Multi-station cooperation point cloud level fusion perception method, device and system, electronic equipment and storage medium

Abstract

The invention provides a multi-station cooperation point cloud level fusion sensing method, device and system, electronic equipment and a storage medium, and relates to the technical field of wireless. According to the method, a sensing overlapping coverage area is determined, so that all multi-station cloud data does not need to be adopted during target matching, only first multi-station cloud data in the sensing overlapping coverage area needs to be utilized, and the target matching efficiency is improved. The method can reduce the amount of point cloud data needing target matching, and greatly reduces the complexity and mismatching probability of a point cloud level fusion algorithm. Moreover, the reliability evaluation values of the data points are introduced in the method, and the reliability of each data point is applied to weighted clustering, so that the fusion diversity gain of the sensing overlapping coverage areas of the plurality of base stations can be improved, the point cloud level fusion precision is further improved, and the accuracy of the target position obtained after point cloud fusion is improved.

Description

Technical Field

[0001] This application relates to the field of wireless technology, and in particular to a multi-station collaborative point cloud-level fusion sensing method, device, system, electronic device, and storage medium. Background Technology

[0002] Communication-sensing fusion (or simply sensing) is one of the key technological directions in the evolution of 5G-A. Based on shared hardware and software resources, it utilizes a portion of time and frequency resources to transmit sensing signals and, through processing the received point cloud data, senses the motion state of targets, including key information such as distance, angle, position, speed, and direction. Unlike radar, sensing fusion technology achieves seamless coverage by networking base stations with both communication and sensing capabilities (i.e., integrated sensing base stations). In overlapping coverage areas with multiple sensing stations, diversity gain can be obtained through multi-station collaborative fusion and deduplication, improving sensing accuracy and other performance aspects.

[0003] However, since the reliability of point cloud data from different base stations varies, and the reliability of different point cloud data from the same base station also varies, base stations in abnormal states such as interference may directly report erroneous point cloud data. Therefore, if point cloud data from different base stations are directly fused, the accuracy of the target location obtained after point cloud fusion will be low, thereby reducing the accuracy of the target location. Summary of the Invention

[0004] This application provides a multi-station collaborative point cloud-level fusion sensing method, device, system, electronic device, and storage medium to address the deficiencies in related technologies.

[0005] This application provides a multi-station collaborative point cloud-level fusion sensing method, applied to a first type of baseband processing unit, including: Determine the overlapping coverage area of ​​multiple base stations working together; Based on the first multi-site cloud data within the overlapping coverage area of ​​the perception, target matching is performed to determine the second multi-site cloud data belonging to the same target. Determine the reliability assessment value of each data point in the second multi-site cloud data, and based on the reliability assessment value of each data point in the second multi-site cloud data, perform weighted clustering on the second multi-site cloud data to determine the current trajectory point of the target corresponding to the second multi-site cloud data.

[0006] According to the multi-site collaborative cloud-level fusion sensing method provided in this application, the step of determining the current trajectory point of the target corresponding to the second multi-site cloud data by performing weighted clustering on the second multi-site cloud data based on the reliability assessment value of each data point in the second multi-site cloud data includes: Based on the reliability assessment values ​​of each data point in the second multi-site cloud data, the initial cluster center of the second multi-site cloud data is determined; Using the initial cluster center as the current cluster center, and based on the distance between each data point in the second multi-site cloud data and the current cluster center, the data points are clustered to obtain clusters; Based on the reliability assessment value of each data point in the cluster, the current cluster center is iteratively updated until the position change of the new cluster center in multiple consecutive iterations is less than a preset threshold. Then, the new cluster center is taken as the current trajectory point.

[0007] According to the multi-station collaborative point cloud-level fusion sensing method provided in this application, the step of iteratively updating the current cluster center based on the reliability assessment value of each data point in the cluster includes: In each iteration, the reliability assessment value of each data point in the cluster is used as a weight to perform a weighted average of the location information of each data point in the cluster, thereby obtaining the location information of the new cluster center.

[0008] According to the multi-site collaborative point cloud-level fusion sensing method provided in this application, the step of performing target matching based on the first multi-site cloud data within the overlapping sensing coverage area to determine the second multi-site cloud data belonging to the same target includes: Based on the centroid distance and distance threshold of point cloud clusters of different single-site cloud data in the first multi-site cloud data, preliminary target matching is performed to obtain a combination of point cloud clusters that are suspected to belong to the same target. Based on the historical trajectory information and current velocity of the corresponding target in the point cloud cluster combination, the trajectory points of the corresponding target are predicted to obtain the predicted trajectory points. Based on at least one of the distance between the predicted trajectory point and the current trajectory point, the radar cross section of the corresponding target, and the area difference of the point cloud cluster combination, the point cloud cluster combination is filtered to obtain the filtering result. Based on the filtering results, the second multi-site cloud data is determined.

[0009] According to the multi-site collaborative point cloud-level fusion sensing method provided in this application, the step of determining the second multi-site point cloud data based on the filtering result includes: If the filtering result includes only one point cloud cluster combination, then the point cloud cluster combination is used as the second multi-site cloud data; If the filtering result includes multiple point cloud cluster combinations, then the point cloud cluster combination with the smallest centroid distance among the multiple point cloud cluster combinations is taken as the second multi-point point cloud data.

[0010] According to the multi-station collaborative point cloud-level fusion sensing method provided in this application, the distance threshold is determined based on the environmental complexity within the sensing overlap coverage area and the number of targets per unit area.

[0011] According to the multi-station cooperative point cloud-level fusion sensing method provided in this application, determining the overlapping coverage area of ​​multiple base stations includes: Receive the initial point cloud data collected when each base station is turned on. The initial point cloud data corresponding to each base station includes the point cloud data of the target device with the positioning module. Based on the initial point cloud data corresponding to each base station and the positioning information of the positioning module, the sensing overlap coverage area is determined.

[0012] According to the multi-site collaborative cloud-level fusion sensing method provided in this application, the step of determining the current trajectory point of the target corresponding to the second multi-site cloud data by performing weighted clustering on the second multi-site cloud data based on the reliability assessment value of each data point in the second multi-site cloud data includes: Based on the distance between each data point in the second multi-site cloud data and the centroid of the second multi-site cloud data, abnormal data points in the second multi-site cloud data are removed.

[0013] This application also provides a multi-station collaborative point cloud-level fusion sensing method, applied to a second type of baseband processing unit, including: Obtain cloud data from a single site; Calculate the overall reliability influencing factors of the single-site cloud data and the single-point reliability influencing factors of each data point in the single-site cloud data; Based on the overall reliability influencing factors and the single-point reliability influencing factors, the reliability assessment value of each data point in the single-site cloud data is determined, and the single-site cloud data and the reliability assessment value of each data point in the single-site cloud data are reported to the first type of baseband processing unit.

