Optimization method and device for vehicle sensing target screening, electronic equipment and automobile

Abstract

The application relates to the technical field of automatic driving systems, in particular to a vehicle sensing target screening optimization method and device, electronic equipment and a car, wherein the method comprises the following steps: acquiring a first vehicle target set to be screened; filtering out targets identified by a single radar from the first vehicle target set to obtain a second vehicle target set; judging whether a map corresponding to the first vehicle target set is a high-precision map; if the map is a high-precision map, identifying an isolation mark according to vehicle position information, and obtaining a filtering range according to the isolation mark; if the map is not a high-precision map, identifying an isolation mark according to a sensing fusion result, and obtaining a filtering range according to the isolation mark; filtering out targets in the filtering range from the second vehicle target set to obtain a third vehicle target set; and then filtering out targets of a selected target type from the third vehicle target set to obtain a fourth vehicle target set. The embodiment of the application is used for filtering out misidentified targets.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving system technology, specifically to an optimization method for vehicle sensor target selection, an optimization device for vehicle sensor target selection, an electronic device, an autonomous driving domain controller, and a vehicle. Background Technology

[0002] One of the key aspects of autonomous driving technology is target detection in the driving environment, which is also the foundation and core of scene reconstruction. The real-time performance and accuracy of target detection are fundamental to ensuring correct decision-making by the driver or the autonomous driving system. While existing vehicles are equipped with various types of sensors, these sensors currently have limited capabilities and significant differences, leading to unstable target recognition in the driving environment. Issues such as target loss, inconsistent vehicle type identification, or misidentification can occur, posing a challenge to effective target reconstruction. Furthermore, the massive amount of data generated by various sensors places high demands on the real-time performance of data processing.

[0003] While existing technologies offer optimized refactoring methods, they do not address the problems encountered in actual mass production, nor do they mention scenario-based optimization solutions. Summary of the Invention

[0004] The purpose of this invention is to provide an optimized method, apparatus, electronic device, and automobile for vehicle sensing target screening, so as to at least solve the problems of high error rate and low reliability in vehicle sensing target identification in the prior art.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] An optimized method for vehicle sensing target filtering includes: acquiring a first set of vehicle targets to be filtered; filtering out targets identified by a single radar from the first set of vehicle targets to obtain a second set of vehicle targets; determining whether the map corresponding to the first set of vehicle targets is a high-precision map; if the map is a high-precision map, identifying isolation markers based on vehicle location information, and obtaining a filtering range based on the isolation markers; if the map is not a high-precision map, identifying isolation markers based on sensor fusion results, and obtaining a filtering range based on the isolation markers; filtering out targets within the filtering range from the second set of vehicle targets to obtain a third set of vehicle targets; and filtering out targets of a selected target type from the third set of vehicle targets to obtain a fourth set of vehicle targets.

[0007] Based on the aforementioned technical means, the vehicle target set is screened and filtered out from the aspects of data acquisition device attributes, spatial dimensions, and target attributes, so as to avoid falsely detected target information as much as possible and provide an accurate data foundation for target reconstruction.

[0008] Furthermore, targets identified by a single radar include targets detected by a single radar that are not fused with output data from other target detection devices.

[0009] Based on the aforementioned technical means, the targets identified by a single radar are explained and defined, providing a basis for judgment to eliminate the high-risk factor of misidentification caused by single radar identification.

[0010] Furthermore, the process of identifying isolation markers based on vehicle location information and obtaining a filtering range based on the isolation markers includes: acquiring isolation markers in the high-precision map that are within a preset range of the vehicle location information, wherein the isolation markers include lane dividers, lane dividers, or the edge of the road on the same side; determining the distance between the isolation markers and the vehicle location information based on the high-precision map; and defining the area outside the distance from the vehicle as the filtering range.

[0011] Using the aforementioned technical methods, the information in the high-precision map is used to filter out misidentified targets, thereby improving the accuracy of the filtering.

[0012] Furthermore, the process of identifying isolation markers based on the sensor fusion results and determining the filtering range based on the isolation markers includes: identifying lane lines and lane barriers based on the sensor fusion results; obtaining lane barriers and the edge of the road on the same side based on the lane lines; using the lane barriers, lane barriers, and the edge of the road on the same side as the isolation markers; and determining the area outside the isolation markers as the filtering range.

[0013] Based on the aforementioned technical means, a filtering method for misidentified targets in non-high-precision maps is provided, expanding the applicable scenarios of this filtering method.

[0014] Furthermore, the selected target type is obtained through the following steps: obtaining the target type attribute output by the front-end recognition software and the user's selection of the type attribute; obtaining the selected target type based on the user's selection.

