same geographical area can be curve-fitted to obtain a smooth elevation curve for the same road.

[0073] As another example, a Gaussian process regression model can be used to perform curve fitting based on the elevation observations of the second trajectory point and the distance along the road to obtain a smooth elevation curve. The Gaussian process regression model uses the distance along the road as the input variable and the elevation observations as the output variable. For example, the Gaussian process regression model uses the distance along the road as the input variable and the elevation observations as the output variable, employs a radial basis function kernel as the covariance function, and determines the hyperparameters of the radial basis function kernel by maximizing the marginal log-likelihood. Based on the determined hyperparameters, the distance along the road from the second trajectory point, and the elevation observations, the mean elevation predictions for multiple prediction points are output. Curve fitting is then performed based on the mean elevation predictions for multiple prediction points to obtain a smooth elevation curve for the same road.

[0074] Step S104: Based on the smoothed elevation curves of each road within the same geographical area, obtain the slope map data for the same geographical area in the slope map.

[0075] As an example, the slope data of the same road can be obtained by differentiating the smoothed elevation curve of the same road. Thus, slope map data of the same geographical area can be generated based on the slope data of each road within the same geographical area.

[0076] The map construction method of this disclosure acquires multidimensional features (including elevation observations and road distances) of each first trajectory point in multiple vehicle driving trajectories within the same geographical area, and performs outlier removal and downsampling processing on each first trajectory point using local density to obtain second trajectory points. This effectively eliminates elevation anomalies caused by GPS signal interference, signal obstruction, or urban canyon effects, while avoiding the impact of excessive redundancy or uneven distribution of trajectory points on the accuracy of elevation curve generation. Based on this, a smooth elevation curve is generated using the road distance and elevation observations of the second trajectory points, ensuring the continuity and stability of road elevation representation. Finally, slope map data is constructed based on the smooth elevation curves of each road, which not only significantly reduces the dependence on professional surveying equipment and achieves low-cost, wide-coverage, and high-frequency updateable slope map production, but also has higher road surface elevation representation accuracy in complex road sections such as bridges, tunnels, and overpasses compared to slope calculation methods based on coarse-resolution DEMs or simple elevation averages in related technologies, thus providing more reliable longitudinal environmental perception input for intelligent driving systems.

[0077] As one possible implementation method, Figure 2 This is a schematic flowchart illustrating another map construction method provided for an exemplary embodiment of the present disclosure. It should be noted that this map construction method can be executed alone, or it can be executed together with any embodiment of the present disclosure or possible implementations thereof, or it can be executed together with any technical solution in related technologies. The embodiments of the present disclosure do not impose any limitations on this.

[0078] like Figure 2 As shown, the map construction method may include the following steps S201 to S206: Step S201: Obtain multidimensional features of each first trajectory point in multiple vehicle driving trajectories located within the same geographical area; wherein, the multidimensional features include at least the elevation observation value, road distance, and geographical location of the first trajectory point.

[0079] The "distance along the road" indicates the cumulative path distance from the starting point of the vehicle's travel trajectory to the first trajectory point. It should be noted that the explanation of step S201 can be found in the relevant descriptions of any embodiment of this disclosure, and will not be repeated here.

[0080] Step S202: Determine the first trajectory point on any road within the same geographical area; wherein any road is divided into multiple road segments.

[0081] In this embodiment of the disclosure, a first trajectory point located on any road within the same geographical area can be determined based on the geographical location of each first trajectory point. This arbitrary road is divided into multiple segments; for example, the arbitrary road can be divided into multiple segments at equal intervals based on a set segment length.

[0082] The segment length is set to a pre-defined segment length. This disclosure does not limit the specific value of the segment length. For example, the value of the segment length can be 40 meters (m), 50 meters, 60 meters, etc.

[0083] Step S203: Based on the distance along the road of the first trajectory point on the same road segment, perform outlier removal processing on the first trajectory point on the same road segment to obtain candidate trajectory points.

[0084] In this embodiment of the disclosure, outlier removal processing can be performed on the first trajectory point on each road segment. That is, outlier removal processing can be performed on the first trajectory point on the same road segment based on the road distance of the first trajectory point on the same road segment to obtain the candidate trajectory points retained on the same road segment.

[0085] For example, based on the distance along the road from the first trajectory point on the same road segment, the first trajectory point on the same road segment with an excessively small or large elevation observation value can be removed to obtain the candidate trajectory points retained on the same road segment.

[0086] One possible implementation method for outlier removal is as follows: First trajectory points on the same road segment are sorted in ascending order based on their distance along the road, resulting in a first sorted sequence. Upper and lower elevation limits are determined based on the elevation observations of each trajectory point in the first sorted sequence. Then, the first trajectory points on the same road segment are filtered according to these upper and lower elevation limits to obtain candidate trajectory points. The elevation observations of these candidate trajectory points must fall between the upper and lower elevation limits.

[0087] As an example, the first element in the first sorted sequence can be used as the lower elevation limit, and the last element in the first sorted sequence can be used as the upper elevation limit.

[0088] As another example, the first and third quartiles can be determined based on the elevation observations of each first trajectory point in the first sorting sequence; the interquartile range can be determined based on the difference between the third and first quartiles; and the upper and lower elevation limits can be determined based on the first, third, and interquartile ranges.

[0089] For example, if the elevation observation value is labeled as y, the first quartile is Q1, and the third quartile is Q3, then: Q1 = percentile(y, 25), Q3 = percentile(y, 75). In this case, the interquartile range (IQR) = Q3 - Q1. In this disclosure, first trajectory points whose elevation observation value y satisfies the following condition can be filtered to obtain retained candidate trajectory points: (1) In summary, by sorting the first trajectory points in ascending order based on their distance along the road within the same road segment, and dynamically determining the upper and lower limits of elevation based on the distribution characteristics of the elevation observations of the sorted first trajectory points (such as the quartiles and interquartile ranges of the elevation observations), adaptive and unsupervised removal of abnormal elevation observations is achieved. Therefore, without the need for a preset fixed threshold, it can effectively identify and filter out outlier elevation observations caused by GPS multipath effects, signal jumps, or urban canyons, while retaining normal elevation observations that conform to the actual longitudinal slope trend of the road segment. Compared to global thresholding or simple mean filtering, this disclosure can achieve stronger robustness and adaptability to different slope characteristics (such as gentle and steep road segments), significantly improving the quality of candidate trajectory points.

[0090] Step S204: In response to the existence of a target road segment among multiple road segments, the candidate trajectory points are downsampled based on the local density of the candidate trajectory points on the target road segment to obtain the second trajectory point.

[0091] Local density is used to characterize the density of trajectory points surrounding the first trajectory point.

[0092] In this embodiment of the disclosure, the number of candidate trajectory points on each segment of any of the aforementioned roads can be counted. If a target segment exists on any of the roads, the candidate trajectory points can be downsampled based on their local density on the target segment to obtain a second trajectory point. The number of candidate trajectory points on the target segment is greater than a first predetermined number. The first predetermined number is a pre-set threshold value. This disclosure does not limit the specific value of the first predetermined number. For example, if the first predetermined number is denoted as N_max, then the value of N_max can be, for example, 40, 50, 60, etc.

[0093] Among them, candidate trajectory points on multiple road segments other than the target road segment can all be retained as second trajectory points.

[0094] For example, first trajectory points with relatively low local density on the target road segment can be filtered to obtain retained second trajectory points.

[0095] As one possible implementation method, the downsampling processing method is as follows: based on the elevation observation values ​​of each candidate trajectory point on the target road segment, determine the k nearest neighbor trajectory points of each candidate trajectory point; determine the local density of each candidate trajectory point according to the average distance from each candidate trajectory point to its k nearest neighbor trajectory points; based on the local density of each candidate trajectory point on the target road segment, sort the candidate trajectory points on the target road segment in descending order to obtain a second sorting sequence; and take the second set number (or set proportion) of candidate trajectory points that are ranked first in the second sorting sequence as the second trajectory points.

[0096] Where k is a positive integer greater than 1, and the local density of each candidate trajectory point is negatively correlated with the mean distance from that candidate trajectory point to its k nearest neighbors. For example, using the elevation observation value y as a feature to construct a KNN model, the local density of each candidate trajectory point is calculated as follows: (2) Among them, density i It refers to the local density of the i-th candidate trajectory point on the target road segment. This refers to the average distance from the i-th candidate trajectory point to its k nearest neighbors. The value of k can be, for example, k = min(5, number of points), where the number of points refers to the number of candidate trajectory points. It should be noted that... The purpose is to avoid the denominator being 0.

[0097] The second set quantity is also a pre-set quantity threshold. It should be noted that this disclosure does not restrict the size relationship between the first set quantity and the second set quantity. For example, the second set quantity can be greater than the first set quantity, or the second set quantity can be less than or equal to the first set quantity.

[0098] In summary, by defining local density inversely based on the mean of k-nearest neighbor distances, and sorting and truncating downsampling candidate trajectory points accordingly, we can prioritize retaining spatially more centrally distributed trajectory points (i.e., those with closer and higher density neighbors) in densely populated areas, while eliminating marginal or isolated trajectory points. This effectively suppresses data redundancy, reduces the computational complexity of subsequent elevation curve fitting, and enhances the reliability and representativeness of subsequent smoothed elevation curves, while ensuring that key elevation information is not lost.

[0099] Step S205: Generate a smooth elevation curve for the same road based on the road distance and elevation observation values ​​of the second trajectory point on the same road within the same geographical area.

[0100] Step S206: Based on the smoothed elevation curves of each road within the same geographical area, obtain the slope map data for the same geographical area in the slope map.

[0101] It should be noted that the explanations of steps S205 to S206 can be found in the relevant descriptions in any embodiment of this disclosure, and will not be repeated here.

[0102] The map construction method of this disclosure avoids elevation misjudgment caused by the mixing of trajectory points across roads or road segments by removing outliers and downsampling at the road segment granularity. At the same time, downsampling is only performed on target road segments where the number of candidate trajectory points exceeds a first set number, which not only ensures the data integrity of sparse road segments, but also effectively controls the data redundancy of high-density road segments. Thus, while ensuring the accuracy of elevation curve generation, it improves computational efficiency and reduces storage overhead.

[0103] As one possible implementation method, Figure 3 This is a schematic flowchart illustrating another map construction method provided as an exemplary embodiment of the present disclosure. It should be noted that this map construction method can be executed alone, or it can be executed together with any embodiment of the present disclosure or any possible implementation thereof, or it can be executed together with any technical solution in the related art. The embodiments of the present disclosure do not impose any limitations on this.

[0104] like Figure 3 As shown, the map construction method may include the following steps S301 to S312: Step S301: Obtain multidimensional features of each first trajectory point in multiple vehicle driving trajectories located within the same geographical area; wherein, the multidimensional features include at least the elevation observation value, road distance, and geographical location of the first trajectory point.

[0105] The distance along the road is used to indicate the cumulative path distance from the starting point of the vehicle's travel trajectory to the first trajectory point. It should be noted that the explanation of step S301 can be found in the relevant descriptions of any embodiment of this disclosure, and will not be repeated here.

[0106] Step S302: Resample the road geometry of any road within the same geographical area to obtain multiple first sampling points.

[0107] The first sampling point, also known as the effective sampling point, is located within the same geographical area mentioned above.

[0108] In this embodiment of the disclosure, road geometry can be resampled for any road within the same geographical area to obtain multiple resampled points. For each resampled point, it can be determined whether the geographical location of the resampled point is within the same geographical area. If so, it is taken as the first sampling point.

