Cloud global sensing system

Abstract

The invention relates to a cloud global sensing system which is used for sensing a target road section, and the target road section is divided into at least two sub-road sections. The system comprises a data acquisition device which is correspondingly arranged on each sub-road section and is used for acquiring perception data of the corresponding sub-road section; wherein the types of the data acquisition devices correspondingly arranged on at least two sub-road sections are different; at least two sub-universities, each sub-university corresponding to one sub-road segment, the sub-universities being used for acquiring the sensing data of the corresponding sub-road segment and sensing the sensing target based on the sensing data to obtain the sensing result of the sensing target in the sub-road segment; and the cloud universe is used for obtaining the sensing result corresponding to each sub universe, integrating the sensing results corresponding to each sub universe, and obtaining the sensing result of the sensing target in the target road section. The cloud global sensing system can be compatible with sensing results of various sensors, the sensing results of a plurality of road sections are spliced, and the tracking rate of a sensing target in a large range is improved.

Description

Technical Field

[0001] This application relates to the field of vehicle-road collaborative technology, and particularly to a cloud global perception system. Background Art

[0002] Recently, the widely concerned autonomous driving technology uses sensors such as on-vehicle lidar, millimeter-wave radar, ultrasonic radar, and video cameras to sense the surrounding environment of the vehicle, and realizes the recognition of the surrounding environment of the vehicle through on-vehicle edge computing, and then realizes autonomous driving of the perceived target, and has also made great progress. However, road traffic is an extremely complex giant system, and the road traffic environment changes rapidly. It is impossible to fully and quickly master the dynamic traffic environment only by traditional traffic perception means and the limited sensors installed on autonomous driving vehicles. Moreover, due to the requirements of on-vehicle sensors for small size and high cost, they cannot be popularized on the vehicles of the vast number of travelers. The vehicle-road collaborative technology has emerged as the times require.

[0003] In vehicle-road collaborative technology, the road traffic conditions can be sensed by roadside sensors. At present, there are many types of roadside sensors, and the efficiency of target tracking in a large range is low. Summary of the Invention

[0004] Based on this, it is necessary to provide a cloud global perception system with a high tracking rate for the above technical problems.

[0005] In a first aspect, this application provides a cloud global perception system, which is used to perceive a target road section, and the target road section is divided into at least two sub-road sections; the system includes:

[0006] A data acquisition device correspondingly set for each sub-road section, which is used to acquire the perception data of the corresponding sub-road section; wherein, there are at least two sub-road sections with different types of data acquisition devices correspondingly set;

[0007] At least two sub-global regions, each sub-global region corresponding to a sub-road section, which is used to acquire the perception data of the corresponding sub-road section, and perform perception target perception based on the perception data to obtain the perception result of the perception target in the sub-road section;

[0008] The cloud global region is used to acquire the corresponding perception results of each sub-global region, and integrate the corresponding perception results of each sub-global region to obtain the perception result of the perception target in the target road section.

[0009] In one of the embodiments, the data acquisition device includes an image acquisition module and a point cloud data acquisition module;

[0010] The image acquisition module is used to acquire the image data of the cross-section of the corresponding sub-road section;

[0011] A point cloud data acquisition module, which is used to acquire point cloud data within a corresponding sub-section;

[0012] The image data is used to determine the perception targets in the corresponding sub-section, and the point cloud data is used to determine the perception results of the perception targets in the corresponding sub-section.

[0013] In one embodiment, the system further includes a main control module;

[0014] The main control module is used to acquire image data, determine the corresponding relationship between the image acquisition module and the point cloud data acquisition module according to the image data, and send the corresponding relationship to the cloud universe;

[0015] The cloud universe is further used to determine the perception results of the perception targets in each sub-section according to the corresponding relationship, frame the perception results and send them to the main control module.

[0016] In one embodiment, multiple data acquisition devices are arranged in the corresponding sub-section of the sub-universe; correspondingly, the sub-universe is specifically used to process the perception data collected by each data acquisition device according to the installation position connection sequence of the data acquisition devices in the corresponding sub-section, so as to obtain the perception results of the perception targets in the corresponding sub-section.