[0015] Based on the aforementioned technical means, filtering by target type attributes further improves the reliability of the filtered targets.

[0016] Furthermore, the method also includes: outputting the fourth vehicle target set to a rendering carrier.

[0017] Based on the aforementioned technical means, the intended use of the vehicle target set is provided, making this solution more relevant to real-world scenarios.

[0018] An optimization device for vehicle sensing target selection, the device comprising:

[0019] The set acquisition module is used to acquire the first set of vehicle targets to be filtered;

[0020] The first filtering module is used to filter out targets identified by a single radar from the first vehicle target set to obtain a second vehicle target set.

[0021] The map determination module is used to determine whether the map corresponding to the first vehicle target set is a high-precision map;

[0022] The first range module is used to identify isolation markers based on vehicle location information and obtain the filtering range based on the isolation markers if the map is a high-precision map.

[0023] The second range module is used to identify isolation markers based on the sensor fusion results and obtain the filtering range based on the isolation markers if the map is not a high-precision map.

[0024] The second filtering module is used to filter out targets within the filtering range from the second vehicle target set to obtain a third vehicle target set; and

[0025] The third filtering module is used to filter out targets of a selected target type from the third vehicle target set to obtain a fourth vehicle target set.

[0026] Based on the above technical means, the steps in the corresponding methods are implemented using corresponding functional modules. The vehicle target set is screened and filtered out from the aspects of data acquisition device attributes, spatial dimensions, and target attributes, so as to avoid falsely detected target information and provide an accurate data foundation for target reconstruction.

[0027] An electronic device includes: one or more processors; and a storage device for storing one or more programs, which, when executed by the one or more processors, cause the electronic device to implement the optimized method for vehicle sensing target selection as described above.

[0028] An autonomous driving domain controller is configured to implement the aforementioned optimized method for vehicle sensing target selection.

[0029] An automobile, the automobile including the aforementioned electronic equipment or the aforementioned autonomous driving domain controller.

[0030] The beneficial effects of this invention are:

[0031] (1) The present invention designs a filtering mechanism based on single radar target identification, taking into account the limitations of single-modal data types in the data acquisition process, and effectively reduces the misidentification of targets in single-modal data lacking data fusion.

[0032] (2) The present invention designs a filtering mechanism based on spatial dimension; by utilizing the advantages of existing high-precision maps, the stability of the filtering effect is improved and the probability of recognition error is reduced.

[0033] (3) The present invention designs a target type-based filtering mechanism, which filters according to the front-end identification type, effectively removing unreliable targets. Attached Figure Description

[0034] Figure 1 This is a flowchart of the optimized method for vehicle sensing target selection according to the present invention;

[0035] Figure 2 This is a schematic diagram illustrating the steps for obtaining the filtering range under a high-precision map in this invention;

[0036] Figure 3 This is a schematic diagram illustrating the steps for obtaining the filtering range under non-high-precision maps in this invention;

[0037] Figure 4 This diagram illustrates the implementation steps of the optimized method for vehicle sensing target selection according to the present invention.

[0038] Figure 5 This is a schematic diagram of the structure of the vehicle sensing target selection optimization device of the present invention.

[0039] Among them, 10-set acquisition module; 20-first filtering module; 30-map judgment module; 40-first range module; 50-second range module; 60-second filtering module; 70-third filtering module. Detailed Implementation

[0040] The embodiments of the present invention will be described below with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention and not for limiting the scope of protection of the present invention.

[0041] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Therefore, the drawings only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0042] This embodiment proposes an optimized method for vehicle sensing target selection. Figure 1 This is a flowchart of the optimized method for vehicle sensing target selection according to the present invention. Figure 1 As shown, the method includes:

[0043] Step 101: Obtain the first set of vehicle targets to be filtered; the first set of vehicle targets can be obtained in ways such as generating it through the vehicle's relevant detection system, importing it from other storage media, or obtaining it through existing communication methods, etc., which are not limited here.

[0044] Step 102: Filter out targets identified by a single radar from the first vehicle target set to obtain a second vehicle target set. Targets identified by a single radar in this step include targets detected by a single radar that have not been fused with output data from other target detection devices. This is because using a single radar for sensing and detection is inaccurate; it is necessary to combine it with other target detection devices such as cameras to perform target data fusion to improve accuracy. However, if the target is only detected by a single radar target detector, there is a high probability of false identification, such as detecting nearby obstacles, which has low accuracy.