[0109] Step S303: Based on multiple first sampling points, any road is divided into multiple intervals.

[0110] For example, any road is divided into multiple intervals by using the distance between geographically adjacent first sampling points as the interval.

[0111] Step S304: Based on the multidimensional features of the first trajectory point, determine the target interval covering the first trajectory point from multiple intervals.

[0112] The target interval refers to the interval covered by elevation observations.

[0113] Step S305: Determine the coverage rate of any road to the elevation observation value based on the proportion of the target interval.

[0114] The percentage of the target interval = the number of target intervals / the total number of intervals.

[0115] In this embodiment of the disclosure, the proportion of the target interval can be used as the coverage rate of any road for elevation observations. For example, the formula for calculating coverage is as follows: coverage = |Number of intervals covered by elevation observations| / |Total number of intervals|; (3) In some embodiments, it can be determined whether to generate and output slope data corresponding to any road based on the relationship between the coverage rate of any road to the elevation observation and a set probability. The set probability is a pre-set probability threshold; for example, the set probability value could be 0.5, 0.6, etc. For example, if the coverage rate of any road to the elevation observation is greater than or equal to the set probability, it can be determined that slope data corresponding to that road needs to be generated and output, and in this case, subsequent steps S308 to S312 can be executed; conversely, if the coverage rate of any road to the elevation observation is less than the set probability, it can be determined that slope data corresponding to that road does not need to be generated and output, and in this case, subsequent step S307 can be executed.

[0116] Step S306: Determine whether the coverage rate is higher than or equal to the set probability. If yes, proceed to steps S308 to S312. If no, proceed to step S307.

[0117] It should be noted that step S307 and steps S308 to S312 are two possible implementations in parallel, and one of them can be selected for execution.

[0118] Step S307: Refuse to generate and output slope data corresponding to any road.

[0119] In this embodiment of the disclosure, if the coverage of any road to the elevation observation value is less than a set probability, that road can be skipped and the corresponding slope data can not be output, so as to avoid outputting unreliable slope data and improve the quality of map construction.

[0120] Understandably, when the coverage rate is lower than the set probability, actively refusing to generate and output the slope data of that road can effectively prevent the slope estimation distortion caused by insufficient coverage of trajectory points (such as only some road sections have elevation observation values, or there are large blind spots), significantly improving the quality controllability and credibility of the slope map, and ensuring that the slope data of each road output is based on sufficient observation.

[0121] Step S308: Determine the first trajectory point whose geographical location is on any road; wherein, any road is divided into multiple road segments.

[0122] In this embodiment of the disclosure, if the coverage of any road to the elevation observation value is greater than or equal to a set probability, a first trajectory point whose geographical location is on any road can be further determined.

[0123] Step S309: Based on the distance along the road of the first trajectory point on the same road segment, perform outlier removal processing on the first trajectory point on the same road segment to obtain candidate trajectory points.

[0124] Step S310: In response to the existence of a target road segment among multiple road segments, the candidate trajectory points are downsampled based on the local density of the candidate trajectory points on the target road segment to obtain the second trajectory point.

[0125] The number of candidate trajectory points on the target road segment is greater than a first predetermined number. Local density is used to characterize the density of trajectory points surrounding the first trajectory point.

[0126] Step S311: Generate a smooth elevation curve for the same road based on the road distance and elevation observation values ​​of the second trajectory point on the same road within the same geographical area.

[0127] Step S312: Based on the smoothed elevation curves of each road within the same geographical area, obtain the slope map data for the same geographical area in the slope map.

[0128] It should be noted that the explanations of steps S308 to S312 can be found in the relevant descriptions in any embodiment of this disclosure, and will not be repeated here.

[0129] The map construction method of this disclosure resamples the road geometry to generate multiple first sampling points before outlier removal and downsampling, and divides the road into multiple intervals accordingly. Then, it determines the target interval covered by the first trajectory points based on their multidimensional features and calculates the proportion of the target interval as the coverage rate of elevation observations. This can achieve a quantitative assessment of the spatial coverage integrity of the driving trajectory of crowdsourced vehicles, objectively reflecting whether a road has obtained sufficient elevation observations from trajectory points. It avoids forcibly outputting road slope data when elevation observations are sparse or locally missing, thereby ensuring the reliability and representativeness of the subsequent slope map generation.

[0130] As one possible implementation method, Figure 4 This is a schematic flowchart illustrating another map construction method provided as an exemplary embodiment of the present disclosure. It should be noted that this map construction method can be executed alone, or it can be executed together with any embodiment of the present disclosure or possible implementations thereof, or it can be executed together with any technical solution in related technologies. The embodiments of the present disclosure do not impose any limitations on this.

[0131] like Figure 4 As shown, the map construction method may include the following steps S401 to S406: Step S401: Obtain multidimensional features of each first trajectory point in multiple vehicle driving trajectories located within the same geographical area; wherein, the multidimensional features include at least the elevation observation value and the distance along the road of the first trajectory point.

[0132] Among them, the distance along the road is used to indicate the cumulative path distance from the starting point of the vehicle's travel trajectory to the first trajectory point.

[0133] Step S402: Based on the multidimensional features and local density of each first trajectory point, outlier removal and downsampling are performed on the first trajectory points to obtain the second trajectory points.

[0134] The local density is used to characterize the density of trajectory points surrounding the first trajectory point. It should be noted that explanations of steps S401 to S402 can be found in the relevant descriptions of any embodiment of this disclosure, and will not be repeated here.

[0135] Step S403: Based on the Gaussian process regression model, the hyperparameters in the Gaussian process regression model are determined by using the distance along the road from the second trajectory point on the same road as the input variable and the elevation observation value of the second trajectory point on the same road as the output variable.

[0136] As an example, the Gaussian Process Regression (GPR) model uses the distance along the road as the input variable and the elevation observation as the output variable. It uses the radial basis function kernel as the covariance function and determines the hyperparameters of the kernel function by maximizing the marginal log-likelihood.

[0137] For example, taking the radial basis function kernel as the Matern 2.5 kernel, the Matern 2.5 kernel generates second-order differentiable functions, suitable for modeling smooth but not infinitely differentiable natural terrain such as road elevation profiles. The formula for the Matern 2.5 kernel function is expressed as follows: (4) Where r = |x - x'| is the input distance, x is the position of the second trajectory point (distance along the road), and x' is the known input position of the observation point / reference point; l is the length-scale parameter, and σ² is the output scale parameter. Both l and σ² are hyperparameters of the Gaussian process regression model.

[0138] Step S404: Based on the hyperparameters, the distance along the road from the second trajectory point on the same road, and the observed elevation values, determine the average elevation prediction of multiple prediction points within the distance range along the road.

[0139] Wherein, the range of distances along the road is the minimum value x among the distances along the road from the second trajectory point on the same road. min up to the maximum value x max .

[0140] In this embodiment of the disclosure, a Gaussian process regression model can be used to predict the elevation values ​​of multiple prediction points within the distance range along the road based on hyperparameters, the distance along the road from the second trajectory point on the same road, and the observed elevation values, so as to obtain the average elevation prediction value of multiple prediction points.

[0141] Step S405: Generate a smooth elevation curve for the same road based on the average elevation prediction of multiple prediction points.

[0142] For example, curve fitting is performed using the distance along the road from multiple prediction points as the abscissa and the average elevation prediction of multiple prediction points as the ordinate to generate a smooth elevation curve for the same road.

[0143] As one possible implementation, the input variable of the Gaussian process regression model can be the normalized distance along the road, and the output variable can be the normalized elevation observation value. In this disclosure, for step S405, the mean elevation prediction of multiple prediction points can be denormalized to obtain the elevation prediction value of multiple prediction points, and a smooth elevation curve of the same road can be generated based on the elevation prediction value of multiple prediction points.

[0144] For example, the normalization method is as follows: normalize the distance x and elevation observation y along the road using the following formula (5): (5) in, μ is the normalized distance along the path. x , σ x Let σ be the mean and standard deviation of the distance x along the road (where σ is the mean and standard deviation of the distance x along the road). x Can be added (to prevent division by zero), similarly, μ is the normalized distance along the path. y , σ y These are the mean and standard deviation of the elevation observations, respectively (where σ is the mean and standard deviation of the elevation observations). y Can be added (to prevent division by zero).

[0145] For example, a smooth elevation curve can be generated by performing curve fitting with the distance along the road from multiple prediction points as the abscissa and the predicted elevation values ​​of multiple prediction points as the ordinate, in order to generate a smooth elevation curve for the same road.

[0146] In summary, by normalizing the input variables (distance along the road) and the output variables (elevation observations), the impact of dimensional differences and numerical scale inconsistencies on the training and prediction stability of the Gaussian process regression model is effectively eliminated, improving the convergence speed of hyperparameter optimization and the model's generalization ability. After the prediction is completed, the mean of the elevation prediction is denormalized to restore the real physical units, ensuring the engineering usability of the final smooth elevation curve.

[0147] As one possible implementation, the Gaussian process regression model can also output the prediction variance of multiple prediction points. For example, the prediction variance can be calculated in linear time complexity based on the Cholesky decomposition of the covariance matrix. In this disclosure, the confidence level of the smooth elevation curve of the same road can also be determined based on the prediction variance of multiple prediction points. Based on the confidence level of the smooth elevation curve of the same road, the reliability of the slope data of the same road in the slope map data can be determined.

[0148] In summary, by utilizing the prediction variance output by the Gaussian process regression model, the uncertainty of elevation estimation at each prediction point is quantified, and the confidence level of the entire smooth elevation curve is constructed accordingly. This confidence level is then transferred to the reliability assessment of the slope data of the corresponding road segment in the slope map. This enables a closed-loop transfer of uncertainty from data modeling to downstream applications, allowing the intelligent driving system to dynamically adjust its control strategy based on the reliability of the slope data (such as conservatively estimating energy consumption or braking distance in low-confidence areas). This significantly enhances the safety and robustness of the intelligent driving system in complex or insufficient elevation observation scenarios.

[0149] Step S406: Based on the smooth elevation curves of each road within the same geographical area, obtain the slope map data for the same geographical area in the slope map.

[0150] It should be noted that the explanation of step S406 can be found in the relevant description in any embodiment of this disclosure, and will not be repeated here.

[0151] The map construction method of this disclosure adopts a Gaussian process regression model, using the road distance of the second trajectory point as the input variable and the elevation observation value as the output variable, and optimizes the model hyperparameters. It can make full use of the spatial correlation between the second trajectory points to perform probabilistic and non-parametric smooth interpolation of the road elevation. Compared with the linear interpolation or spline fitting methods in related technologies, this disclosure effectively suppresses the oscillation or overfitting phenomenon caused by local noise or sparse distribution of the second trajectory points while maintaining the overall smoothness of the elevation curve. This generates an elevation curve that is more consistent with the actual longitudinal profile of the road, and significantly improves the geometric quality of the slope calculation.

[0152] As one possible implementation method, Figure 5This is a schematic flowchart illustrating another map construction method provided as an exemplary embodiment of the present disclosure. It should be noted that this map construction method can be executed alone, or it can be executed together with any embodiment of the present disclosure or any possible implementation thereof, or it can be executed together with any technical solution in the related art. The embodiments of the present disclosure do not impose any limitations on this.

[0153] like Figure 5 As shown, the map construction method may include the following steps S501 to S507: Step S501: Obtain multidimensional features of each first trajectory point in multiple vehicle driving trajectories within the geographical range of the target map; wherein, the multidimensional features include at least the elevation observation value and the distance along the road of the first trajectory point.