[0014] According to the multi-site collaborative point cloud-level fusion sensing method provided in this application, the overall reliability influencing factors include the mean square error of the single-site cloud data; The factors affecting the reliability of a single point include the signal-to-noise ratio of the data point, the deviation from the trajectory prediction point, and the density of surrounding points. The process of determining the reliability assessment value of each data point in the single-site cloud data based on the overall reliability influencing factors and the single-point reliability influencing factors includes: The mean square error of the single-site cloud data, the signal-to-noise ratio of each data point in the single-site cloud data, the deviation from the trajectory prediction point, and the density of surrounding points are normalized respectively to obtain the normalization results. The results of each normalization process are weighted and summed to obtain the reliability assessment value of each data point in the single-site cloud data.

[0015] According to the multi-site collaborative cloud-level fusion sensing method provided in this application, the step of reporting the single-site cloud data and the reliability assessment value of each data point in the single-site cloud data to the first type of baseband processing unit includes: Based on the reliability assessment value of each data point in the single-site cloud data, non-abnormal data points within each data point of the single-site cloud data are determined; Accordingly, the step of reporting the single-site cloud data and the reliability assessment values ​​of each data point in the single-site cloud data to the first type of baseband processing unit includes: The data of the non-abnormal points in the single-site cloud data and the reliability assessment value are reported to the first type of baseband processing unit.

[0016] This application also provides a multi-station collaborative point cloud-level fusion sensing device, applied to a first type of baseband processing unit, comprising: The area determination module is used to determine the sensing overlap coverage area of ​​multiple base stations working together; The target matching module is used to perform target matching based on the first multi-site cloud data within the perceived overlapping coverage area, and to determine the second multi-site cloud data belonging to the same target; The weighted clustering module is used to determine the reliability assessment value of each data point in the second multi-site cloud data, and based on the reliability assessment value of each data point in the second multi-site cloud data, to perform weighted clustering on the second multi-site cloud data to determine the current trajectory point of the target corresponding to the second multi-site cloud data.

[0017] This application also provides a multi-station collaborative point cloud-level fusion sensing device, applied to a second type of baseband processing unit, comprising: The data acquisition module is used to acquire cloud data from a single site. The influencing factor calculation module is used to calculate the overall reliability influencing factors of the single-site cloud data and the single-point reliability influencing factors of each data point in the single-site cloud data. The reliability assessment module is used to determine the reliability assessment value of each data point in the single-site cloud data based on the overall reliability influencing factors and the single-point reliability influencing factors, and to report the single-site cloud data and the reliability assessment value of each data point in the single-site cloud data to the first type of baseband processing unit.

[0018] This application also provides a multi-station collaborative point cloud-level fusion sensing system, including: a first type of baseband processing unit and a second type of baseband processing unit, wherein the first type of baseband processing unit and the second type of baseband processing unit are communicatively connected; The first type of baseband processing unit is used to execute the above-described multi-station collaborative point cloud-level fusion sensing method; The second type of baseband processing unit is used to execute the above-described multi-station collaborative point cloud-level fusion sensing method.

[0019] This application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the multi-station collaborative point cloud-level fusion sensing method as described above.

[0020] This application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the multi-station collaborative point cloud-level fusion sensing method as described above.

[0021] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the multi-station collaborative point cloud-level fusion sensing method as described above.

[0022] The multi-station collaborative point cloud-level fusion sensing method, apparatus, system, electronic device, and storage medium provided in this application first determine the overlapping coverage area of ​​multiple base stations' collaborative sensing. Then, based on the first multi-station cloud data within the overlapping coverage area, target matching is performed to determine the second multi-station cloud data belonging to the same target. Finally, the reliability assessment value of each data point in the second multi-station cloud data is determined, and based on the reliability assessment value of each data point in the second multi-station cloud data, weighted clustering is performed on the second multi-station cloud data to determine the current trajectory point of the target corresponding to the second multi-station cloud data. This method, by determining the overlapping coverage area, eliminates the need to use all multi-station cloud data for target matching, requiring only the first multi-station cloud data within the overlapping coverage area. This reduces the amount of point cloud data that needs to be matched, significantly lowering the complexity of the point cloud-level fusion algorithm and the probability of mismatch. Moreover, the introduction of the reliability assessment value of data points and the application of the reliability of each data point to weighted clustering can improve the fusion diversity gain of the overlapping coverage area of ​​multiple base stations, further improving the point cloud-level fusion accuracy and the accuracy of the target location obtained after point cloud fusion. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in this application or related technologies, the drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is one of the flowcharts of the multi-station collaborative point cloud-level fusion sensing method provided in this application.

[0025] Figure 2 This is the second flowchart of the multi-station collaborative point cloud-level fusion sensing method provided in this application.

[0026] Figure 3 This is one of the structural schematic diagrams of the multi-station collaborative point cloud-level fusion sensing device provided in this application.

[0027] Figure 4 This is the second structural schematic diagram of the multi-station collaborative cloud-level fusion sensing device provided in this application.

[0028] Figure 5 This is a schematic diagram of the structure of the multi-station collaborative cloud-level fusion sensing system provided in this application.

[0029] Figure 6 This is a schematic diagram of the structure of the electronic device provided in this application. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0031] Existing sensor fusion technologies mainly include trajectory-level fusion, point cloud-level fusion, and signal-level fusion. However, trajectory-level fusion schemes do not fully utilize the underlying multi-site cloud data, resulting in limited multi-site collaboration gains. Therefore, further research is needed on point cloud-level fusion schemes to further improve sensing accuracy by fully utilizing point cloud data from different base stations at different angles.

[0032] However, existing point cloud fusion solutions do not consider the reliability of point cloud data, but directly fuse point cloud data from different base stations. This results in low accuracy of the target location obtained after point cloud fusion, thereby reducing the accuracy of the target location.

[0033] Based on this, this application provides a multi-station collaborative point cloud-level fusion perception method.

[0034] Figure 1 This is a flowchart illustrating a multi-station collaborative point cloud-level fusion sensing method provided in an embodiment of this application, such as... Figure 1 As shown, this method is applied to a first type of baseband processing unit, including: S11, determine the overlapping coverage area of ​​multiple base stations working together; S12, based on the first multi-site cloud data within the perceived overlapping coverage area, perform target matching to determine the second multi-site cloud data belonging to the same target; S13, determine the reliability assessment value of each data point in the second multi-site cloud data, and based on the reliability assessment value of each data point in the second multi-site cloud data, perform weighted clustering on the second multi-site cloud data to determine the current trajectory point of the target corresponding to the second multi-site cloud data.

[0035] Specifically, the multi-station collaborative point cloud-level fusion sensing method provided in this application embodiment is executed by a first type of baseband unit (BBU). The 5G-A distributed base station architecture may include a first type of baseband unit and a second type of baseband unit, which are deployed in their respective base stations and work together.

[0036] The first type of baseband processing unit can be the master baseband processing unit (MasterBBU) in the 5G-A distributed base station architecture, and the second type of baseband processing unit can be the slave baseband processing unit (SlaveBBU) in the 5G-A distributed base station architecture.

[0037] The first type of baseband processing unit may include one unit, while the second type of baseband processing unit may include multiple units. Each second type of baseband processing unit can collect single-site cloud data from its respective base station and determine the reliability estimate of each data point in the single-site cloud data. It then reports the single-site cloud data and the reliability estimates of each data point to the first type of baseband processing unit, which receives the single-site cloud data to form multi-site cloud data. Simultaneously, the first type of baseband processing unit also receives the reliability estimates of each data point within each single-site cloud data set.