[0045] Step 103: Determine whether the map corresponding to the first vehicle target set is a high-precision map. A high-precision map, also known as an "autonomous driving map" or "intelligent vehicle base map," refers to a navigation map with high resolution and high abundance of features, achieving both absolute and relative accuracy at the centimeter level (10 to 20 centimeters). High-precision maps contain rich and accurate road detail information, such as lane lines, lane center lines, and lane attribute changes. Furthermore, the lane model must include mathematical parameters such as road curvature, slope, heading, and cross slope. Currently, high-precision maps are typically only available on some urban expressways or highways and other structured roads.

[0046] Step 104: If the map is a high-precision map, identify the isolation markers based on the vehicle location information, and obtain the filtering range based on the isolation markers; since the high-precision map contains more detailed road information, it can achieve better filtering effect and lower error rate when combined with high-precision vehicle location information.

[0047] Step 104': If the map is not a high-precision map, identify the isolation markers based on the sensor fusion results, and determine the filtering range based on the isolation markers. In the case of a non-high-precision map, it is necessary to identify the isolation markers based on the detection results of sensor fusion, and then determine the filtering range. The isolation markers in this step are the same as those defined in Step 104, including lane dividers, lane barriers, or the edge of the road on this side. Among them, lane dividers include, but are not limited to: single or double yellow lines used to separate opposing lanes, and traffic signs indicating that entry is not permitted according to traffic regulations. Lane barriers include lane guardrails, isolation fences, cones, stone blocks, and obstacles used to mark areas that are not allowed to be entered or crossed. The edge of the road on this side includes the boundary of the road in the direction of travel.

[0048] Step 105: Filter out targets within the filtering range from the second vehicle target set to obtain a third vehicle target set. This step filters targets based on the filtering range, which is determined according to the isolation markers in the driving environment. This step provides a filtering rule in a spatial dimension.

[0049] Step 106: Filter out targets of the selected target type from the third vehicle target set to obtain a fourth vehicle target set. The target type in this step is based on the target type identified by the software. This software refers to various existing target recognition or target fusion software that labels the generated targets according to internal judgment rules. When the front-end recognition software can identify the output target as "other" or "uncertain," the target is considered unreliable and can be filtered out.

[0050] In another embodiment, Figure 2 This is a schematic diagram illustrating the steps for obtaining the filtering range under a high-precision map in this invention. For example... Figure 2 As shown, the isolation marker is identified based on the vehicle location information, and the filtering range is obtained based on the isolation marker, including:

[0051] Step 201: Obtain the isolation markers in the high-precision map that are within a preset range of the vehicle location information. The isolation markers include lane isolation lines, lane barriers, or the edge of the road on this side. Since there are a lot of road details in the high-precision map, combined with the well-positioned vehicle location information, misidentification of the target can be filtered directly through the high-precision map.

[0052] Step 202: Determine the distance between the isolation marker and the vehicle location information based on the high-precision map. For example, based on the vehicle location information, the distance to each lane line can be obtained from the high-precision map. If the vehicle is currently in the leftmost lane, then the left line of the vehicle is the lane isolation line separating oncoming lanes, and targets outside this left line can be filtered out. The distance between the vehicle and this left line is obtained through the high-precision map. Similarly, if the vehicle is located in the second lane from the left based on the vehicle location information, the left line of the lane to the left of the left lane can be used as the lane isolation line, and targets outside it can be filtered out. The same applies if the vehicle is located in the rightmost lane, which will not be elaborated here. In other scenarios, lane isolation is achieved using lane barriers or fences, which cover the lane isolation lines. In this case, the lane barriers can be used to determine the distance.

[0053] Step 203: Define the area outside the stated distance from the vehicle as the filtering range. Based on the distance obtained from the high-precision map and vehicle location information, targets in the oncoming lane are filtered out to avoid their identification and instability.

[0054] In this embodiment, target filtering is performed using high-precision maps and precise positioning. Compared to filtering based solely on sensor information, the filtering effect is more stable and the probability of error is lower.

[0055] In another embodiment, Figure 3 This is a schematic diagram illustrating the steps for obtaining the filtering range under non-high-precision maps in this invention. For example... Figure 3 As shown, isolation markers are identified based on the sensor fusion results, and the filtering range is obtained based on the isolation markers, including:

[0056] Step 301: Identify lane lines and lane dividers in the sensor fusion results; In this embodiment, it is not possible to obtain a large amount of accurate road information from the map, so it is necessary to identify road information based on the sensor fusion results, wherein the road information includes lane lines and lane dividers.