[0154] Among them, the distance along the road is used to indicate the cumulative path distance from the starting point of the vehicle's travel trajectory to the first trajectory point.

[0155] Step S502: Based on the multidimensional features and local density of each first trajectory point, outlier removal and downsampling are performed on the first trajectory points to obtain the second trajectory points.

[0156] Local density is used to characterize the density of trajectory points surrounding the first trajectory point.

[0157] Step S503: Generate a smooth elevation curve for the same road based on the road distance and elevation observation value of the second trajectory point on the same road within the geographical range of the target map block.

[0158] Step S504: Obtain the slope map data of the target tile in the slope map based on the smooth elevation curves of each road within the same geographical area; wherein, the slope map includes multiple tiles, and any tile is used to indicate the corresponding geographical area.

[0159] It should be noted that the explanations of steps S501 to S504 can be found in the relevant descriptions in any embodiment of this disclosure, and will not be repeated here.

[0160] Step S505: Based on the tile identifiers of multiple tiles in the slope map, determine the neighboring tiles of the target tile from the multiple tiles.

[0161] For example, the grid coordinates (x, y) of multiple tiles can be determined based on the tile identifiers of multiple tiles, and the 8-neighbor directions (dx, dy) ∈ {-1, 0, 1}² \ {(0,0)} can be traversed to determine the neighboring tiles of the target tile, where the grid coordinates of the neighboring tiles can be (x+dx, y+dy).

[0162] Step S506: Determine the common boundary roads between the target tile and its neighboring tiles.

[0163] Boundary roads refer to roads that appear in multiple tiles and are located at the boundaries of multiple tiles.

[0164] Step S507: Smooth the slope data corresponding to the boundary roads in the slope map data of the target tile and the slope map data of the neighboring tiles.

[0165] The generation principle of slope map data for neighboring tiles is similar to that for target tiles, and will not be elaborated here.

[0166] In this embodiment of the disclosure, a smoothing algorithm can be used to smooth the slope data corresponding to the boundary roads in the slope map data of the target tile and the slope map data of neighboring tiles, so as to avoid the occurrence of slope discontinuity at the boundary roads of adjacent tiles and improve the construction quality of the slope map.

[0167] As one possible implementation, the smoothing process could be as follows: Obtain the first slope value of at least one second sampling point on the boundary road from the slope map data of the target tile, and obtain the second slope value of at least one third sampling point on the boundary road from the slope map data of neighboring tiles; use a fusion algorithm to fuse the first slope value corresponding to the second sampling point and the second slope value of the third sampling point to obtain a fused slope value; update the first slope value in the slope map data of the target tile and the second slope value in the slope map data of neighboring tiles based on the fused slope value.

[0168] The distance between the second and third sampling points is less than a set distance, meaning the second and third sampling points are adjacent. The second sampling point is the boundary sampling point of the target tile, and the third sampling point is the boundary sampling point of the adjacent tile.

[0169] The fusion algorithms include, but are not limited to, weighted summation and mean averaging.

[0170] In summary, this approach not only eliminates slope inconsistencies caused by minor geometric deviations or estimation differences resulting from independent tile modeling, but also enhances the robustness and accuracy of slope estimation for boundary roads through data fusion. Especially in situations with uneven coverage of multi-source trajectories or the presence of local noise, the cross-tile slope fusion mechanism effectively suppresses random errors in single-tile estimation, ensuring the smoothness and reliability of the slope map at a global scale.

[0171] In any embodiment of this disclosure, when the slope map data of the target tile and / or neighboring tiles indicates the existence of a fourth sampling point without a slope value on the boundary road, the proportion of the fourth sampling point can be further calculated. The proportion of the fourth sampling point is calculated as: (Number of fourth sampling points / Total number of sampling points on the boundary road). If the proportion of the fourth sampling point is greater than a set proportion, the slope value of the fourth sampling point is interpolated and filled using an interpolation algorithm (such as linear interpolation or spline interpolation) to obtain the interpolated slope value of the fourth sampling point. Therefore, in this disclosure, the slope data of the boundary road in the slope map data can be updated based on the interpolated slope value of the fourth sampling point to improve the completeness of the slope map. That is, by using the interpolated slope value, the slope value of the fourth sampling point on the boundary road in the slope map data is assigned, resulting in updated slope map data.

[0172] The set percentage is a pre-set, relatively small percentage threshold. For example, the set percentage value is 9%, 10%, etc.

[0173] In other words, considering that the coverage of the first trajectory point in the vehicle trajectory data is limited, there may be invalid sampling points (referred to as the fourth sampling point in this disclosure) on some roads that are not covered by elevation observations. In order to improve the integrity of the road slope data, an interpolation algorithm can be used to interpolate and fill the slope of the invalid sampling points.

[0174] In summary, the interpolation and filling mechanism effectively addresses the issue of missing slope data caused by gaps in trajectory coverage, data processing interruptions, or tile splicing. It avoids breaks in road longitudinal profile information due to the absence of slope values ​​at local sampling points, ensuring that the slope map maintains complete spatial coverage and continuous representation even at boundary roads. Furthermore, interpolation and filling based on nearby valid slope values ​​enables reasonable estimation of the slope of missing sampling points even in the absence of direct elevation observations, maintaining the consistency of road geometry. This improves the completeness of the slope map, providing usable elevation-derived information for downstream intelligent driving functions (such as longitudinal control and energy consumption prediction), significantly enhancing the robustness and reliability of intelligent driving systems in edge or weak observation scenarios.

[0175] The map construction method of this disclosure, after generating slope map data for a target tile, actively identifies its neighboring tiles and the common boundary roads between them, and then performs cross-tile smoothing processing on the slope data on the boundary roads. This effectively solves the problem of boundary discontinuity caused by independent calculation during the tile-based construction of large-scale slope maps. Therefore, it ensures that adjacent tiles have consistent slope representations on shared road geometry, avoiding slope jumps or breaks caused by tile cutting, thereby improving the spatial continuity and topological integrity of the overall slope map and providing a seamless and consistent longitudinal environmental perception foundation for intelligent driving systems.

[0176] As one possible implementation method, Figure 6 This is a schematic flowchart illustrating another map construction method provided for an exemplary embodiment of the present disclosure. It should be noted that this map construction method can be executed alone, or it can be executed together with any embodiment of the present disclosure or possible implementations thereof, or it can be executed together with any technical solution in related technologies. The embodiments of the present disclosure do not impose any limitations on this.

[0177] like Figure 6 As shown, based on any of the above embodiments, the same geographical area can be determined using the following steps S601 to S604: Step S601: Determine the offset based on the number of binary bits of the tile identifier of the target tile belonging to the same geographical area, and determine the map zoom level based on the difference between the number of binary bits and the offset.

[0178] For example, if the map zoom level is set to zoom, the number of bits for the tile identifier is n bits, and the offset is offset, then: (6) Step S602: Determine the target code for indicating the geographic location of the target tile based on the tile identifier, map zoom level, and offset.

[0179] As an example, taking the image as tiles and the target encoding as Morton encoding, we have: (7) Here, tile_id refers to the tile identifier, and the flag bit is 1<<(zoom + offset).

[0180] Step S603: Perform a bit-interleaving inverse operation on the target encoding to decode the target encoding into two-dimensional grid coordinates.

[0181] Using the example above, a one-dimensional Morton code can be decoded into two-dimensional grid coordinates (xpart, ypart) through a bit interleaving inverse operation. For example, even-numbered bits can be extracted from the Morton code to obtain the x component, and even-numbered bits can be extracted from the Morton code after shifting it right by 1 bit to obtain the y component. The bit extraction operation can be achieved through 5 masking and shifting operations.

[0182] Step S604: Determine the same geographical area based on the map zoom level and two-dimensional grid coordinates.

[0183] As one possible approach, the geographic scale factor can be determined based on the map zoom level, where the geographic scale factor is negatively correlated with the map zoom level; and the geographic extent of the target tile can be determined based on the geographic scale factor and the values ​​of each coordinate component in the two-dimensional grid coordinates.

[0184] For example, if the geographic scale factor is labeled as "scale", then the geographic extent of the target tile can be represented as: [NW, NE, SE, SW, NW], where NW = (lon min lat max ) ← Top left, NE = (lon max lat max ) ← Top right, SE = (lon max lat min ) ← Bottom right, SW = (lon min lat min ) ← Bottom left, NW = (lon min lat max ) ← Return to the starting point, close the loop; where: (8) Here, lon is an abbreviation for longitude, which represents longitude (east-west direction); lat is an abbreviation for latitude, which represents latitude (north-south direction).

[0185] As another possible implementation, the geographic scale factor can be determined based on the map zoom level; the geographic scale factor is negatively correlated with the map zoom level; the geographic boundary of the target tile is determined based on the geographic scale factor, the values ​​and signs of each coordinate component in the two-dimensional grid coordinates; and the geographic boundary is expanded to obtain the same geographic range.

[0186] As an example, firstly, negative coordinates can be handled using two's complement: if ,but ;like ,but Next, calculate the geographic scale factor: (degree / grid cell); then, based on the geographic scale factor and (xpart, ypart), calculate the geographic boundary [NW, NE, SE, SW, NW] of the target tile; finally, expand the geographic boundary of the target tile to obtain the geographic range of the target tile, so as to ensure that the boundary road segments are completely covered.

[0187] For example, the center point (center_lon, center_lat) of the geographic boundary of the target map tile can be calculated; the vertices of the geographic boundary can be converted to coordinates in the ENU (East-North-Sky) coordinate system with the center point as the origin; an expansion operation can be performed on the geographic boundary in the ENU coordinate system to extend the geographic boundary outward by a set distance; and the vertices of the expanded geographic boundary can be converted back to the original coordinate system to obtain the geographic range of the target map tile.

[0188] In summary, by determining a geographic scale factor negatively correlated with the map zoom level and combining it with the values ​​and signs of the coordinate components of the two-dimensional grid coordinates, the geographic boundary of the target tile can be accurately calculated. Then, this boundary is expanded to obtain the final, unified geographic area. This effectively solves the problem of road segment truncation or incomplete coverage caused by roads crossing tile boundaries. Specifically, the expansion strategy ensures that even if the geometric part of a road is outside the tile edge, its key road segments can still be completely included within the geographic area of ​​the current tile, thus guaranteeing data integrity in subsequent steps such as trajectory point matching, elevation curve fitting, and cross-tile smoothing.

[0189] The map construction method of this disclosure determines the offset based on the number of binary bits of the target tile's identifier, and determines the map zoom level based on the difference between the number of binary bits and the offset. This allows the tile's hierarchical information to be directly parsed from its identifier itself, without relying on additional metadata. The target code is jointly determined using the tile identifier, map zoom level, and offset. By performing a bit-interleaving inverse operation on the target code, it is decoded into two-dimensional grid coordinates, achieving a deterministic mapping from the tile identifier to standardized spatial coordinates. Finally, the geographical extent of the target tile can be uniquely determined by combining the map zoom level and the two-dimensional grid coordinates. Therefore, the spatial location information of the target tile can be embedded within its encoding structure, making the geographical extent determination process completely self-contained, reversible, and computationally efficient. This avoids dependence on external indexes or configuration files, significantly improving the deployment flexibility and processing consistency of the slope map construction system in distributed or edge computing environments.

[0190] As one possible implementation method, Figure 7 This is a schematic flowchart illustrating another map construction method provided as an exemplary embodiment of the present disclosure. It should be noted that this map construction method can be executed alone, or it can be executed together with any embodiment of the present disclosure or any possible implementation thereof, or it can be executed together with any technical solution in the related art. The embodiments of the present disclosure do not impose any limitations on this.