[0038] First, step S11 is executed to determine the overlapping sensing coverage area of ​​multiple base stations cooperating. Since each base station is equipped with a second type of baseband processing unit and has a corresponding sensing coverage area, the overlapping sensing coverage area refers to the intersection of the sensing coverage areas of at least two base stations. Here, the overlapping sensing coverage area may include one or more, and each overlapping sensing coverage area corresponds to the second type of baseband processing unit deployed by at least two base stations.

[0039] The sensing coverage area can be predetermined and stored by the UAV, and the identification information (ID) of each second-type baseband processing unit corresponding to the sensing overlapping coverage area can also be recorded. For example, the UAV can measure the sensing reference signal strength or channel state information from different base stations and report the measurement report to the first-type baseband processing unit. The first-type baseband processing unit can calculate the sensing overlapping coverage area of ​​multiple base stations based on the sensing reference signal strength or channel state information.

[0040] Then, step S12 is executed. Since the point cloud angle, velocity, and distance characteristics of different base stations are different, it is necessary to determine whether the cloud data of multiple sites belong to the same target through target matching.

[0041] During the target matching process, the first multi-site cloud data within the overlapping coverage area can be determined first. This first multi-site cloud data includes the single-site cloud data of each base station corresponding to the overlapping coverage area within the overlapping coverage area.

[0042] Subsequently, using the first multi-site cloud data within the overlapping sensing coverage area, target matching is performed to determine the second multi-site cloud data belonging to the same target. One or more targets may exist within the overlapping sensing coverage area, and each target corresponds to a set of second multi-site cloud data. This second multi-site cloud data includes combinations of point cloud clusters belonging to the same target within each individual site cloud data segment of the first multi-site cloud data.

[0043] Here, the point cloud clusters with the smallest centroid distance among the individual site cloud data within the first multi-site cloud data can be selected as the second multi-site cloud data belonging to the same target. The centroid distance can be calculated using Euclidean distance.

[0044] Finally, step S13 is executed to determine the reliability assessment value of each data point in the second multi-site cloud data. This reliability assessment value is a numerical value obtained by evaluating the reliability of the data points, which can be evaluated and reported by each type of baseband processing unit. The reliability of a data point refers to its accuracy and reasonableness. The higher the reliability estimate of a data point, the higher its reliability. The reliability estimate of a data point can be jointly determined by the signal-to-clutter ratio (SCR) of the data point, its deviation from the trajectory prediction point, the density of surrounding points, and the mean square error of the single-site cloud data where the data point is located in the second multi-site cloud data. Here, the trajectory prediction point can be predicted using the Kalman filter algorithm. The density of surrounding points refers to the density of data points within a circle with a radius of meters around the data point.

[0045] For example, the mean-square error (MSE), signal-to-noise ratio (SNR) of each data point in a single-site cloud data point, deviation from the trajectory prediction point, and density of surrounding points can be normalized to obtain the normalization results. Here, the normalization methods for different influencing factors may differ depending on their impact on the reliability of the data points. For instance, for the MSE and deviation from the trajectory prediction point in single-site cloud data, smaller values ​​correspond to higher reliability; therefore, they can be normalized by adding one and taking the reciprocal. For the SNR and density of surrounding points, larger values ​​correspond to higher reliability; therefore, they can be normalized by using the ratio to the maximum value.

[0046] Subsequently, by weighted summing of the normalization results, the reliability assessment value of each data point in the single-site cloud data can be obtained.

[0047] In addition, the first type of baseband processing unit can also integrate a reliability assessment algorithm, which is calculated based on the single-site cloud data reported by each second type of baseband processing unit after the second type of baseband processing unit reports the single-site cloud data. No specific limitation is made here.

[0048] Determining the trajectory points of a target can be understood as locating the target. The reliability assessment values ​​of each data point in the second multi-site cloud data can be used as weights. A weighted clustering algorithm is then used to perform weighted clustering on each data point in the second multi-site cloud data, so that the cluster centers are biased towards data points with higher reliability, thereby determining the current trajectory points of the target corresponding to the second multi-site cloud data.

[0049] Here, weighted clustering algorithms can include weighted K-means and weighted density-based clustering (Density-Based Spatial Clustering of Applications with Noise, DBSCAN). If the weighted K-means algorithm is used, since the second set of multi-site cloud data belongs to the same target, K is set to 1, meaning there is only one cluster center. Its implementation principle is that the weight of each data point affects the calculation of the cluster center. When updating the cluster center, the location information of the new cluster center can be a weighted average of the location information of each data point.

[0050] If weighted DBSCAN is used, its implementation principle is that the weight of each data point can adjust the density contribution of the data point. For example, data points with higher weights are considered to have a lower threshold for being considered "core points".

[0051] The multi-station collaborative point cloud-level fusion sensing method provided in this application embodiment first determines the overlapping coverage area of ​​multiple base stations' collaborative sensing. Then, based on the first multi-station cloud data within the overlapping coverage area, target matching is performed to determine the second multi-station cloud data belonging to the same target. Finally, the reliability assessment value of each data point in the second multi-station cloud data is determined, and based on the reliability assessment value of each data point in the second multi-station cloud data, weighted clustering is performed on the second multi-station cloud data to determine the current trajectory point of the target corresponding to the second multi-station cloud data. This method, by determining the overlapping coverage area, eliminates the need to use all multi-station cloud data for target matching, requiring only the first multi-station cloud data within the overlapping coverage area. This reduces the amount of point cloud data that needs to be matched, significantly lowering the complexity of the point cloud-level fusion algorithm and the probability of mismatch. Moreover, the introduction of the reliability assessment value of data points and the application of the reliability of each data point to weighted clustering can improve the fusion diversity gain of the overlapping coverage area of ​​multiple base stations, further improving the point cloud-level fusion accuracy and the accuracy of the target location obtained after point cloud fusion.

[0052] Based on the above embodiments, the step of determining the current trajectory point of the target corresponding to the second multi-site cloud data by performing weighted clustering on the second multi-site cloud data based on the reliability assessment values ​​of each data point in the second multi-site cloud data includes: Based on the reliability assessment values ​​of each data point in the second multi-site cloud data, the initial cluster center of the second multi-site cloud data is determined; Using the initial cluster center as the current cluster center, and based on the distance between each data point in the second multi-site cloud data and the current cluster center, the data points are clustered to obtain clusters; Based on the reliability assessment value of each data point in the cluster, the current cluster center is iteratively updated until the position change of the new cluster center in multiple consecutive iterations is less than a preset threshold. Then, the new cluster center is taken as the current trajectory point.

[0053] Specifically, when performing weighted clustering on the second multi-site cloud data, a weighted clustering algorithm can be introduced. First, the reliability assessment value of each data point in the second multi-site cloud data is used to determine the initial cluster centers. For example, the reliability assessment value of each data point in the second multi-site cloud data can be used as the weight of each data point, and the data point with the largest weight in the second multi-site cloud data can be selected as the initial cluster center.