[0057] Step 302: Obtain the lane divider line and the edge of the road on this side based on the lane lines, and use the lane divider line, lane divider line, and edge of the road on this side as the isolation markers; this step mainly determines the isolation markers using lane lines. For example: if there are three lanes in the current driving direction, and the vehicle is in the middle lane, it is necessary to filter out target information that is far from the left-left lane line (lane divider line) and the right-right lane line (edge ​​of the road on this side). This is because the usability of this target information is poor for display or driving safety.

[0058] Step 303: Determine the area outside the isolation marker as the filtering range. In some scenarios, the filtering range can be determined based on lane dividers. For example, if there is a guardrail next to the lane, targets outside the guardrail need to be filtered out. In other scenarios, the filtering range can be determined by combining lane dividers and lane dividers. For example, if the left lane has been detected as a lane divider and the guardrail is around the left lane, then targets outside the left lane can be determined to be misidentified or filtered out for scene reconstruction.

[0059] In this embodiment, although the filtering effect is reduced compared to the previous embodiment which uses high-precision maps and precise positioning for target filtering, it reduces the amount of data in the vehicle target set and improves the accuracy of scene reconstruction by filtering out unreliable targets.

[0060] In another embodiment, the selected target type is obtained through the following steps: acquiring the target type attribute output by the front-end recognition software and the user's selection of the type attribute; and obtaining the selected target type based on the user's selection. The target type in this step is based on the target type identified by the software, which can be any existing target recognition or target fusion software that labels generated targets according to internal judgment rules. When the front-end recognition software can identify the output target as "other" or "uncertain," the target is considered unreliable and can be filtered out.

[0061] In another embodiment, the method further includes outputting the fourth set of vehicle targets to a rendering carrier. This additional step is used to present the filtered target results. The output carrier can be an image format, such as *.jpg, *.png, etc.; or a video format, such as *.avi, *.mov, *.mp4, etc., depending on the actual needs of the reconstruction software. The rendering effect depends on the set quality and needs to be balanced with the rendering time.

[0062] Figure 4 This diagram illustrates the implementation steps of the optimized method for vehicle sensing target selection according to the present invention. Figure 4 As shown, in this embodiment, the method includes: receiving sensor / map / location information; filtering out single radar targets; determining whether the target meets the high-precision map range and the positioning is good; if so, filtering the target based on the high-precision positioning; otherwise, filtering based on lane lines / fences; filtering out targets of other / uncertain types; and outputting to the rendering carrier.

[0063] This embodiment also proposes an optimized device for vehicle sensing target selection, such as... Figure 5 As shown, the device includes:

[0064] The set acquisition module 10 is used to acquire the first set of vehicle targets to be filtered.

[0065] The first filtering module 20 is used to filter out targets identified by a single radar from the first vehicle target set to obtain a second vehicle target set;

[0066] Map determination module 30 is used to determine whether the map corresponding to the first vehicle target set is a high-precision map;

[0067] The first range module 40 is used to identify isolation markers based on vehicle location information and obtain a filtering range based on the isolation markers if the map is a high-precision map.

[0068] The second range module 50 is used to identify isolation markers based on the sensor fusion results and obtain the filtering range based on the isolation markers if the map is not a high-precision map.

[0069] The second filtering module 60 is used to filter out targets within the filtering range from the second vehicle target set to obtain a third vehicle target set; and

[0070] The third filtering module 70 is used to filter out targets of a selected target type from the third vehicle target set to obtain a fourth vehicle target set.

[0071] This embodiment also proposes an electronic device, which includes: one or more processors; and a storage device for storing one or more programs, which, when executed by the one or more processors, enable the electronic device to implement the optimized method for vehicle sensing target screening as described above.

[0072] This embodiment also proposes an Autonomous Driving Domain Controller (ADC), which is configured to implement the aforementioned optimized method for vehicle sensor target selection. The aforementioned optimized method for vehicle sensor target selection in this invention is implemented in the ADC. As an intelligent computing platform, the Autonomous Driving Domain Controller (ADC) is designed for Level 3 to Level 5 autonomous driving applications. It integrates computationally intensive sensor data processing and sensor fusion with control strategy development into a single control unit, and facilitates the establishment of a structured and organized vehicle controller network.

[0073] This embodiment also proposes a vehicle that includes the aforementioned electronic equipment or the aforementioned autonomous driving domain controller. When the vehicle uses the aforementioned electronic equipment or autonomous driving domain controller, the processing of scene reconstruction is better, significantly improving the safety of autonomous driving or assisted driving.

[0074] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a microcontroller, chip, or processor to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, portable hard drive, read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0075] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit. This integrated unit can be implemented in hardware or software. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application.