[0191] like Figure 7 As shown, the map construction method may include the following steps S701 to S707: Step S701: Determine the target road network data located within the same geographical area from the map road network data, and construct a road network topology map based on the target road network data; wherein, the nodes in the road network topology map are used to indicate the connection points between connected roads, and the edges between nodes are used to indicate the topological relationships between connected roads.

[0192] The road network data can be road network data from a standard map (SD Map).

[0193] The target road network data includes attribute data for each road within the same geographical area. The attribute data includes, but is not limited to: road_id (unique road identifier), roadclass (road grade), direction (traffic direction, such as 1=one-way, 2=two-way, 3=reverse), f_road_id / t_road_id (connection relationship between preceding and succeeding roads), length (length in meters), max_speed (speed limit in km / h), and the geometric coordinate sequence of the road centerline.

[0194] The road network topology graph can also be called a node-link graph. In the road network topology graph, nodes indicate the connection points (i.e., road endpoints) between connected roads; edges between nodes indicate the topological relationships between connected roads. Optionally, the edges between nodes have corresponding direction codes, which indicate the direction of road traffic. The direction of traffic includes one-way, two-way, and reverse.

[0195] In this embodiment of the disclosure, target road network data located within the same geographical area can be determined from map road network data, and a road network topology map can be constructed based on the target road network data.

[0196] For example, define two endpoint labels for each road: (road_id, "from") and (road_id, "to"); use a union-find data structure to merge topologically connected endpoints: traverse the f_road_id (predecessor road ID) and t_road_id (successor road ID) of each road, and perform the following merge operations: merge the "from" endpoint of the current road with the "to" endpoint of the predecessor road: union(my_from, other_to); merge the "to" endpoint of the current road with the "from" endpoint of the successor road: union(my_to, other_from). Assign a unique node identifier node_id to each merged endpoint; where the coordinates of each node are the average of the coordinates of all merged endpoints; direction mapping: map the direction encoding of the SD Map (e.g., 1=one-way, 2=two-way, 3=reverse) to the encoding of the map matching library (e.g., 1=forward, 0=two-way, -1=reverse).

[0197] Step S702: Obtain multiple vehicle driving trajectories located within the same geographical area and containing elevation observation values, and sample key points for each vehicle driving trajectory to obtain each first trajectory point in each vehicle driving trajectory.

[0198] Among them, the length of the vehicle's driving trajectory is greater than a set length threshold, and multiple trajectory points in the vehicle's driving trajectory have corresponding confidence levels, where the confidence level can be the original confidence level or the normalized confidence level.

[0199] For example, the initial features of each trajectory point in the vehicle's driving trajectory are, for example, [longitude, latitude, elevation observation, timestamp, confidence level, distance along the road, road ID], where the distance along the road and the road ID can be initialized fixed values ​​(such as -1) and filled in by subsequent steps.

[0200] In this embodiment of the disclosure, multiple vehicle driving trajectories located within the same geographical area and containing elevation observation values ​​are obtained, and key point sampling is performed on each vehicle driving trajectory to obtain each first trajectory point in each vehicle driving trajectory.

[0201] For example, a 3D distance-based equal-interval downsampling algorithm can be used to sample key points for each vehicle trajectory, obtaining each first trajectory point in each vehicle trajectory. The sampling interval is marked as d. key Then, starting from the first trajectory point of each vehicle's driving trajectory, the 3D distance between subsequent trajectory points and the previously saved trajectory point can be checked sequentially. Only when the 3D distance is ≥ d... key When this happens, the trajectory point is retained. This allows for the removal of overly dense sampling points, reducing the amount of subsequent computation.

[0202] Step S703: Match each first trajectory point to the road network topology map to obtain the map matching result of each first trajectory point; wherein, the map matching result is used to indicate the road sign of the road matched by the first trajectory point, the geographical location of the first trajectory point after being projected onto the matched road, and the road distance of the first trajectory point.

[0203] The geographical location can be the projected coordinates (prj_lng, prj_lat) of the first trajectory point onto the road network topology map that matches the first trajectory point, including the projected longitude and projected latitude.

[0204] Among them, the distance along the road is the cumulative path distance of the first trajectory point along the projected road after the vehicle's driving trajectory is projected onto the road network topology map.

[0205] In this embodiment of the disclosure, a map matching algorithm can be used to match each first trajectory point to a road network topology map to obtain the map matching result of each first trajectory point. The map matching algorithm includes, but is not limited to, map matching algorithms based on Hidden Markov Models (HMMs).

[0206] Step S704: Generate multidimensional features for each first trajectory point based on the map matching results of each first trajectory point.

[0207] For example, the initial features of each first trajectory point can be updated based on the map matching results of each first trajectory point to obtain the multidimensional features of each first trajectory point.

[0208] Step S705: Based on the multidimensional features and local density of each first trajectory point, outlier removal and downsampling are performed on the first trajectory points to obtain the second trajectory points.

[0209] Local density is used to characterize the density of trajectory points surrounding the first trajectory point.

[0210] Step S706: Generate a smooth elevation curve for the same road based on the road distance and elevation observation values ​​of the second trajectory point on the same road within the same geographical area.

[0211] Step S707: Based on the smoothed elevation curves of each road within the same geographical area, obtain the slope map data for the same geographical area in the slope map.

[0212] It should be noted that the explanations of steps S705 to S707 can be found in the relevant descriptions in any embodiment of this disclosure, and will not be repeated here.

[0213] The map construction method of this disclosure extracts target road network data from map road network data and constructs a road network topology map. After sampling key points of vehicle driving trajectories containing elevation observation values, the data is matched to the road network topology map to obtain map matching results for each collected trajectory point. Based on this, structured multidimensional features are generated. This method can achieve accurate mapping of trajectory points from original GPS coordinates to road semantic space, effectively solving the problem of road classification ambiguity caused by positioning drift or signal noise of original trajectory points. It ensures that subsequent elevation modeling is strictly based on the actual driving road, avoids invalid observation interference across roads or non-roads, and significantly improves the road-level accuracy and semantic consistency of slope map construction.

[0214] In any embodiment of this disclosure, using vehicle travel trajectory as GPS trajectory and tiles in the slope map as examples, this disclosure provides a slope map construction scheme based on crowdsourced trajectories. The overall process is as follows: Figure 8 As shown, it mainly includes the following steps S1 to S9: Step S1: Tile Region Determination. Purpose: To decode NDS (NavData Standard, a standardized map data format widely used in the field of intelligent connected vehicles) tile identifiers into geographic extent polygons, and extend them outward to cover boundary road segments.

[0215] Detailed process: (1) Extract map zoom level: Calculate the number of binary bits nbits of the tile identifier (tile_id), determine the offset (offset = 32 when nbits > 32, otherwise offset = 16), and calculate the map zoom level: ; (2) Extract Morton code: Calculate the flag bit flag = 1 << (zoom + offset), and extract the Morton code: ; (3) Morton decoding: The one-dimensional Morton code is decoded into two-dimensional grid coordinates (xpart, ypart) through the inverse bit interleaving operation. The specific bit operation sequence is as follows: extract the even-numbered bits of the Morton code to obtain the x-component, and extract the even-numbered bits of the Morton code after shifting it right by 1 bit to obtain the y-component. The bit extraction operation is implemented through 5 masking and shifting operations.

[0216] (4) Signed coordinate recovery: Negative coordinates are handled using two's complement: if ,but ;like ,but .

[0217] (5) Geographic boundaries converted to latitude and longitude: Calculate geographic scale factors: ; Calculate the geographic boundary according to formula (8), then the polygon of the geographic boundary can be represented as [NW, NE, SE, SW, NW], where NW = (lon min lat max ) ← Top left, NE = (lon max lat max ) ← Top right, SE = (lon max lat min )← Bottom right, SW = (lon min lat min ) ← Bottom left, NW = (lon min lat max ) ← Return to the starting point, close the loop.

[0218] (6) Polygon expansion: Expand the polygon of the geographical boundary of the tile by 50 meters to ensure that the boundary road segment is completely covered. The specific steps are as follows: calculate the center point (center_lon, center_lat) of the polygon; convert the vertices of the polygon from the WGS84 coordinate system (world geodetic coordinate system) to the ENU (East-North-Sky) coordinate system with the center point as the origin; process the polygon in the ENU coordinate system to generate a new polygon that has been expanded by 50 meters; convert the vertices of the expanded polygon in the ENU coordinate system back to the WGS84 coordinate system to obtain the geographical range of the expanded tile polygon.

[0219] Step S2: Road Network Data Acquisition. Objective: To acquire standard map (SDMap) road network data within the polygonal geographic area of ​​the extended tiles.

[0220] Detailed process: (1) Query the current standard map version number sdmap_version.

[0221] (2) Use the polygon of the expanded tile as the query range to obtain the road network data within the polygon. Each road (link) contains the following attributes: road_id (unique road identifier), roadclass (road level), direction (traffic direction: 1=one-way, 2=two-way, 3=reverse), f_road_id / t_road_id (previous / successor road connection relationship), length (length, meters), max_speed (speed limit, km / h), and the geometric coordinate sequence of the road centerline.

[0222] (3) If the road network data is empty (no road data), output an empty result and terminate the processing.

[0223] Step S3: Track Data Acquisition and Preprocessing. Objective: To acquire crowdsourced GPS tracks and perform quality filtering and formatting.

[0224] Detailed process: (1) Batch trajectory retrieval: Starting from the current time, trace back up to a maximum of T. max (Values ​​are, for example, 100 days). GPS tracks within the polygonal geographic area of ​​the outer tile are retrieved in batches of 7 days each. Filtering criteria: Only GPS tracks containing elevation data (needZ=True) and with a positioning status of 2 (low precision) or 3 (high precision) are retrieved. Early termination condition: Retrieving GPS tracks stops when the accumulated number of GPS tracks exceeds 10 times the number of road links in the road network. The format of each track point in the original GPS track is [longitude, latitude, timestamp, unknown field, original confidence value, elevation observation value].

[0225] (2) 3D cumulative distance calculation and length filtering: For each GPS track, with the first track point as the origin of the ENU coordinate system, calculate and sum the 3D Euclidean distances between all adjacent track points to obtain the total length of the GPS track: ; Where i is the total number of trajectory points in the GPS trajectory, (Δe i ,Δn i ,Δh i () represents the difference between the eastward, northward, and celestial components of the i-th segment in the ENU coordinate system. The i-th segment refers to the line segment between the i-th trajectory point and the (i+1)-th trajectory point. In this disclosure, segments with a length less than L can be discarded. min A short GPS track (e.g., 100m).

[0226] (3) Keyframe Sampling: Equal-interval downsampling based on 3D distance. Starting from the first trajectory point, the 3D ENU distance between subsequent trajectory points and the previously retained trajectory point is checked sequentially, and only when the distance is greater than or equal to d is the 3D ENU distance checked. key (For example, a value of 5m) retains the trajectory point. This step removes overly dense sampling points, reducing the amount of subsequent calculations.

[0227] (4) Confidence mapping: Mapping the original confidence scores to confidence scores. raw Mapping to normalized intervals: ; (5) Sorting and numbering: Sort in descending order of trajectory length (longer trajectory first), and assign consecutive integer IDs (0, 1, 2, ...).