[0054] Following this, the iterative optimization process for cluster centers begins. The initial cluster centers are used as the current cluster centers, and the distances between each data point in the second multi-site cloud data and the current cluster centers are calculated. This distance can be Euclidean distance, and the calculation formula is as follows: ; in, Let be the Euclidean distance between data point i and the current cluster center j, and k be the dimension index. Since each data point is a three-dimensional data point, the maximum value of k is 3. and These are the coordinates of data point i and the current cluster center j in the kth dimension, respectively.

[0055] The distance corresponding to each data point is compared with the clustering threshold, and data points whose cluster size is less than the clustering threshold are assigned to the cluster with the current cluster center. The clustering threshold can be set as needed and is not specifically limited here.

[0056] Subsequently, the reliability assessment value of each data point in the cluster is used as a weight to iteratively update the current cluster center, i.e., the cluster center is continuously recalculated to obtain new cluster centers. When the position change of the new cluster center in multiple consecutive iterations is less than a preset threshold, the weighted clustering algorithm is considered to have converged, and the new cluster center is taken as the current trajectory point of the target. The specific number of iterations can be set as needed, such as 5 times, 10 times, etc., and the preset threshold can also be set as needed; no specific limitation is made here.

[0057] The positional change of the new cluster center in each of the consecutive iterations is less than the preset threshold, which can be understood as the maximum positional change of the new cluster center in each of the consecutive iterations being less than the preset threshold.

[0058] In this embodiment, the cluster center is determined by the reliability assessment value of each data point, which can ensure that the cluster center is closer to the data points with high reliability and improve the accuracy of point cloud fusion.

[0059] Based on the above embodiments, the iterative update of the current cluster center based on the reliability assessment value of each data point in the cluster includes: In each iteration, the reliability assessment value of each data point in the cluster is used as a weight to perform a weighted average of the location information of each data point in the cluster, thereby obtaining the location information of the new cluster center.

[0060] Specifically, during the iterative update of the current cluster centers, in each iteration, the reliability assessment value of each data point in the cluster can be used as a weight to perform a weighted average of the location information of each data point in the cluster, thus obtaining the location information of the new cluster centers, i.e.: ; Where m is the number of data points in the cluster. For the location information of the new cluster centers, Let i be the reliability assessment value for the i-th data point. This represents the location information of the i-th data point.

[0061] In this embodiment, the reliability assessment value of each data point in the cluster is used as a weight. By performing a weighted average of each data point, the location information of the new cluster center is determined. This ensures that the cluster center is closer to the data points with high reliability, thereby improving the accuracy of point cloud fusion.

[0062] In existing point cloud fusion schemes, target matching selects the point cloud cluster with the smallest distance between centroids as the point cloud data belonging to the same target, which can easily lead to mismatches and affect the accuracy of point cloud fusion.

[0063] Based on this, and building upon the above embodiments, the step of performing target matching based on the first multi-site cloud data within the perceived overlapping coverage area to determine the second multi-site cloud data belonging to the same target includes: Based on the centroid distance and distance threshold of point cloud clusters of different single-site cloud data in the first multi-site cloud data, preliminary target matching is performed to obtain a combination of point cloud clusters that are suspected to belong to the same target. Based on the historical trajectory information and current velocity of the corresponding target in the point cloud cluster combination, the trajectory points of the corresponding target are predicted to obtain the predicted trajectory points. Based on at least one of the distance between the predicted trajectory point and the current trajectory point, the radar cross section of the corresponding target, and the area difference of the point cloud cluster combination, the point cloud cluster combination is filtered to obtain the filtering result. Based on the filtering results, the second multi-site cloud data is determined.

[0064] Specifically, during target matching, the first multi-site cloud data can be spatiotemporally aligned. At this point, inter-site clock synchronization can be ensured through air interface calibration, and the timestamps of the first multi-site cloud data can be pre-aligned through interpolation to ensure temporal consistency. Spatial consistency of the first multi-site cloud data is ensured through calibration of engineering parameters such as latitude and longitude.

[0065] Subsequently, based on spatiotemporal alignment, the centroid distance between point cloud clusters of different single-site cloud data in the first multi-site cloud data can be calculated, that is, the centroid distance between each point cloud cluster in one single-site cloud data and each point cloud cluster in another single-site cloud data can be calculated. This centroid distance can be Euclidean distance.

[0066] Subsequently, the centroid distances of point cloud clusters from different single-site cloud datasets within the first multi-site cloud dataset are compared with distance thresholds. Point cloud clusters with centroid distances less than the distance thresholds are considered to belong to the same target. Here, a point cloud cluster combination can include one or more, and each point cloud cluster combination corresponds to a target. The targets corresponding to different point cloud cluster combinations may be the same target or different targets, requiring further screening.

[0067] Subsequently, the historical trajectory information and current velocity of the corresponding target for each point cloud cluster can be used to predict the trajectory points of the corresponding target, thus obtaining the predicted trajectory points. The historical trajectory information of the corresponding target is the true trajectory information, and filtering algorithms such as Kalman filtering and particle filtering can be used to predict the trajectory points of the corresponding target, thereby obtaining the predicted trajectory points.

[0068] Next, the distance between the predicted trajectory point and the current trajectory point is calculated; this distance can be Euclidean distance. The current trajectory point is also a true trajectory point. By comparing the distance between the predicted and current trajectory points with a set threshold, point cloud cluster combinations can be filtered. For example, for a given point cloud cluster combination, if the distance between the predicted trajectory point and the current trajectory point of the corresponding target in that cluster is greater than the set threshold, it indicates that the point cloud cluster combination may be a mismatch caused by measurement errors, clutter interference, or other reasons. Filtering it out can improve the accuracy of target matching results.

[0069] In low-altitude scenarios, where the target is low-altitude, such as a drone, its radar cross-section (RCS), while following a certain fluctuation pattern, generally has an area threshold range, for example, it can be 0.01m. 2 ~0.1m 2 If the value exceeds this range, it is considered a mismatch caused by clutter, interference, or other factors. Based on this, it can also be determined whether the radar cross-section of the target corresponding to the point cloud cluster is within the area threshold range. If it is outside the area threshold range, the point cloud cluster is filtered out.

[0070] Furthermore, although the areas of point cloud clusters at different angles from multiple stations may vary, the differences will not be orders of magnitude. The area difference between point cloud cluster combinations can also be considered, i.e., the area difference between individual point cloud clusters within a combination. This area difference can be estimated based on the point cloud distribution. If the area difference of a point cloud cluster combination exceeds an area threshold, it is considered a false match and filtered out.

[0071] The filtration result can be obtained by using at least one of the three filtration methods mentioned above.

[0072] Based on the number of point cloud cluster combinations included in the filtering result, if the filtering result contains only one point cloud cluster combination, this filtering result can be directly used as the second multi-site cloud data. That is, the second multi-site cloud data includes point cloud clusters corresponding to the same target from different single-site cloud data. If the filtering result contains multiple point cloud cluster combinations, one point cloud cluster combination can be selected from this filtering result as the second multi-site cloud data.

