Differential Extraction Method, System and Storage Medium Combining Crowdsourcing Historical Data

Abstract

The present invention discloses a differential extraction method, system and storage medium combining crowdsourcing historical data, including: obtaining SLAM data within a period of time, clustering the current SLAM data and the unclustered data in the previous period of time, putting the data with unsuccessful clustering into the unclustered data of the historical data, and waiting to be clustered with the data in the next period of time; performing aggregation processing on the data with successful clustering, and performing differential extraction on the data with successful aggregation in the current period of time and the data with successful aggregation in the previous period of time, identifying the newly added and changed data in the current period of time relative to the previous period of time, directly outputting the newly added data and putting it into the newly added part of the historical data, putting the changed data into the updated part of the historical data, and calculating the output confidence of the changed data, and only outputting the updated data when the calculated output confidence is greater than the preset confidence. The present invention solves the problems that the results of historical data cannot be effectively utilized and the calculation amount is high at the present stage.

Description

Technical Field

[0001] The present invention belongs to the technical fields of autonomous driving and high-precision maps, and particularly relates to a differential extraction method, system, and storage medium combining crowdsourcing historical data. Background Art

[0002] With the increasing amount of data collected by crowdsourcing, how to produce high-precision maps with the least amount of data calculation and effectively combine historical data to reduce the calculation amount is a very important issue. The current method is to directly add historical data for algorithm calculation, but as the data volume gradually increases, this method will lead to an increase in the amount of computation. Repeating the calculation of the already processed data will also result in an increase in time cost. Therefore, how to effectively utilize the results of historical data to reduce the calculation cost and increase the confidence of the results is crucial.

[0003] Currently, the differential extraction method is mainly used to extract new data or compare different versions of data. For example, a high-precision navigation map accuracy registration method and device disclosed in patent document CN108195382A, which obtains new road data by performing differential processing on navigation road network data and comparing it with the data of the original navigation master library and the basic road network data after matching and rectification. Another example is an image processing device, a camera device, and an image processing method disclosed in patent document CN104350735A, which extracts differential noise components by performing differential processing on the captured image and a smoothed image generated by smoothing the captured image. Another example is a graph information processing device, a map information processing method, and a map information processing program disclosed in CN110914888A, which extracts differential data by comparing the moving trajectory images of the trajectory information of multiple moving bodies with scalar-form images based on vector-form data that are different from the above-mentioned moving trajectory images. The above methods all obtain the differences between data through the differential method, but lack the differential use of the data generation results.

[0004] Therefore, it is necessary to develop a differential extraction method, system, and storage medium combining crowdsourcing historical data. Summary of the Invention

[0005] The purpose of the present invention is to provide a differential extraction method, system, and storage medium combining crowdsourcing historical data, which can solve the problems that the results of historical data cannot be effectively utilized and the calculation amount is high at the present stage.

[0006] In a first aspect, a differential extraction method combining crowdsourcing historical data according to the present invention includes the following steps:

[0007] For the first time period, obtain the SLAM data within this time period, clean and cluster the SLAM data, aggregate the successfully clustered results, output the successfully aggregated data and put the successfully aggregated data into the newly added historical data, and put the unsuccessfully clustered data into the unclustered data of the historical data;

[0008] For non-first time periods, obtain the SLAM data within this time period, clean the SLAM data, use the cleaned data and the unclustered data from the previous time period as new inputs for clustering, and put the unsuccessfully clustered data into the unclustered data of the historical data, waiting to be clustered with the data from the next time period; perform aggregation processing on the successfully clustered data, and perform differential extraction on the successfully aggregated data in this time period and the successfully aggregated data in the previous time period to identify the newly added and changed data in this time period relative to the previous time period. Directly output the newly added data and put it into the newly added historical data, put the changed data into the updated historical data, and calculate the output confidence of the changed data. Only output the changed data when the calculated output confidence is greater than the preset confidence.

[0009] Optionally, the data differential extraction is performed by referring to all fields of the data object. If a field has changed, it is considered that the data has been updated, and the updated data is put into the updated historical data. If the data object has never appeared in the historical data, it is considered newly added.

[0010] Optionally, the clustering is to cluster multiple objects of the same subtype at the same position into a cluster.

[0011] Optionally, the aggregation is to aggregate multiple objects of the same subtype at the same position that have been clustered into a cluster into one object.

[0012] In a second aspect, a differential extraction system combining crowdsourcing historical data according to the present invention includes a memory and a controller. The memory stores a computer-readable program, and when the computer-readable program is called by the controller, it can execute the steps of the differential extraction method for combining crowdsourcing historical data according to the present invention.

[0013] In a third aspect, a storage medium according to the present invention stores a computer-readable program, and when the computer-readable program is called by the controller, it can execute the steps of the differential extraction method for combining crowdsourcing historical data according to the present invention.

[0014] The present invention has the following advantages: Aiming at the problem that the current differential extraction method only acts on the differences between analysis data and lacks the reuse of data generation results, the present invention combines the characteristics of crowdsourcing data collection to propose a differential extraction method that can effectively convert historical results. The present invention can not only reduce the amount of data operation, but also improve the confidence of data results. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the embodiments. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0016] Figure 1 It is a flowchart of the clustering method for the parent brand and non-parent brand tags in this embodiment;

[0017] Figure 2 It is a flowchart of the clustering method for the sub-brand tags in this embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] The present invention will be described in detail below with reference to the drawings.