[0228] (6) Format conversion: The format of each trajectory point in each GPS trajectory is converted into the following 7-dimensional vector [longitude, latitude, elevation observation value, timestamp, confidence level, distance along the road, road ID], where the distance along the road and road ID are initialized to -1 (to be filled in step S4 / S5).

[0229] Step S4: Road Network Matching. Purpose: To match (adapt) the trajectory points in the GPS track to the corresponding road segments on the road network.

[0230] Detailed process: (1) Road network topology construction: Extract the node-link graph from the SD Map obtained in step S2. Define two endpoint labels for each road: (road_id, "from") and (road_id, "to"). Merge topologically connected endpoints using a union-find data structure: traverse the f_road_id (predecessor road ID) and t_road_id (successor road ID) of each road and perform the following merge operation: The "from" endpoint of the current road is merged with the "to" endpoint of the preceding road: union(my_from, other_to); The "to" endpoint of the current road is merged with the "from" endpoint of the subsequent road: union(my_to, other_from); Assign a unique node_id to each merged endpoint. The coordinates of each node are the average of the coordinates of all merged endpoints. Direction mapping: Map the direction encoding of the SD Map (e.g., 1=one-way, 2=two-way, 3=reverse) to the encoding of the map matching library (e.g., 1=forward, 0=two-way, -1=reverse).

[0231] (2) Hidden Markov Model (HMM) map matching: An HMM-based map matching algorithm is used. The main parameters are as follows: GPS buffer radius = 30 meters (search for candidate road segments within a 30m range of each track point in the GPS track); top number of candidate road segments k = 30 (each trajectory point has a maximum of 30 candidate road segments); Enable heading angle correction (use_heading_inf=True). When the distance between the front and rear track points is ≤5m, heading angle correction can be ignored. Enable hierarchical subnet matching (use_sub_net=True, is_hierarchical=True); Multi-core parallel execution; Output: Map matching results for each trajectory point, including the matched road ID (link_id), projected coordinates (prj_lng, prj_lat), and distance along the road (route_dis).

[0232] Step S5: Track Projection Update. Purpose: To write the map matching results back into the GPS track and establish the association between each elevation observation and the specific road location.

[0233] Detailed process: For each GPS track point, the map matching result (only the main match sub_seq == 0) is used. The latitude and longitude of the track point are updated to the matching projected coordinates (prj_lng, prj_lat), and the road distance (route_dis) and road ID (link_id) are written to the corresponding positions in the track coordinate array. After this step, the multidimensional features of each track point become: [projected longitude, projected latitude, elevation observation value, timestamp, confidence score, road distance, road ID].

[0234] Step S6: Calculate the slope of each road segment.

[0235] Objective: To independently perform elevation fitting and slope calculation for each road segment. The detailed process for slope calculation segment by segment is as follows: Figure 9 As shown, it mainly includes the following sub-steps: Step S6.1, Road geometry resampling: Transform the road coordinates from the WGS84 coordinate system to the ENU coordinate system (with the tile center as the reference point). Calculate the cumulative distance sequence along the road D = [0, cumsum(||P_{i+1} - P_i||)], where cumsum refers to the position of the i-th trajectory point. Generate the target distance array d. target = [0, Δs, 2Δs, ..., D max ], where Δs is the slope sampling interval, D max This refers to the road length of the current road in the SD Map. Resampled ENU coordinates are obtained by interpolating at the target distance using a linear interpolation function.

[0236] Step S6.2, Tile Member Filtering: For each resampled point, determine whether it falls within the polygonal geographic area of ​​the tile. If so, mark it as a valid sampled point (valid_mask); skip roads without valid sampled points.

[0237] Step S6.3, Coverage Assessment: Collect elevation observations for all trajectory points matched to the current road. Calculate the number of intervals covered by elevation observations, using the intervals between valid sampling points as the unit. The coverage formula is, for example: coverage = |Number of intervals covered by elevation observations| / |Total number of intervals|; If coverage < 0.5, skip the road and do not output slope data (insufficient observation data coverage).

[0238] Step S6.4, IQR outlier removal and K-Nearest Neighbors (KNN) density downsampling: The input consists of the distance along the path from the trajectory point x_train and the elevation observation value y_train.

[0239] (a) Sort by x and divide the road into several segments every segment_length.

[0240] (b) Perform IQR outlier removal for each road segment (requires a number of trajectory points >= 4): Calculate the first quartile Q1 = percentile(y, 25) and the third quartile Q3 = percentile(y, 75), calculate the interquartile range IQR = Q3 - Q1, and remove outlier trajectory points whose elevation observation y meets the following conditions: ; (c) For each road segment, if the number of segments still exceeds N after cleaning max For each trajectory point, perform KNN downsampling: construct a KNN model using the elevation observation value y as a feature (e.g., k = min(5, number of points)) and calculate the local density of each trajectory point.

[0241] Sort by local density in descending order, retaining the top N. max The trajectory points with the highest local density are rearranged in x-order to maintain the spatial structure.

[0242] In summary, IQR outlier removal combined with KNN density downsampling can solve the problems of abnormal elevation observations contaminating GPS tracks and uneven data density. Segmented IQR filtering specifically removes abnormal elevation observations; KNN density downsampling adaptively retains the most representative observation points in road segments with excessively dense data, preventing overly dense data from dominating the fitting process.

[0243] Step S6.5, Gaussian process elevation fitting: The input is the cleaned observation data (x_train, y_train).

[0244] (a) Perform data normalization according to formula (5).

[0245] (b) Constructing a Gaussian process model: Likelihood function: Gaussian Likelihood; Mean function: Constant Mean; Covariance kernel function: ScaleKernel(MaternKernel(ν=2.5, ard_num_dims=1)); The reason for choosing the Matern 2.5 kernel as the radial basis function kernel is that it generates second-order differentiable functions, which is suitable for modeling smooth but non-infinitely differentiable natural terrain such as road elevation profiles. The formula for the Matern 2.5 kernel function is shown in formula (4).

[0246] The ScaleKernel wrapper layer adds global output variance scaling.

[0247] (c) Model training: Optimizer: Adam, learning rate lr = 0.1; Number of iterations: 300; Loss function: Negative exact marginal log-likelihood, i.e.: ; Wherein, θ includes the hyperparameters of the kernel function (length scale l, output scale σ²) and the likelihood noise variance; X is the mileage coordinate data input during the training phase, used to indicate the spatial location information of each sampling point on the road; y is the measured elevation observation data corresponding one-to-one with the mileage coordinate data X, used to indicate the elevation measurement value of the road surface at each sampling point on the road.

[0248] (d) Prediction: in [x min , x max 3000 prediction points are uniformly generated within the range. An accelerated method is used to speed up the calculation of prediction variance, and the mean of the elevation prediction is obtained. pred And a 95% confidence interval. Then, through inverse normalization, the elevation prediction value is obtained: elevation = mean. pred σ y + μ y .

[0249] In summary, Gaussian process elevation fitting can solve the problems of extracting smooth elevation curves from noisy observations and providing uncertainty quantification. Using a Gaussian process regression model with a Matern 2.5 kernel, elevation noise in GPS tracks is modeled as the noise variance parameter of Gaussian likelihood. The optimal smoothness (length scale) and noise level are automatically learned by maximizing the marginal likelihood.

[0250] Step S6.6, Finite Difference Slope Calculation: Interpolate the smoothed elevation curve predicted by the Gaussian process at the target distance location, and then calculate the slope: Internal points (central difference), such as the slope of the i-th point: ; Where i = 1, 2, ..., N-2, elev i+1 This is the predicted elevation value for the (i+1)th point. It represents the distance along the road from the (i+1)th point.

[0251] The slope of the first endpoint (forward difference) is: ; The slope of the endpoint (backward difference) is: .

[0252] The slope unit is m / m (dimensionless ratio), which can be multiplied by 100 to convert to percentage slope.

[0253] Step S7: Cross-tile fusion, mainly includes the following sub-steps: Objective: To eliminate slope discontinuities at the boundaries of adjacent tiles.

[0254] Step S7.1, Neighbor Tile Discovery: Decode the tile identifier of the current tile into Morton code, and then decode it into grid coordinates (x, y) (using the same decoding method as in step S1). Traverse the 8-neighbor directions (dx, dy) ∈ {-1, 0, 1}²\ {(0,0)}: Calculate the coordinates (x+dx, y+dy) of the neighboring tiles, perform boundary checks, convert the signed coordinates back to unsigned code, perform Morton encoding (the reverse process of decoding), and reconstruct the tile identifier.

[0255] Step S7.2, Cross-tile road matching: Pull slope map data for the current tile and all neighboring tiles from the data service. Construct a dictionary with road_id as the key, collecting all elevation profiles of the same road across different tiles.

[0256] Step S7.3, Elevation Mean Fusion: For roads appearing in multiple tiles (boundary roads), average all valid elevation values ​​(or slope values) at each distance sampling point: ; Step S7.4, Linear Interpolation Filling: Calculate the merged progress = number of valid elevation points / total number of sampling points. If progress ≥ 0.9, then perform linear interpolation filling on the remaining empty None / NaN values.

[0257] Step S7.5, Slope recalculation: Based on the fused elevation data, the slope is recalculated using the same finite difference formula as in step S6.6.

[0258] In summary, cross-tile fusion can solve the boundary discontinuity problem caused by parallel tile processing. By combining 8-neighbor Morton coding, neighbor tile discovery, cross-tile elevation mean fusion, and linear interpolation filling, the spatial continuity of the slope map at tile boundaries is ensured.

[0259] Step S8: Batch processing. Purpose: To perform batch slope calculations on multiple tiles.

[0260] Detailed process: For a predefined list of tile IDs, the complete pipeline of steps S1-S7 is executed one by one through subprocesses, with each tile being processed independently.

[0261] Step S9: Route tile discovery.

[0262] Objective: To automatically determine the set of tiles for generating a slope map based on the user-input origin and destination addresses.

[0263] Detailed process: (1) Input the starting point and ending point addresses, and call the route planning service to obtain the route coordinate sequence.

[0264] (2) Convert each route coordinate point to an NDS tile ID based on the lon and lat indicated by the geographic boundaries of the tile and the map zoom level.

[0265] (3) Remove duplicates from the set of tile IDs to obtain the minimum set of tiles that covers the entire route.

[0266] (4) Perform steps S1-S7 sequentially or in batches for each tile.

[0267] In summary, the solution provided in this disclosure has at least the following advantages: 1. Low cost and high precision: Utilizing a Gaussian process regression model, high-precision slope information is extracted from low-precision crowdsourced GPS trajectory elevation data, requiring no specialized measurement equipment, and the marginal cost of data acquisition is almost zero. 2. Strong robustness: The dual mechanism of segmented IQR outlier filtering + KNN density downsampling effectively resists abnormal elevation observations caused by GPS signal multipath, urban canyons, sensor failures, etc., without the need for manual data cleaning. 3. Quantifiable uncertainty: The Gaussian process regression model naturally outputs a 95% confidence interval for prediction, providing a basis for data quality assessment for downstream intelligent driving systems and improving system safety. 4. Spatial continuity: Cross-tile elevation mean fusion + linear interpolation filling mechanism eliminates slope jumps at tile boundaries, ensuring a continuous experience for route-level applications. 5. High scalability: A multi-level parallel architecture (multi-core map matching + multi-threaded road segment processing + multi-tile batch processing) enables efficient computation of large-scale road networks. 6. Adaptive Quality Control: The coverage assessment mechanism ensures that slope data is output only when data is abundant; the time window expansion strategy automatically expands the data collection range when data is sparse. 7. Continuous Update Capability: Based on continuously collected crowdsourced GPS tracks, the slope map can be updated periodically to reflect road changes (new construction, reconstruction, repaving, etc.) without the need to redeploy a professional surveying fleet.