[0073] In this embodiment, the point cloud cluster combination obtained by preliminary target matching is filtered by at least one of the following: the distance between the predicted trajectory point and the current trajectory point, the radar cross section of the corresponding target, and the area difference of the point cloud cluster combination. Combining the preliminary target matching with the filtering algorithm can balance matching accuracy and complexity, reduce the risk of mismatch, and thus improve the point cloud fusion accuracy.

[0074] Based on the above embodiments, determining the second multi-site cloud data based on the filtering results includes: If the filtering result includes only one point cloud cluster combination, then the point cloud cluster combination is used as the second multi-site cloud data; If the filtering result includes multiple point cloud cluster combinations, then the point cloud cluster combination with the smallest centroid distance among the multiple point cloud cluster combinations is taken as the second multi-point point cloud data.

[0075] Specifically, in the embodiments of this application, when the filtering results include multiple point cloud cluster combinations, the point cloud cluster combination with the smallest centroid distance among the multiple point cloud cluster combinations can be directly used as the second multi-site cloud data as a fallback solution to ensure that, under any circumstances, the second multi-site cloud data belonging to the same target can be obtained.

[0076] Based on the above embodiments, the distance threshold is determined based on the environmental complexity within the perceived overlapping coverage area and the number of targets per unit area.

[0077] Specifically, in this embodiment, the distance threshold for the initial target matching application can be dynamically adjusted based on the environmental complexity within the overlapping sensing coverage area and the number of targets per unit area. Here, the environmental complexity can be obtained by weighted summation of information such as the number of clutter, clutter type, and clutter intensity within the overlapping sensing coverage area. The clutter type can include dynamic clutter and static clutter, and the clutter intensity can be the power of the clutter.

[0078] An evaluation metric can be obtained by weighting and summing the environmental complexity and the number of targets. If the evaluation metric is greater than the metric threshold, the distance threshold can be increased to improve the probability of successful target matching, thereby enhancing perception accuracy through point cloud-level fusion. If the evaluation metric is less than or equal to the metric threshold, the distance threshold can be decreased to reduce false matches and algorithm complexity. The magnitude of increasing or decreasing the distance threshold can be set as needed and is not limited here.

[0079] For example, in scenarios with sparse targets and simple environments, the distance threshold should be increased; in scenarios with dense targets and complex environments, the distance threshold should be decreased.

[0080] In this embodiment, the distance threshold is determined by sensing the environmental complexity within the overlapping coverage area and the number of targets per unit area. This allows for adaptive adjustment of the distance threshold, thereby increasing the probability of successful target matching and reducing false matching and algorithm complexity.

[0081] Based on the above embodiments, determining the sensing overlap coverage area of ​​multiple base stations cooperating includes: Receive the initial point cloud data collected when each base station is turned on. The initial point cloud data corresponding to each base station includes the point cloud data of the target device with the positioning module. Based on the initial point cloud data corresponding to each base station and the positioning information of the positioning module, the sensing overlap coverage area is determined.

[0082] Specifically, in this embodiment of the application, the sensing overlap coverage area of ​​multiple base stations cooperating can be determined in advance through the following steps: First, when each base station is activated, a target device equipped with a positioning module can repeatedly operate within the sensing coverage area of ​​each base station. Here, the positioning module can be a Global Positioning System (GPS) module, a Real-time Kinematic (RTK) module, or another module. The target device can be a drone or other low-altitude target; if the target device is a drone, its operation represents the drone's flight.

[0083] During the repeated operation of the target device, the second type of baseband processing unit deployed at each base station can collect the corresponding single-site cloud data and report it to the first type of baseband processing unit. The first type of baseband processing unit can receive the initial point cloud data collected by the second type of baseband processing unit when each base station is turned on. The initial point cloud data corresponding to each base station includes the point cloud data of the target device.

[0084] The first type of baseband processing unit can determine the overlapping coverage areas of each sensing module by using the initial point cloud data corresponding to each base station and the positioning information of the positioning module.

[0085] Meanwhile, the first type of baseband processing unit can also determine the identification information of the second type of baseband processing unit corresponding to each sensing overlap coverage area.

[0086] After each base station is activated, the trajectory information of targets sensed by the base station can be further utilized to iteratively update the sensing coverage area of ​​the base station to cope with changes in coverage area caused by environmental changes. Furthermore, overlapping sensing coverage areas can also be updated to improve sensing accuracy.

[0087] In this embodiment of the application, by determining the overlapping coverage area of ​​perception, the amount of point cloud data that needs to be matched to the target can be reduced, which greatly reduces the complexity of the point cloud-level fusion algorithm and the probability of mismatch.

[0088] Based on the above embodiments, the step of determining the current trajectory point of the target corresponding to the second multi-site cloud data by performing weighted clustering on the second multi-site cloud data based on the reliability assessment values ​​of each data point in the second multi-site cloud data includes: Based on the distance between each data point in the second multi-site cloud data and the centroid of the second multi-site cloud data, abnormal data points in the second multi-site cloud data are removed.

[0089] Specifically, before performing weighted clustering on the second multi-site cloud data, a unified preliminary clustering can be performed on the second multi-site cloud data to remove outlier data points.

[0090] Here, outlier data points in the second multi-site cloud data can be removed based on the distance between each data point and the centroid of the second multi-site cloud data. For example, the distance corresponding to each data point in the second multi-site cloud data can be compared with a given threshold, and data points whose distances are greater than the given threshold are identified as outliers and removed. The remaining data points in the second multi-site cloud data, excluding outliers, constitute the preliminary clustering result.

[0091] In this embodiment of the application, by removing abnormal data points from the second multi-site cloud data, the quality of the second multi-site cloud data can be guaranteed, and the impact of abnormal data points on subsequent weighted clustering can be reduced.

[0092] like Figure 2 As shown, based on the above embodiments, this application also provides a multi-station collaborative point cloud-level fusion sensing method, applied to a second type of baseband processing unit, including: S21, acquire cloud data for a single site; S22, Calculate the overall reliability influencing factors of the single-site cloud data and the single-point reliability influencing factors of each data point in the single-site cloud data; S23, based on the overall reliability influencing factors and the single-point reliability influencing factors, determine the reliability assessment value of each data point in the single-site cloud data, and report the single-site cloud data and the reliability assessment value of each data point in the single-site cloud data to the first type of baseband processing unit.

[0093] Specifically, the multi-station collaborative point cloud-level fusion sensing method provided in this application embodiment is executed by a second type of baseband processing unit, which can be a subordinate baseband processing unit in a 5G-A distributed base station architecture. Multiple second type baseband processing units may be included, and each second type baseband processing unit can execute this method.

[0094] First, execute step S21 to collect single-site cloud data from the assigned base station. Then, step S22 is executed to calculate the overall reliability influencing factors of the single-site cloud data and the individual reliability influencing factors of each data point in the single-site cloud data. The overall reliability influencing factors of the single-site cloud data may include the mean square error of the single-site cloud data, and the individual reliability influencing factors of each data point may include the signal-to-noise ratio of the data point, the deviation from the trajectory prediction point, and the density of surrounding points.