[0019] As Figure 1 shown, a differential extraction method combining crowdsourcing historical data includes the following steps:

[0020] For the first time period, obtain the SLAM data within this time period, clean and cluster the SLAM data, perform aggregation processing on the successfully clustered results, output the successfully aggregated data and put the successfully aggregated data into the new addition of historical data, and put the unsuccessfully clustered data into the unclustered data of historical data.

[0021] In this embodiment, the time period can be one day or other time cycles, and the setting of the time period is adjusted according to specific situations.

[0022] For non-first time periods, obtain the SLAM data within the time period, clean the SLAM data, use the cleaned data and the unclustered data from the previous time period as new inputs for clustering, and put the data for which clustering is unsuccessful into the unclustered data of the historical data, waiting to be clustered with the data in the next time period; perform aggregation processing on the data for which clustering is successful, and perform differential extraction on the data successfully aggregated in the current time period and the data successfully aggregated in the previous time period to identify the newly added and changed data in the current time period relative to the previous time period. Directly output the newly added data and put it into the newly added part of the historical data, put the changed data into the updated part of the historical data, and calculate the output confidence of the changed data. Only output the changed data when the calculated output confidence is greater than the preset confidence level.

[0023] In this embodiment, the data differential extraction is performed by referring to all fields of the data object. If a field has changed, it is considered that the data has been updated, and the updated data is put into the updated part of the historical data. If the data object has never appeared in the historical data, it is considered newly added.

[0024] In this embodiment, the clustering is to cluster multiple objects with the same subtype of arrows at the same position into a cluster. Assuming arrow types, there are subtypes such as right turn, straight + left turn, straight + right turn, left turn + right turn, left front, right front, straight + U-turn, left turn + U-turn, left U-turn, right U-turn, no left turn, etc., then cluster the right turn arrow subtypes at the same position into a cluster, and the same for the other subtypes.

[0025] In this embodiment, the aggregation is as follows: Aggregate multiple objects with the same subtype at the same position that are clustered into a cluster into one object. Specifically, aggregate multiple left turn arrow subtypes at the same position that are clustered into a cluster into one left turn arrow subtype for output, and the same for the other subtypes.

[0026] In this embodiment, a differential extraction system combining crowdsourcing historical data includes a memory and a controller. The memory stores a computer-readable program, and when the computer-readable program is called by the controller, it can execute the steps of the differential extraction method for combining crowdsourcing historical data as described in this embodiment.

[0027] In this embodiment, a storage medium stores a computer-readable program, and when the computer-readable program is called by the controller, it can execute the steps of the differential extraction method for combining crowdsourcing historical data as described in the present utility model.

[0028] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention shall be equivalent replacement methods and are all included in the protection scope of the present invention.

Claims

1. A differential extraction method combining crowdsourcing historical data, characterized in that: Including the following steps: For the first time period, obtain the SLAM data within this time period, clean and cluster the SLAM data, aggregate the successfully clustered results, output the successfully aggregated data and put the successfully aggregated data into the newly added historical data, and put the unsuccessfully clustered data into the unclustered data of the historical data; For non-first time periods, obtain the SLAM data within this time period, clean the SLAM data, use the cleaned data and the unclustered data in the previous time period as new inputs for clustering, and put the unsuccessfully clustered data into the unclustered data of the historical data, waiting to be clustered with the data in the next time period; Aggregate the successfully clustered data, and perform differential extraction on the successfully aggregated data in this time period and the successfully aggregated data in the previous time period to identify the newly added and changed data in this time period relative to the previous time period. Directly output the newly added data and put it into the newly added historical data, put the changed data into the updated historical data, and calculate the output confidence of the changed data. Only output the changed data when the calculated output confidence is greater than the preset confidence.

2. The differential extraction method combining crowdsourcing historical data according to claim 1, characterized in that: The data differential extraction is to refer to all fields of the data object. If the field has changed, it is considered that the data has been updated, and the updated data is put into the updated historical data. If the data object has never appeared in the historical data, it is considered newly added.

3. The differential extraction method combining crowdsourcing historical data according to claim 1 or 2, characterized in that: The clustering is to cluster multiple objects of the same subtype at the same position into a cluster.

4. The differential extraction method combining crowdsourcing historical data according to claim 1 or 2, characterized in that: The aggregation is to aggregate multiple objects of the same subtype at the same position that have been clustered into a cluster into one object.

5. A differential extraction system that combines crowdsourcing historical data, characterized in that: Including a memory and a controller, the memory stores a computer-readable program, and when the computer-readable program is called by the controller, it can execute the steps of the differential extraction method for combining crowdsourcing historical data according to any one of claims 1 to 4.

6. A storage medium, characterized in that: It stores a computer-readable program, and when the computer-readable program is called by the controller, it can execute the steps of the differential extraction method for combining crowdsourcing historical data according to any one of claims 1 to 4.

Images