[0268] The above are various embodiments of the map construction method. This disclosure also provides a map application method, namely a vehicle control method. Figure 10 A schematic flowchart of a vehicle control method provided for an exemplary embodiment of this disclosure.

[0269] like Figure 10 As shown, the vehicle control method may include the following steps S1001 to S1002: Step S1001: Obtain slope map data of the tiles covered by at least one planned route of the vehicle from the slope map.

[0270] The slope map was constructed using any of the aforementioned map construction methods. The planned route was obtained by trajectory planning based on the vehicle's destination.

[0271] In this embodiment of the disclosure, the slope map data of the tiles covered by at least one planned route of the vehicle can be queried from the slope map.

[0272] Step S1002: Perform driving control on the vehicle based on the slope map data of the blocks covered by at least one planned route.

[0273] In this embodiment of the disclosure, vehicle driving control can be performed based on the slope map data of the blocks covered by the planned route.

[0274] In any embodiment of this disclosure, the vehicle control method is, for example, as follows: determining the energy consumption information required for the vehicle to travel along the planned route based on the slope map data of the blocks covered by each planned route; determining the navigation route with the lowest energy consumption from each planned route based on the energy consumption information of each planned route; determining the slope data of at least one segment in the navigation route based on the slope map data of the blocks covered by the navigation route; and controlling the vehicle's driving based on the slope data of at least one segment.

[0275] Among them, energy consumption is positively correlated with slope, that is, the steeper the slope, the higher the energy consumption.

[0276] Understandably, by using slope map data based on the tiles covered by each planned route, the energy consumption information of each planned route is quantitatively calculated, and the navigation route with the lowest energy consumption is selected accordingly, thus realizing intelligent path decision-making aimed at energy conservation. Furthermore, for the selected navigation route, detailed slope data of at least one segment is extracted, and the vehicle's driving control parameters (such as motor torque, braking pressure, and cruising speed) are dynamically adjusted accordingly, so that the vehicle can execute the optimal control strategy under different working conditions such as uphill, downhill, or flat roads. This not only effectively extends the vehicle's driving range but also improves driving comfort and active safety.

[0277] As an application scenario, taking the vehicle's driving trajectory as a GPS trajectory and the tiles in the slope map as tiles, the map construction method and vehicle control method provided in this disclosure can be applied to multi-device interaction scenarios. For example, they mainly involve the following three types of device interaction and collaboration: Data collection terminal: Mass-produced vehicles (fleet) equipped with GPS receivers continuously collect GPS tracks containing elevation data during daily driving and upload them to a track server in the cloud; In the cloud: retrieve GPS tracks and standard maps (SD Map) from the track server, execute the slope map generation pipeline (map matching → Gaussian process fitting → cross-tile fusion), and store the results in the map database service; On the consumer side: Intelligent driving vehicles or navigation clients query the slope map data of the tiles covered by the required route through relevant interfaces for longitudinal control planning, energy consumption prediction, etc.

[0278] For example, the user (or intelligent driving planning module) inputs the starting point and the end point (or destination) address → the route planning service returns the route coordinates → the route coordinates are converted into a set of tile IDs → a slope calculation pipeline is performed on each tile → cross-tile fusion eliminates boundary discontinuities → the slope data of each road segment is provided to the map consumer through relevant interfaces for driving control.

[0279] The vehicle control method of this disclosure obtains the slope data of the blocks covered by the planned route of the vehicle from the slope map and directly uses the slope data for vehicle driving control. This achieves a deep integration of refined road longitudinal slope prior information with vehicle dynamic control strategy, providing reliable input for energy management, power distribution and braking strategy, thereby significantly improving the predictability, energy efficiency and safety of vehicle control.

[0280] To implement the above embodiments, this disclosure also proposes a map building apparatus. Figure 11 This is a schematic diagram of the structure of a map building apparatus provided for an exemplary embodiment of the present disclosure.

[0281] like Figure 11 As shown, the map building device 1100 may include: a first acquisition module 1110, a processing module 1120, a first generation module 1130, and a second generation module 1140.

[0282] The first acquisition module 1110 is used to acquire multidimensional features of each first trajectory point in multiple vehicle driving trajectories located within the same geographical area; wherein the multidimensional features include at least the elevation observation value and the road distance of the first trajectory point; the road distance is used to indicate the cumulative path distance from the starting point of the vehicle driving trajectory to the first trajectory point; the processing module 1120 is used to perform outlier removal and downsampling processing on the first trajectory points according to the multidimensional features and local density of each first trajectory point to obtain second trajectory points; wherein the local density is used to characterize the density of the surrounding trajectory points of the first trajectory point; the first generation module 1130 is used to generate a smooth elevation curve of the same road according to the road distance and elevation observation value of the second trajectory points on the same road within the same geographical area; the second generation module 1140 is used to obtain slope map data of the same geographical area in the slope map according to the smooth elevation curves of each road within the same geographical area.

[0283] In one implementation of this disclosure, the multidimensional feature further includes a geographic location. A processing module 1120 is configured to: determine a first trajectory point located on any road within the same geographic area; wherein any road is divided into multiple road segments; perform outlier removal processing on the first trajectory point on the same road segment based on the road distance of the first trajectory point on the same road segment to obtain candidate trajectory points; in response to the existence of a target road segment among the multiple road segments, perform downsampling processing on the candidate trajectory points based on the local density of the candidate trajectory points on the target road segment to obtain a second trajectory point; wherein the number of candidate trajectory points on the target road segment is greater than a first predetermined number.

[0284] In one implementation of this disclosure, the processing module 1120 is configured to: sort the first trajectory points on the same road segment in ascending order according to their road distance to obtain a first sorting sequence; determine an upper elevation limit and a lower elevation limit based on the elevation observation values ​​of each first trajectory point in the first sorting sequence; and filter the first trajectory points on the same road segment according to the upper elevation limit and the lower elevation limit to obtain candidate trajectory points; wherein the elevation observation values ​​of the candidate trajectory points are located between the upper elevation limit and the lower elevation limit.

[0285] In one implementation of this disclosure, the processing module 1120 is configured to: determine the k nearest neighbors of each candidate trajectory point based on the elevation observation values ​​of candidate trajectory points on the target road segment; wherein k is a positive integer greater than 1; determine the local density of each candidate trajectory point based on the average distance from each candidate trajectory point to the k nearest neighbors; wherein the local density is negatively correlated with the average; sort the candidate trajectory points on the target road segment in descending order based on the local density to obtain a second sorting sequence; and take the second set number of candidate trajectory points at the top of the second sorting sequence as the second trajectory points.

[0286] In one implementation of this disclosure, the map building device 1100 may further include: a first determining module, configured to resample the road geometry of any road to obtain multiple first sampling points; wherein the first sampling points are located within the same geographical area; divide any road based on the multiple first sampling points to obtain multiple intervals; determine a target interval covering the first trajectory points from the multiple intervals according to the multidimensional features of the first trajectory points; determine the coverage rate of any road to elevation observations according to the proportion of the target intervals; and determine whether to generate and output slope data corresponding to any road according to the relationship between the coverage rate and a set probability.

[0287] In one implementation of this disclosure, the first generation module 1130 is configured to: determine the hyperparameters in the Gaussian process regression model based on the road distance of the second trajectory point on the same road as the input variable and the elevation observation value of the second trajectory point on the same road as the output variable; determine the average elevation prediction value of multiple prediction points within the road distance range based on the hyperparameters, the road distance of the second trajectory point on the same road, and the elevation observation value; wherein the road distance range is the minimum to the maximum value among the road distances of the second trajectory points on the same road; and generate a smooth elevation curve of the same road based on the average elevation prediction value of the multiple prediction points.

[0288] In one implementation of this disclosure, the Gaussian process regression model also outputs the prediction variance of multiple prediction points. The map building device 1100 may further include: a second determining module, used to determine the confidence level of the smooth elevation curve of the same road based on the prediction variance of multiple prediction points; and to determine the reliability of the slope data of the same road in the slope map data based on the confidence level of the smooth elevation curve of the same road.

[0289] In one implementation of this disclosure, the input variable is the normalized distance along the road, and the output variable is the normalized elevation observation value. The first generation module 1130 is used to: perform inverse normalization processing on the average elevation prediction value of multiple prediction points to obtain the elevation prediction value of multiple prediction points; and generate a smooth elevation curve of the same road based on the elevation prediction value of multiple prediction points.

[0290] In one implementation of this disclosure, the slope map includes multiple tiles, each tile indicating a corresponding geographical area; the map building device 1100 may further include: a smoothing module, used to determine neighboring tiles of the target tile belonging to the same geographical area from the multiple tiles based on the tile identifiers of the multiple tiles in the slope map; determine the common boundary road between the target tile and the neighboring tiles; and smooth the slope data corresponding to the boundary road in the slope map data of the target tile and the slope map data of the neighboring tiles.

[0291] In one implementation of this disclosure, a smoothing module is configured to: obtain a first slope value of at least one second sampling point on a boundary road from the slope map data of the target tile; obtain a second slope value of at least one third sampling point on a boundary road from the slope map data of neighboring tiles; wherein the distance between the second sampling point and the third sampling point is less than a set distance; fuse the first slope value of the second sampling point and the second slope value of the third sampling point to obtain a fused slope value; and update the first slope value in the slope map data of the target tile and the second slope value in the slope map data of neighboring tiles based on the fused slope value.

[0292] In one implementation of this embodiment, the map building device 1100 may further include: an interpolation module, configured to, in response to the slope map data indicating the existence of a fourth sampling point without a slope value on the boundary road, and the proportion of the fourth sampling point being greater than a set proportion, interpolate and fill the slope value of the fourth sampling point to obtain the interpolated slope value of the fourth sampling point; and an update module, configured to update the slope data of the boundary road in the slope map data according to the interpolated slope value of the fourth sampling point.

[0293] In one implementation of this disclosure, the same geographical area is determined using the following modules: a third determining module, configured to determine the offset based on the number of binary bits of the tile identifier of the target tile belonging to the same geographical area, and determine the map zoom level based on the difference between the number of binary bits and the offset; determine the target code for indicating the geographical location of the target tile based on the tile identifier, the map zoom level, and the offset; perform a bit interleaving inverse operation on the target code to decode the target code into two-dimensional grid coordinates; and determine the same geographical area based on the map zoom level and the two-dimensional grid coordinates.

[0294] In one implementation of this disclosure, the first acquisition module 1110 is configured to: determine target road network data located within the same geographical area from map road network data, and construct a road network topology map based on the target road network data; wherein, nodes in the road network topology map are used to indicate connection points between connected roads, and edges between nodes are used to indicate topological relationships between connected roads; acquire multiple vehicle driving trajectories located within the same geographical area and containing elevation observation values, and perform key point sampling on each vehicle driving trajectory to obtain each first trajectory point in each vehicle driving trajectory; match each first trajectory point to the road network topology map to obtain map matching results for each first trajectory point; wherein, the map matching results are used to indicate the road signage of the road matched by the first trajectory point, the geographical location of the first trajectory point after projection onto the matched road, and the road distance of the first trajectory point; and generate multidimensional features for each first trajectory point based on the map matching results for each first trajectory point.

[0295] It should be noted that the explanations and descriptions of any of the foregoing map building method embodiments also apply to the map building apparatus of this embodiment, and will not be repeated here.