[0076] The optional embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the embodiments of the present invention are not limited to the specific details described above. Within the scope of the technical concept of the embodiments of the present invention, various simple modifications can be made to the technical solutions of the embodiments of the present invention, and these simple modifications all fall within the protection scope of the embodiments of the present invention. It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, the embodiments of the present invention will not further describe the various possible combinations.

[0077] Furthermore, various different embodiments of the present invention can be combined in any way, as long as they do not violate the spirit of the embodiments of the present invention, they should also be regarded as the content disclosed by the embodiments of the present invention.

Claims

1. An optimized method for vehicle sensing target selection, characterized in that, The method includes: Obtain the first set of vehicle targets to be filtered; Targets identified by a single radar are filtered out from the first set of vehicle targets to obtain a second set of vehicle targets; the targets identified by a single radar include targets detected by a single radar that have not been fused with the output data of other target detection devices. Determine whether the map corresponding to the first set of vehicle targets is a high-precision map; If the map is a high-precision map, the isolation signs are identified based on the vehicle location information, and the filtering range is obtained based on the isolation signs. The isolation signs include lane isolation lines, lane barriers, or the edge of the road on this side. If the map is not a high-precision map, the isolation markers are identified based on the sensor fusion results, and the filtering range is obtained based on the isolation markers, with lane barriers, lane isolation lines and the edge of the road on this side as the isolation markers; Targets within the filtering range are filtered out from the second vehicle target set to obtain a third vehicle target set; Targets of the selected target type are filtered out from the third vehicle target set to obtain a fourth vehicle target set; the selected target type is obtained through the following steps: obtaining the target type attribute output by the front-end recognition software and the user's selection of the type attribute; the selected target type is obtained according to the user's selection.

2. The optimized method for vehicle sensing target selection according to claim 1, characterized in that, Identify the isolation marker based on the vehicle location information, and obtain the filtering range based on the isolation marker, including: Obtain isolation markers in the high-precision map that are within a preset range of the vehicle's location information. The isolation markers include lane isolation lines, lane barriers, or the edge of the road on this side. The distance between the isolation marker and the vehicle location information is determined based on the high-precision map; The area outside the distance from the vehicle is defined as the filtering range.

3. The optimized method for vehicle sensing target selection according to claim 1, characterized in that, Isolation markers are identified based on the sensor fusion results, and the filtering range is obtained based on the isolation markers, including: The sensor fusion results identified lane lines and lane dividers. Based on the lane lines, lane separation lines and the edge of the road on this side are obtained, and the lane barriers, lane separation lines, and the edge of the road on this side are used as the isolation markers; The area outside the isolation marker is defined as the filtering range.

4. The optimized method for vehicle sensing target selection according to claim 1, characterized in that, The method further includes: outputting the fourth vehicle target set to a rendering carrier.

5. An optimized device for vehicle sensing target selection, characterized in that, The device includes: The set acquisition module is used to acquire the first set of vehicle targets to be filtered; The first filtering module is used to filter out targets identified by a single radar from the first vehicle target set to obtain a second vehicle target set; the targets identified by a single radar include: targets detected by a single radar and not fused with the output data of other target detection devices. The map determination module is used to determine whether the map corresponding to the first vehicle target set is a high-precision map; The first range module is used to identify isolation markers based on vehicle location information and obtain a filtering range based on the isolation markers if the map is a high-precision map; the isolation markers include lane isolation lines, lane barriers, or the edge of the road on this side. The second range module is used to identify isolation markers based on the sensor fusion results if the map is not a high-precision map, and to obtain the filtering range based on the isolation markers; the isolation markers are lane barriers, lane lines and the edge of the road on this side. The second filtering module is used to filter out targets within the filtering range from the second vehicle target set to obtain a third vehicle target set; and The third filtering module is used to filter out targets of a selected target type from the third vehicle target set to obtain a fourth vehicle target set; the selected target type is obtained through the following steps: obtaining the target type attribute output by the front-end recognition software and the user's selection of the type attribute; obtaining the selected target type according to the user's selection.

6. An electronic device, characterized in that, The electronic device includes: One or more processors; A storage device for storing one or more programs, which, when executed by the one or more processors, cause the electronic device to implement the optimized method for vehicle sensing target screening as described in any one of claims 1 to 4.

7. An autonomous driving domain controller, characterized in that, The autonomous driving domain controller is configured to implement the optimized method for vehicle sensing target selection as described in any one of claims 1 to 4.

8. A car, characterized in that, The vehicle includes the electronic device as described in claim 6 or the autonomous driving domain controller as described in claim 7.

Images