[0095] Finally, step S23 is executed to determine the reliability assessment value of each data point in the single-site cloud data using the overall reliability influencing factors and the single-point reliability influencing factors. For example, the reliability assessment value of each data point in the single-site cloud data can be determined by weighted summation of the overall reliability influencing factors and the single-point reliability influencing factors. Alternatively, the overall reliability influencing factors and the single-point reliability influencing factors can be normalized separately, and the results of each normalization process can be weighted summation to determine the reliability assessment value of each data point in the single-site cloud data.

[0096] After determining the reliability assessment value of each data point, the second type of baseband processing unit can also report the single-site cloud data and the reliability assessment value of each data point in the single-site cloud data to the first type of baseband processing unit. This allows the first type of baseband processing unit to use the reliability assessment value of each data point to perform weighted clustering on the second multi-site cloud data belonging to the same target within the overlapping coverage area of ​​the perception, and to determine the current trajectory point of the target corresponding to the second multi-site cloud data.

[0097] This application provides a multi-station collaborative cloud-level fusion sensing method. By calculating the overall reliability influencing factors of single-site cloud data and the individual reliability influencing factors of each data point within the single-site cloud data, a reliability assessment value for each data point is calculated. This quantifies the reliability of each data point, thereby helping the first type of baseband processing unit to determine the current trajectory point of the target using the reliability assessment value. Furthermore, by having the second type of baseband processing unit calculate the reliability assessment value of each data point and report it to the first type of baseband processing unit for application, distributed data processing can be achieved, reducing the data processing load of the first type of baseband processing unit.

[0098] Based on the above embodiments, the overall reliability influencing factors include the mean square error of the single-site cloud data; The factors affecting the reliability of a single point include the signal-to-noise ratio of the data point, the deviation from the trajectory prediction point, and the density of surrounding points. The process of determining the reliability assessment value of each data point in the single-site cloud data based on the overall reliability influencing factors and the single-point reliability influencing factors includes: The mean square error of the single-site cloud data, the signal-to-noise ratio of each data point in the single-site cloud data, the deviation from the trajectory prediction point, and the density of surrounding points are normalized respectively to obtain the normalization results. The results of each normalization process are weighted and summed to obtain the reliability assessment value of each data point in the single-site cloud data.

[0099] Specifically, when calculating the reliability assessment value of each data point, the mean square error of the single-site cloud data, the signal-to-noise ratio (SNR) of each data point in the single-site cloud data, the deviation from the trajectory prediction point, and the density of surrounding points can be normalized to obtain the normalization results. In this embodiment, the mean square error of the single-site cloud data and the deviation of the data point from the trajectory prediction point are normalized by adding one and taking the reciprocal. The SNR of the data point and the density of surrounding points are normalized by the ratio to the maximum value. Subsequently, by weighting and summing the results of each normalization process, the reliability assessment value of each data point in the single-site cloud data can be obtained.

[0100] In one possible embodiment, the formula for calculating the reliability assessment value of the j-th data point is: ; in, Let be the reliability assessment value for the j-th data point, and MSE be the mean square error of the cloud data at the single site where the j-th data point is located. This is the normalized result of MSE. Let J be the signal-to-noise ratio of the j-th data point. This represents the maximum signal-to-noise ratio (SNR) of all data points in the second multi-site cloud data set. for The normalization result, Let j be the deviation between the j-th data point and the trajectory prediction point. for The normalization result, Let the density of the surrounding points of the j-th data point be denoted as . This represents the maximum value of the density of surrounding points for all data points in the second multi-site cloud data. for The normalization result, , , , These are the weights for each normalization result, which can be configured based on human experience.

[0101] In this embodiment, the mean square error of single-site cloud data, the signal-to-noise ratio of each data point in the single-site cloud data, the deviation from the trajectory prediction point, and the density of surrounding points are comprehensively considered. Through normalization processing, weighted summation, and other methods, the reliability assessment value of each data point in the single-site cloud data is finally obtained, which can make the obtained reliability assessment value more accurately represent the reliability of the data point.

[0102] Based on the above embodiments, the step of reporting the single-site cloud data and the reliability assessment values ​​of each data point in the single-site cloud data to the first type of baseband processing unit includes: Based on the reliability assessment value of each data point in the single-site cloud data, non-abnormal data points within each data point of the single-site cloud data are determined; Accordingly, the step of reporting the single-site cloud data and the reliability assessment values ​​of each data point in the single-site cloud data to the first type of baseband processing unit includes: The data of the non-abnormal points in the single-site cloud data and the reliability assessment value are reported to the first type of baseband processing unit.

[0103] Specifically, in this embodiment of the application, abnormal data points in the cloud data of a single site can be removed and non-abnormal data points can be retained by using the reliability assessment value of each data point.

[0104] Here, the reliability assessment value of each data point in the single-site cloud data can be compared with the reliability threshold. Data points with a reliability assessment value greater than or equal to the reliability threshold are retained as non-abnormal data points, while data points with a reliability assessment value less than the reliability threshold are removed as abnormal data points.

[0105] Furthermore, the data of non-abnormal points in the single-site cloud data and their reliability assessment values ​​can be reported to the first type of baseband processing unit to improve the data quality received by the first type of baseband processing unit and avoid the impact of abnormal data points in the single-site cloud data on the point cloud-level fusion results.

[0106] like Figure 3 As shown, based on the above embodiments, this application also provides a multi-station collaborative point cloud-level fusion sensing device, applied to a first type of baseband processing unit, including: The area determination module 31 is used to determine the sensing overlap coverage area of ​​multiple base stations cooperating; The target matching module 32 is used to perform target matching based on the first multi-site cloud data within the perceived overlapping coverage area, and determine the second multi-site cloud data belonging to the same target; The weighted clustering module 33 is used to determine the reliability assessment value of each data point in the second multi-site cloud data, and to perform weighted clustering on the second multi-site cloud data based on the reliability assessment value of each data point in the second multi-site cloud data to determine the current trajectory point of the target corresponding to the second multi-site cloud data.

[0107] Based on the above embodiments, the multi-station collaborative point cloud-level fusion sensing device provided in this application embodiment, wherein the weighted clustering module is specifically used for: Based on the reliability assessment values ​​of each data point in the second multi-site cloud data, the initial centroid of the second multi-site cloud data is determined; Using the initial centroid as the current centroid, and based on the distance between each data point in the second multi-site cloud data and the current centroid, the data points are clustered to obtain clusters; Based on the reliability assessment value of each data point in the cluster, the current centroid is iteratively updated until the position change of the new centroid in multiple consecutive iterations is less than a preset threshold, and then the new centroid is taken as the current trajectory point.

[0108] Based on the above embodiments, the multi-station collaborative point cloud-level fusion sensing device provided in this application embodiment, wherein the weighted clustering module is specifically used for: In each iteration, the reliability assessment value of each data point in the cluster is used as a weight to perform a weighted average of the position information of each data point in the cluster, thereby obtaining the position information of the new centroid.