[0296] In the map building apparatus of this disclosure, multidimensional features (including elevation observations and road distances) of each first trajectory point in multiple vehicle driving trajectories within the same geographical area are acquired. Then, outlier removal and downsampling processing are performed on each first trajectory point using local density to obtain second trajectory points. This effectively eliminates elevation anomalies caused by GPS signal interference, signal obstruction, or urban canyon effects, while avoiding the impact of excessive redundancy or uneven distribution of trajectory points on the accuracy of elevation curve generation. Based on this, a smooth elevation curve is generated using the road distance and elevation observations of the second trajectory points, ensuring the continuity and stability of road elevation representation. Finally, slope map data is constructed based on the smooth elevation curves of each road. This not only significantly reduces the reliance on professional surveying equipment and achieves low-cost, wide-coverage, and high-frequency updateable slope map production, but also provides higher road surface elevation representation accuracy in complex road sections such as bridges, tunnels, and elevated roads compared to slope calculation methods based on coarse-resolution DEMs or simple elevation averages in related technologies. This provides a more reliable longitudinal environmental perception input for intelligent driving systems.

[0297] To implement the above embodiments, this disclosure also proposes a vehicle control device. Figure 12 This is a schematic diagram of the structure of a vehicle control device provided for an exemplary embodiment of the present disclosure.

[0298] like Figure 12 As shown, the vehicle control device 1200 may include a second acquisition module 1210 and a control module 1220.

[0299] The second acquisition module 1210 is used to acquire slope map data of the tiles covered by at least one planned route of the vehicle from the slope map; wherein the slope map is constructed using the map building device 1100 in the aforementioned embodiment; the control module 1220 is used to control the vehicle's driving based on the slope map data of the tiles covered by at least one planned route.

[0300] In one implementation of this disclosure, the control module 1220 is configured to: determine the energy consumption information required for the vehicle to travel along the planned route based on the slope map data of the blocks covered by each planned route; determine the navigation route from each planned route based on the energy consumption information of each planned route; determine the slope data of at least one segment in the navigation route based on the slope map data of the blocks covered by the navigation route; and perform driving control on the vehicle based on the slope data of at least one segment.

[0301] It should be noted that the foregoing explanation of the vehicle control method embodiment also applies to the vehicle control device of this embodiment, and will not be repeated here.

[0302] In the vehicle control device of this disclosure, the slope data of the blocks covered by the planned route of the vehicle is obtained from the slope map, and the slope data is directly used for vehicle driving control. This realizes the deep integration of refined road longitudinal slope prior information with vehicle dynamic control strategy, providing reliable input for energy management, power distribution and braking strategy, thereby significantly improving the predictability, energy efficiency and safety of vehicle control.

[0303] To implement the above embodiments, this disclosure also proposes an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the map building method or vehicle control method as described in any of the foregoing embodiments.

[0304] Figure 13 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure. (Refer to...) Figure 13The electronic device 1300 may include one or more of the following components: a processing component 1302, a memory 1304, a power component 1306, a multimedia component 1308, an audio component 1310, an input / output (I / O) interface 1312, a sensor component 1314, and a communication component 1316.

[0305] Processing component 1302 typically controls the overall operation of electronic device 1300, such as operations associated with display, telephone calls, data communication, camera operation, and recording operations. Processing component 1302 may include one or more processors 1320 to execute instructions to perform all or part of the steps of the methods described above. Furthermore, processing component 1302 may include one or more modules to facilitate interaction between processing component 1302 and other components. For example, processing component 1302 may include a multimedia module to facilitate interaction between multimedia component 1308 and processing component 1302.

[0306] The memory 1304 is configured to store various types of data to support the operation of the electronic device 1300. The memory 1304 can be implemented by any type of volatile or non-volatile storage device or a combination thereof.

[0307] Power component 1306 provides power to various components of electronic device 1300. Power component 1306 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to electronic device 1300.

[0308] The multimedia component 1308 includes a screen that provides an output interface between the electronic device 1300 and the user. In some embodiments, the multimedia component 1308 includes a front-facing camera and / or a rear-facing camera. Each front-facing and rear-facing camera may be a fixed optical lens system or have focal length and optical zoom capabilities.

[0309] Audio component 1310 is configured to output and / or input audio signals. For example, audio component 1310 includes a microphone (MIC) configured to receive external audio signals when electronic device 1300 is in an operating mode, such as call mode, recording mode, and voice recognition mode. The received audio signals may be further stored in memory 1304 or transmitted via communication component 1316. In some embodiments, audio component 1310 also includes a speaker for outputting audio signals.

[0310] I / O interface 1312 provides an interface between processing component 1302 and peripheral interface modules, such as keyboards, click wheels, buttons, etc.

[0311] Sensor assembly 1314 includes one or more sensors for providing state assessments of various aspects of electronic device 1300. For example, sensor assembly 1314 can detect the on / off state of electronic device 1300, the relative positioning of components, changes in the position of electronic device 1300 or a component of electronic device 1300, the presence or absence of user contact with electronic device 1300, the orientation or acceleration / deceleration of electronic device 1300, temperature changes of electronic device 1300, the presence of nearby objects, etc.

[0312] Communication component 1316 is configured to facilitate wired or wireless communication between electronic device 1300 and other devices. Electronic device 1300 can access wireless networks based on communication standards, such as WiFi, 4G, or 5G, or combinations thereof. In one exemplary embodiment, communication component 1316 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, communication component 1316 further includes a Near Field Communication (NFC) module to facilitate short-range communication.

[0313] In an exemplary embodiment, the electronic device 1300 may be implemented by one or more chips for performing the methods described above.

[0314] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 1304 including instructions, which can be executed by a processor 1320 of an electronic device 1300 to perform the above-described method. For example, the non-transitory computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CD-ROM), magnetic tape, floppy disk, and optical data storage device, etc.

[0315] To implement the above embodiments, this disclosure also proposes a vehicle, including: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to implement the vehicle control method as described in the foregoing embodiments.

[0316] Figure 14 This is a block diagram illustrating a vehicle 1400 according to an exemplary embodiment. For example, vehicle 1400 may be a hybrid vehicle, a non-hybrid vehicle, an electric vehicle, a fuel cell vehicle, or other types of vehicle. Vehicle 1400 may be an intelligent driving vehicle, a semi-intelligent driving vehicle, or a non-intelligent driving vehicle.

[0317] Reference Figure 14 The vehicle 1400 may include various subsystems, such as an infotainment system 1410, a perception system 1420, a decision control system 1430, a drive system 1440, and a computing platform 1450. The vehicle 1400 may also include more or fewer subsystems, and each subsystem may include multiple components. Furthermore, each subsystem and each component of the vehicle 1400 can be interconnected via wired or wireless means.

[0318] In some embodiments, the infotainment system 1410 may include a communication system, an entertainment system, and a navigation system, etc.

[0319] The perception system 1420 may include several sensors for sensing information about the environment surrounding the vehicle 1400. For example, the perception system 1420 may include a global positioning system (which may be a GPS system, a BeiDou system, or another positioning system), an inertial measurement unit (IMU), a lidar, a millimeter-wave radar, an ultrasonic radar, and a camera device.

[0320] The decision control system 1430 may include a computing system, a vehicle controller, a steering system, a throttle, and a braking system.

[0321] The drive system 1440 may include components that provide powered motion to the vehicle 1400. In one embodiment, the drive system 1440 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.

[0322] Some or all of the functions of the vehicle 1400 are controlled by a computing platform 1450. The computing platform 1450 may include at least one processor 1451 and a memory 1452, the processor 1451 being able to execute instructions 1453 stored in the memory 1452.

[0323] Processor 1451 can be any conventional processor, such as a commercially available CPU. Processors may also include graphics processing units (GPUs), field-programmable gate arrays (FPGAs), systems-on-chips (SoCs), application-specific integrated circuits (ASICs), or combinations thereof.

[0324] The memory 1452 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.

[0325] In addition to instruction 1453, memory 1452 can also store data, such as road maps, route information, vehicle position, direction, speed, and other data. The data stored in memory 1452 can be used by computing platform 1450.

[0326] In this embodiment of the disclosure, processor 1451 may execute instructions 1453 to complete all or part of the steps of the above-described vehicle control method embodiments.

[0327] To implement the above embodiments, this disclosure also proposes a chip, wherein the chip includes an interface circuit and a processing circuit coupled to each other. The interface circuit is used to input or output signals, and the processing circuit is configured to execute the map building method or vehicle control method provided in any of the foregoing embodiments.

[0328] Figure 15 This is a schematic diagram of the structure of a chip proposed as an exemplary embodiment of this disclosure. See also... Figure 15 The diagram shown is a schematic representation of the structure of chip 1500, but it is not limited to this.

[0329] Chip 1500 includes processing circuit 1501, which is configured to execute any of the above map building methods or vehicle control methods.

[0330] In some embodiments, chip 1500 further includes one or more interface circuits 1502. Optionally, interface circuit 1502 is connected to memory 1503, and interface circuit 1502 can be used to receive signals from memory 1503 or other devices, and interface circuit 1502 can be used to send signals to memory 1503 or other devices. For example, interface circuit 1502 can read instructions stored in memory 1503 and send the instructions to processing circuit 1501.

[0331] In some embodiments, the interface circuit 1502 performs at least one of the communication steps such as sending and / or receiving in the above method, while the processing circuit 1501 performs other steps.

[0332] In some embodiments, the terms interface circuit, interface, transceiver pin, transceiver, etc., can be used interchangeably.

[0333] In some embodiments, chip 1500 further includes one or more memories 1503 for storing instructions. Optionally, all or part of the memories 1503 may be located outside of chip 1500.

[0334] To implement the above embodiments, this disclosure also proposes a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the map construction method or vehicle control method as described in any of the foregoing method embodiments.

[0335] To implement the above embodiments, this disclosure also proposes a computer program product having a computer program stored thereon, which, when executed by a processor, implements the map construction method or vehicle control method as described in any of the foregoing method embodiments.

[0336] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this disclosure. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0337] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this disclosure, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0338] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing custom logic functions or processes, and the scope of preferred embodiments of this disclosure includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as will be understood by those skilled in the art to which embodiments of this disclosure pertain.

[0339] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and compact disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0340] It should be understood that various parts of this disclosure can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0341] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0342] Furthermore, the functional units in the various embodiments of this disclosure can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0343] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of the present disclosure have been shown and described above, it is to be understood that the above embodiments are exemplary and should not be construed as limiting the present disclosure. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present disclosure.

Claims

1. A map construction method, characterized in that, include: Obtain multidimensional features of each first trajectory point in multiple vehicle driving trajectories located within the same geographical area; wherein, the multidimensional features include at least the elevation observation value and the road distance of the first trajectory point; the road distance is used to indicate the cumulative path distance from the starting point of the vehicle driving trajectory to the first trajectory point; Based on the multidimensional features and local density of each first trajectory point, outlier removal and downsampling are performed on the first trajectory points to obtain second trajectory points; wherein, the local density is used to characterize the density of the surrounding trajectory points of the first trajectory point; Based on the road distance and elevation observation values ​​of the second trajectory points on the same road within the same geographical area, a smooth elevation curve of the same road is generated; Based on the smoothed elevation curves of each road within the same geographical area, slope map data for the same geographical area in the slope map is obtained.

2. The method according to claim 1, characterized in that, The multidimensional features also include geographical location. The step of performing outlier removal and downsampling on the first trajectory points based on the multidimensional features and local density of each first trajectory point to obtain the second trajectory points includes: Determine the first trajectory point on any road within the same geographical area where the geographical location is located; wherein, any road is divided into multiple road segments; Based on the road distance of the first trajectory point on the same road segment, outlier removal processing is performed on the first trajectory point on the same road segment to obtain candidate trajectory points; In response to the existence of a target road segment among the plurality of road segments, the candidate trajectory points are downsampled based on the local density of the candidate trajectory points on the target road segment to obtain the second trajectory point; wherein the number of candidate trajectory points on the target road segment is greater than a first set number.

3. The method according to claim 2, characterized in that, The step of removing outliers from the first trajectory points on the same road segment based on their road-side distances to obtain candidate trajectory points includes: Based on the distance along the road from the first trajectory points on the same road segment, the first trajectory points on the same road segment are sorted in ascending order to obtain a first sorting sequence; Based on the elevation observation values ​​of each first trajectory point in the first sorting sequence, determine the upper and lower elevation limits; Based on the upper elevation limit and the lower elevation limit, the first trajectory points on the same road segment are filtered to obtain the candidate trajectory points; wherein the elevation observation value of the candidate trajectory points is located between the upper elevation limit and the lower elevation limit.

4. The method according to claim 2, characterized in that, The step of downsampling the candidate trajectory points based on the local density of the candidate trajectory points on the target road segment to obtain the second trajectory point includes: Based on the elevation observation values ​​of candidate trajectory points on the target road segment, determine the k nearest neighbors of each candidate trajectory point; where k is a positive integer greater than 1. The local density of each candidate trajectory point is determined based on the average distance from each candidate trajectory point to its k nearest neighbors; wherein the local density is negatively correlated with the average distance. Based on the local density, the candidate trajectory points on the target road segment are sorted in descending order to obtain a second sorting sequence; The second set number of candidate trajectory points that are at the beginning of the second sorting sequence are taken as the second trajectory points.

5. The method according to any one of claims 2-4, characterized in that, Before performing outlier removal and downsampling on the first trajectory points based on their multidimensional features and local density, the method further includes: Resampling of the road geometry for any one of the roads yields multiple first sampling points; wherein the first sampling points are located within the same geographical area; Based on the multiple first sampling points, any one road is divided into multiple intervals; Based on the multidimensional features of the first trajectory point, a target interval covering the first trajectory point is determined from the plurality of intervals; Based on the proportion of the target interval, determine the coverage rate of any road to the elevation observation value; Based on the relationship between the coverage rate and the set probability, determine whether to generate and output the slope data corresponding to any of the roads.

6. The method according to claim 1, characterized in that, The step of generating a smooth elevation curve for the same road based on the road distance and elevation observation values ​​of the second trajectory points on the same road within the same geographical area includes: Based on the Gaussian process regression model, the hyperparameters in the Gaussian process regression model are determined by taking the distance along the road of the second trajectory point on the same road as the input variable and the elevation observation value of the second trajectory point on the same road as the output variable. Based on the hyperparameters, the distance along the road of the second trajectory point on the same road, and the observed elevation values, the average elevation prediction of multiple prediction points within the distance range is determined; wherein, the distance range is the minimum to maximum distance among the distances along the road of the second trajectory point on the same road. A smooth elevation curve for the same road is generated based on the average elevation prediction of the multiple prediction points.

7. The method according to claim 6, characterized in that, The Gaussian process regression model also outputs the prediction variance of the multiple prediction points, and the method further includes: Based on the prediction variance of the multiple prediction points, determine the confidence level of the smooth elevation curve of the same road; The reliability of the slope data of the same road in the slope map data is determined based on the confidence level of the smooth elevation curve of the same road.

8. The method according to claim 6, characterized in that, The input variable is the normalized distance along the road, and the output variable is the normalized elevation observation value. The step of generating a smooth elevation curve for the same road based on the average elevation prediction of the multiple prediction points includes: The average elevation prediction of the multiple prediction points is inversely normalized to obtain the elevation prediction values ​​of the multiple prediction points. Based on the predicted elevation values ​​of the multiple prediction points, a smooth elevation curve for the same road is generated.

9. The method according to claim 1, characterized in that, The slope map includes multiple tiles, each tile indicating a corresponding geographical area; the method further includes: Based on the tile identifiers of multiple tiles in the slope map, neighboring tiles of the target tile belonging to the same geographical area are determined from the multiple tiles; Determine the common boundary roads between the target tile and its neighboring tiles; The slope data corresponding to the boundary roads in the slope map data of the target tile and the slope map data of the neighboring tiles are smoothed.

10. The method according to claim 9, characterized in that, The smoothing process for the slope data corresponding to the boundary roads in the slope map data of the target tile and the slope map data of the neighboring tiles includes: From the slope map data of the target patch, obtain the first slope value of at least one second sampling point on the boundary road; From the slope map data of the neighboring tiles, obtain the second slope value of at least one third sampling point on the boundary road; wherein the distance between the second sampling point and the third sampling point is less than a set distance; The first slope value of the second sampling point and the second slope value of the third sampling point are fused to obtain a fused slope value; Based on the fused slope value, the first slope value in the slope map data of the target tile and the second slope value in the slope map data of the neighboring tiles are updated.

11. The method according to claim 10, characterized in that, The method further includes: In response to the map slope data indicating the existence of a fourth sampling point without a slope value on the boundary road, and the proportion of the fourth sampling point being greater than a set proportion, the slope value of the fourth sampling point is interpolated and filled to obtain the interpolated slope value of the fourth sampling point. The slope data of the boundary road in the map slope data is updated based on the interpolated slope value of the fourth sampling point.

12. The method according to claim 1, characterized in that, The methods for determining the same geographical area include: The offset is determined based on the number of binary bits of the tile identifier of the target tile belonging to the same geographical area, and the map zoom level is determined based on the difference between the number of binary bits and the offset. Based on the tile identifier, the map zoom level, and the offset, a target code for indicating the geographic location of the target tile is determined; Perform a bit-interleaving inverse operation on the target encoding to decode the target encoding into two-dimensional grid coordinates; The same geographical area is determined based on the map zoom level and the two-dimensional grid coordinates.

13. The method according to claim 1, characterized in that, The acquisition of multidimensional features of each first trajectory point in multiple vehicle driving trajectories located within the same geographical area includes: Target road network data located within the same geographical area are determined from the map road network data, and a road network topology map is constructed based on the target road network data; wherein, the nodes in the road network topology map are used to indicate the connection points between connected roads, and the edges between the nodes are used to indicate the topological relationships between the connected roads; Multiple vehicle driving trajectories located within the same geographical area and containing elevation observation values ​​are obtained, and key point sampling is performed on each vehicle driving trajectory to obtain each first trajectory point in each vehicle driving trajectory; Each of the first trajectory points is matched onto the road network topology map to obtain a map matching result for each of the first trajectory points; wherein, the map matching result is used to indicate the road sign of the road matched by the first trajectory point, the geographical location of the first trajectory point after being projected onto the matched road, and the road distance of the first trajectory point; Based on the map matching results of each of the first trajectory points, multidimensional features of each of the first trajectory points are generated.

14. A vehicle control method, characterized in that, include: From the slope map, obtain slope map data of the tiles covered by at least one planned route of the vehicle; wherein the slope map is constructed using the method described in any one of claims 1-13; The vehicle is controlled to drive based on the slope map data of the blocks covered by the at least one planned route.

15. The method according to claim 14, characterized in that, The step of controlling the vehicle's movement based on the slope map data of the tiles covered by the at least one planned route includes: Based on the slope map data of the blocks covered by each of the planned routes, determine the energy consumption information required for the vehicle to travel along the planned routes; Based on the energy consumption information of each of the planned routes, a navigation route is determined from each of the planned routes; Based on the slope map data of the tiles covered by the navigation route, determine the slope data of at least one segment of the navigation route; The vehicle is controlled to drive based on the slope data of at least one road segment.

16. A map building apparatus, characterized in that, include: The first acquisition module is used to acquire multidimensional features of each first trajectory point in multiple vehicle driving trajectories located within the same geographical area; wherein, the multidimensional features include at least the elevation observation value and the road distance of the first trajectory point; the road distance is used to indicate the cumulative path distance from the starting point of the vehicle driving trajectory to the first trajectory point; The processing module is used to perform outlier removal and downsampling processing on the first trajectory points based on the multidimensional features and local density of each first trajectory point to obtain a second trajectory point; wherein, the local density is used to characterize the density of the surrounding trajectory points of the first trajectory point. The first generation module is used to generate a smooth elevation curve for the same road based on the road distance and elevation observation values ​​of the second trajectory points on the same road within the same geographical area. The second generation module is used to obtain slope map data for the same geographical area in the slope map based on the smooth elevation curves of each road within the same geographical area.

17. The apparatus according to claim 16, characterized in that, The multidimensional features also include geographic location, and the processing module is used for: Determine the first trajectory point on any road within the same geographical area where the geographical location is located; wherein, any road is divided into multiple road segments; Based on the road distance of the first trajectory point on the same road segment, outlier removal processing is performed on the first trajectory point on the same road segment to obtain candidate trajectory points; In response to the existence of a target road segment among the plurality of road segments, the candidate trajectory points are downsampled based on the local density of the candidate trajectory points on the target road segment to obtain the second trajectory point; wherein the number of candidate trajectory points on the target road segment is greater than a first set number.

18. The apparatus according to claim 17, characterized in that, The processing module is used for: Based on the distance along the road from the first trajectory points on the same road segment, the first trajectory points on the same road segment are sorted in ascending order to obtain a first sorting sequence; Based on the elevation observation values ​​of each first trajectory point in the first sorting sequence, determine the upper and lower elevation limits; Based on the upper elevation limit and the lower elevation limit, the first trajectory points on the same road segment are filtered to obtain the candidate trajectory points; wherein the elevation observation value of the candidate trajectory points is located between the upper elevation limit and the lower elevation limit.

19. A vehicle control device, characterized in that, include: The second acquisition module is used to acquire slope map data of the tiles covered by at least one planned route of the vehicle from the slope map; wherein the slope map is constructed using the map construction apparatus according to any one of claims 16-18; The control module is used to control the vehicle's movement based on the slope map data of the blocks covered by the at least one planned route.

20. The vehicle control device according to claim 19, characterized in that, The control module is used for: Based on the slope map data of the blocks covered by each of the planned routes, determine the energy consumption information required for the vehicle to travel along the planned routes; Based on the energy consumption information of each of the planned routes, a navigation route is determined from each of the planned routes; Based on the slope map data of the tiles covered by the navigation route, determine the slope data of at least one segment of the navigation route; The vehicle is controlled to drive based on the slope data of at least one road segment.

21. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, it implements the steps of the method as claimed in any one of claims 1 to 13, and / or implements the steps of the method as claimed in claim 14 or 15.

22. A vehicle, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured as follows: Implement the steps of the method as described in claim 14 or 15.

23. A non-transitory computer-readable storage medium having computer program instructions stored thereon, characterized in that, When executed by a processor, the program instructions implement the steps of the method according to any one of claims 1 to 13, and / or implement the steps of the method according to claim 14 or 15.

24. A chip, characterized in that, The chip includes an interface circuit and a processing circuit that are coupled to each other. The interface circuit is used to input or output signals, and the processing circuit is used to implement the method of any one of claims 1 to 13, and / or to implement the method of claim 14 or 15.

25. A computer program product, characterized in that, Includes a computer program, which, when executed by a processor, implements the steps of the method according to any one of claims 1 to 13, and / or implements the steps of the method according to claim 14 or 15.

Images