[0109] Based on the above embodiments, the multi-station collaborative point cloud-level fusion sensing device provided in this application embodiment, wherein the target matching module is specifically used for: Based on the centroid distance and distance threshold of point cloud clusters of different single-site cloud data in the first multi-site cloud data, preliminary target matching is performed to obtain a combination of point cloud clusters that are suspected to belong to the same target. Based on the historical trajectory information and current velocity of the corresponding target in the point cloud cluster combination, the trajectory points of the corresponding target are predicted to obtain the predicted trajectory points. Based on at least one of the distance between the predicted trajectory point and the current trajectory point, the radar cross section of the corresponding target, and the area difference of the point cloud cluster combination, the point cloud cluster combination is filtered to obtain the filtering result. Based on the filtering results, the second multi-site cloud data is determined.

[0110] Based on the above embodiments, the multi-station collaborative point cloud-level fusion sensing device provided in this application embodiment, wherein the target matching module is specifically used for: If the filtering result includes only one point cloud cluster combination, then the point cloud cluster combination is used as the second multi-site cloud data; If the filtering result includes multiple point cloud cluster combinations, then the point cloud cluster combination with the smallest centroid distance among the multiple point cloud cluster combinations is taken as the second multi-point point cloud data.

[0111] Based on the above embodiments, the multi-station collaborative point cloud-level fusion sensing device provided in this application embodiment determines the distance threshold based on the environmental complexity within the sensing overlap coverage area and the number of targets per unit area.

[0112] Based on the above embodiments, the multi-station collaborative point cloud-level fusion sensing device provided in this application embodiment further includes an overlapping coverage area determination module, used for: Receive the initial point cloud data collected when each base station is turned on. The initial point cloud data corresponding to each base station includes the point cloud data of the target device with the positioning module. Based on the initial point cloud data corresponding to each base station and the positioning information of the positioning module, the sensing overlap coverage area is determined.

[0113] Based on the above embodiments, the multi-station collaborative point cloud-level fusion sensing device provided in this application embodiment further includes a rejection module, used for: Based on the distance between each data point in the second multi-site cloud data and the centroid of the second multi-site cloud data, abnormal data points in the second multi-site cloud data are removed.

[0114] Specifically, the functions of each module in the multi-station collaborative point cloud-level fusion sensing device provided in this application embodiment correspond one-to-one with the operation flow of each step in the above-mentioned method embodiment with the first type of baseband processing unit as the execution subject, and the achieved effect is also the same. For details, please refer to the above embodiments, and this application embodiment will not repeat them.

[0115] like Figure 4 As shown, based on the above embodiments, this application also provides a multi-station collaborative point cloud-level fusion sensing device, applied to a second type of baseband processing unit, including: Data acquisition module 41 is used to acquire cloud data from a single site; Influencing factor calculation module 42 is used to calculate the overall reliability influencing factors of the single site cloud data and the single-point reliability influencing factors of each data point in the single site cloud data. The reliability assessment module 43 is used to determine the reliability assessment value of each data point in the single-site cloud data based on the overall reliability influencing factors and the single-point reliability influencing factors, and to report the single-site cloud data and the reliability assessment value of each data point in the single-site cloud data to the first type of baseband processing unit.

[0116] Based on the above embodiments, the multi-station collaborative point cloud-level fusion sensing device provided in this application embodiment includes the mean square error of the single-site cloud data as an influencing factor on overall reliability. The factors affecting the reliability of a single point include the signal-to-noise ratio of the data point, the deviation from the trajectory prediction point, and the density of surrounding points. The reliability assessment module is specifically used for: The mean square error of the single-site cloud data, the signal-to-noise ratio of each data point in the single-site cloud data, the deviation from the trajectory prediction point, and the density of surrounding points are normalized respectively to obtain the normalization results. The results of each normalization process are weighted and summed to obtain the reliability assessment value of each data point in the single-site cloud data.

[0117] Based on the above embodiments, the reliability assessment module of the multi-station collaborative point cloud-level fusion sensing device provided in this application embodiment is specifically used for: Based on the reliability assessment value of each data point in the single-site cloud data, non-abnormal data points within each data point of the single-site cloud data are determined; The data of the non-abnormal points in the single-site cloud data and the reliability assessment value are reported to the first type of baseband processing unit.

[0118] Specifically, the functions of each module in the multi-station collaborative point cloud-level fusion sensing device provided in this application embodiment correspond one-to-one with the operation flow of each step in the above-mentioned method embodiment with the second type of baseband processing unit as the execution subject, and the achieved effect is also the same. For details, please refer to the above embodiments, and this application embodiment will not repeat them.

[0119] like Figure 5 As shown, based on the above embodiments, this application also provides a multi-station collaborative point cloud-level fusion sensing system, including: a first type of baseband processing unit 51 and at least two second type of baseband processing units 52, wherein the first type of baseband processing unit 51 and the second type of baseband processing unit 52 are communicatively connected. Figure 5 The image only shows the case where the multi-station collaborative point cloud-level fusion sensing system includes two second-type baseband processing units 52.

[0120] The first type of baseband processing unit 51 is used to execute the multi-station collaborative point cloud-level fusion sensing method provided in the above embodiments; The second type of baseband processing unit 52 is used to execute the multi-station collaborative point cloud-level fusion sensing method provided in the above embodiments.

[0121] The multi-station collaborative cloud-level fusion sensing system provided in this application embodiment removes abnormal data points from single-site cloud data based on reliability assessment values ​​using a second type of baseband processing unit. Then, a first type of baseband processing unit determines the overlapping coverage area of ​​the sensing and performs target matching and weighted clustering to ensure that the cluster centroid is closer to the data points with high reliability. This system can improve the fusion diversity gain of the overlapping coverage area of ​​the sensing, thereby improving the sensing accuracy.

[0122] Figure 6 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 6 As shown, the electronic device may include a processor 810, a communications interface 820, a memory 830, and a communication bus 840, wherein the processor 810, the communications interface 820, and the memory 830 communicate with each other via the communication bus 840. The processor 810 can call logical instructions in the memory 830 to execute the multi-station collaborative point cloud-level fusion sensing method provided in the above embodiments.

[0123] Furthermore, the logical instructions in the aforementioned memory 830 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to related technologies, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0124] On the other hand, this application also provides a computer program product, which includes a computer program that can be stored on a computer-readable storage medium. When the computer program is executed by a processor, the computer is able to execute the multi-station collaborative point cloud-level fusion perception method provided in the above embodiments.

[0125] Furthermore, this application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, is implemented to perform the multi-station collaborative point cloud-level fusion sensing method provided in the above embodiments. This computer-readable storage medium can be either a non-transitory computer-readable storage medium or a transient computer-readable storage medium; no specific limitation is made here.

[0126] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0127] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the parts that contribute to the related technology, can be embodied in the form of software products. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0128] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A multi-station cooperative point cloud level fusion perception method, characterized in that, The application is applied to a first type of baseband processing unit, comprising: determining a perception overlap coverage area cooperated by multiple base stations; based on the first multi-site cloud data in the perception overlap coverage area, target matching is performed to determine the second multi-site cloud data belonging to the same target; determining the reliability evaluation value of each data point in the second multi-site cloud data, and based on the reliability evaluation value of each data point in the second multi-site cloud data, the second multi-site cloud data is weighted clustered to determine the current trajectory point of the target corresponding to the second multi-site cloud data.

2. The multi-site cooperative point cloud level fusion perception method of claim 1, wherein, The method for determining the current trajectory point of the target corresponding to the second multi-site cloud data based on the reliability evaluation value of each data point in the second multi-site cloud data, comprising: based on the reliability evaluation value of each data point in the second multi-site cloud data, determining the initial clustering center of the second multi-site cloud data; taking the initial clustering center as the current clustering center, based on the distance between each data point in the second multi-site cloud data and the current clustering center, the clustering of each data point is performed to obtain a clustering cluster; based on the reliability evaluation value of each data point in the clustering cluster, the current clustering center is iteratively updated until the position change of the new clustering center in the continuous iteration process is less than the preset threshold value, and then the new clustering center is taken as the current trajectory point.

3. The multi-station cooperative point cloud level fusion perception method of claim 2, wherein, The method for iteratively updating the current clustering center based on the reliability evaluation value of each data point in the clustering cluster, comprising: in each iteration process, the reliability evaluation value of each data point in the clustering cluster is taken as the weight to perform weighted average on the position information of each data point in the clustering cluster to obtain the position information of the new clustering center.

4. The multi-site cooperative point cloud level fusion perception method of claim 1, wherein, The method for determining the second multi-site cloud data based on the first multi-site cloud data in the perception overlap coverage area, comprising: based on the centroid distance of the point cloud cluster in the different single-site cloud data in the first multi-site cloud data and the distance threshold value, preliminary target matching is performed to obtain a point cloud cluster combination suspected to belong to the same target; based on the historical trajectory information and the current speed of the corresponding target of the point cloud cluster combination, the trajectory point of the corresponding target is predicted to obtain a predicted trajectory point; based on at least one of the distance between the predicted trajectory point and the current trajectory point, the radar scattering cross section of the corresponding target, and the area difference of the point cloud cluster combination, the point cloud cluster combination is filtered to obtain a filtering result; based on the filtering result, the second multi-site cloud data is determined.

5. The multi-station cooperative point cloud level fusion perception method of claim 4, wherein, The method for determining the second multi-site cloud data based on the filtering result, comprising: if the filtering result only includes one point cloud cluster combination, the point cloud cluster combination is taken as the second multi-site cloud data; if the filtering result includes multiple point cloud cluster combinations, the point cloud cluster combination with the smallest centroid distance in the multiple point cloud cluster combinations is taken as the second multi-site cloud data.

6. The multi-station cooperative point cloud level fusion perception method of claim 4, wherein, The distance threshold value is determined based on the environmental complexity in the perception overlap coverage area and the number of targets per unit area.

7. The multi-site cooperative point cloud level fusion perception method according to any one of claims 1-6, characterized in that, The determining the perception overlapping coverage area cooperated by the plurality of base stations comprises: receiving initial point cloud data collected when each base station is started, and each initial point cloud data corresponding to each base station comprises point cloud data of the target device with the positioning module; determining the perception overlapping coverage area based on the initial point cloud data corresponding to each base station and the positioning information of the positioning module.

8. The multi-site cooperative point cloud level fusion perception method according to any one of claims 1-6, characterized in that, The weighting clustering of the second multi-point cloud data based on the reliability evaluation value of each data point in the second multi-point cloud data to determine the current track point of the target corresponding to the second multi-point cloud data previously comprises: eliminating abnormal data points in the second multi-point cloud data based on the distance between each data point in the second multi-point cloud data and the centroid of the second multi-point cloud data. 9.A multi-station cooperative point cloud level fusion perception method, characterized in that, Applied to a second type of baseband processing unit, comprising: acquiring single-point cloud data; calculating overall reliability influence factors of the single-point cloud data and single-point reliability influence factors of each data point in the single-point cloud data; determining reliability evaluation values of each data point in the single-point cloud data based on the overall reliability influence factors and the single-point reliability influence factors, and reporting the single-point cloud data and the reliability evaluation values of each data point in the single-point cloud data to a first type of baseband processing unit.

10. The multi-site cooperative point cloud level fusion perception method of claim 9, wherein, The overall reliability influence factors comprise mean square error of the single-point cloud data; The single-point reliability influence factors comprise signal-to-clutter ratio, deviation value from a track prediction point and peripheral point density of the data point; The determining of the reliability evaluation values of each data point in the single-point cloud data based on the overall reliability influence factors and the single-point reliability influence factors comprises: respectively performing normalization processing on the mean square error of the single-point cloud data, the signal-to-clutter ratio, the deviation value from the track prediction point and the peripheral point density of each data point in the single-point cloud data to obtain each normalization processing result; performing weighted summation on the each normalization processing result to obtain the reliability evaluation value of each data point in the single-point cloud data.

11. The multi-site cooperative point cloud level fusion perception method of claim 9, wherein, The reporting of the single-point cloud data and the reliability evaluation values of each data point in the single-point cloud data to the first type of baseband processing unit previously comprises: determining non-abnormal data points in each data point in the single-point cloud data based on the reliability evaluation values of each data point in the single-point cloud data; Correspondingly, the reporting of the single-point cloud data and the reliability evaluation values of each data point in the single-point cloud data to the first type of baseband processing unit comprises: reporting data and reliability evaluation values of the non-abnormal points in the single-point cloud data to the first type of baseband processing unit.

12. A multi-station cooperative point cloud level fusion perception apparatus, characterized in that, Applied to a first type of baseband processing unit, comprising: a region determining module configured to determine a perception overlapping coverage area cooperated by a plurality of base stations; a target matching module configured to perform target matching based on first multi-point cloud data in the perception overlapping coverage area to determine second multi-point cloud data belonging to the same target; The weighted clustering module is configured to determine a reliability evaluation value of each data point in the second multi-site cloud data, and perform weighted clustering on the second multi-site cloud data based on the reliability evaluation value of each data point in the second multi-site cloud data, to determine a current trajectory point of the target corresponding to the second multi-site cloud data.

13. A multi-station cooperative point cloud level fusion perception apparatus, comprising: The second type of baseband processing unit comprises: The data acquisition module is configured to acquire single-site cloud data. The influence factor calculation module is configured to calculate an overall reliability influence factor of the single-site cloud data and a single-point reliability influence factor of each data point in the single-site cloud data. The reliability evaluation module is configured to determine a reliability evaluation value of each data point in the single-site cloud data based on the overall reliability influence factor and the single-point reliability influence factor, and report the single-site cloud data and the reliability evaluation value of each data point in the single-site cloud data to the first type of baseband processing unit.

14. A multi-station cooperative point cloud level fusion perception system, comprising: The first type of baseband processing unit and the second type of baseband processing unit are communicatively connected. The first type of baseband processing unit is configured to perform the multi-station cooperative point cloud level fusion perception method according to any one of claims 1-8. The second type of baseband processing unit is configured to perform the multi-station cooperative point cloud level fusion perception method according to any one of claims 9-11. The processor executes the computer program to implement the multi-station cooperative point cloud level fusion perception method according to any one of claims 1-11.

15. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The computer program is executed by the processor to implement the multi-station cooperative point cloud level fusion perception method according to any one of claims 1-11.

16. A computer readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the multi-station cooperative point cloud level fusion perception method according to any one of claims 1-11.

17. A computer program product comprising a computer program, characterized in that, ​

Images