[0017] In one embodiment, the perception results include the tracking results and event detection results of the perception targets; correspondingly, the cloud universe includes a splicing module and an event deduplication module;

[0018] The splicing module is used to splice the tracking results of the perception targets in at least two sub-universes according to the connection sequence of the corresponding sub-sections of at least two sub-universes, so as to obtain the tracking results of the same perception target in the target section;

[0019] The event deduplication module is used to deduplicate all event detection results according to the perception targets corresponding to each event detection result.

[0020] In one embodiment, the cloud universe further includes a trajectory processing module;

[0021] The trajectory processing module is used to perform trajectory simulation according to the tracking results of the same perception target in the target section, so as to obtain the motion trajectory of the same perception target.

[0022] In one embodiment, the trajectory processing module is further used to predict the tracking results of any perception target in any sub-section according to the tracking results of any perception target in other sub-sections when the perception data of any perception target is missing in any sub-section.

[0023] In one embodiment, the cloud universe further includes a trajectory smoothing module;

[0024] A trajectory smoothing module for smoothing the motion trajectory of the same perceived target.

[0025] In one embodiment, the cloud-wide domain further includes a preset target detection module;

[0026] The preset target detection module is used to determine the corresponding perceived target as the preset target type when there is a perceived target in the perception result of the target section that meets the conditions of the preset target type.

[0027] In one embodiment, the cloud-wide domain further includes an early warning module;

[0028] The early warning module is used to send an early warning message to the corresponding perceived target when the event detection result of the perceived target in the target section is the preset event type.

[0029] In one embodiment, the sub-domain is further used to assign a global identifier to the perceived target in the corresponding sub-section;

[0030] The cloud-wide domain is further used to assign a global identifier to the corresponding perceived target when there is a perceived target without a global identifier.

[0031] The above cloud-wide domain perception system is used to perceive a target section, and the target section is divided into at least two sub-sections; the system includes: a data acquisition device respectively set for each sub-section, used to acquire the perception data of the corresponding sub-section; wherein, there are at least two sub-sections with different types of data acquisition devices; at least two sub-domains, each sub-domain corresponding to a sub-section, the sub-domain is used to acquire the perception data of the corresponding sub-section, and perform perceived target perception based on the perception data to obtain the perception result of the perceived target in the sub-section; the cloud-wide domain is used to acquire the perception result corresponding to each sub-domain, integrate the perception results corresponding to each sub-domain, and obtain the perception result of the perceived target in the target section. The cloud-wide domain perception system can be compatible with the perception results of various sensors, splice the perception results of multiple sections at the same time, and improve the tracking rate of perceived targets in a large range. Description of the Drawings

[0032] Figure 1 It is a structural block diagram of the cloud-wide domain perception system in one embodiment;

[0033] Figure 2 It is a schematic diagram of sub-section division in one embodiment;

[0034] Figure 3 It is a structural schematic diagram of the data acquisition device in one embodiment;

[0035] Figure 4 It is a schematic diagram of the connection sequence of the installation positions of the data acquisition devices in one embodiment;

[0036] Figure 5 It is a structural block diagram of the cloud-wide area in an embodiment;

[0037] Figure 6 It is a structural block diagram of the cloud-wide area system in an embodiment. Detailed implementation manners

[0038] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0039] The cloud-wide area perception system provided by the embodiment of the present application is used to perceive a target road section, and the target road section is divided into at least two sub-road sections; as Figure 1 shown, the system includes:

[0040] A data acquisition device 1 provided corresponding to each sub-road section, which is used to acquire the perception data of the corresponding sub-road section; wherein, there are at least two types of data acquisition devices 1 provided corresponding to different sub-road sections;

[0041] At least two sub-wide areas 2, each sub-wide area 2 corresponds to a sub-road section, and the sub-wide area 2 is used to acquire the perception data of the corresponding sub-road section and perform perception target perception based on the perception data to obtain the perception result of the perception target in the sub-road section;

[0042] The cloud-wide area 3 is used to acquire the corresponding perception results of each sub-wide area 2, integrate the corresponding perception results of each sub-wide area, and obtain the perception result of the perception target in the target road section.

[0043] Among them, the target road section refers to the entire road section range that the cloud-wide area perception system needs to perceive. The cloud-wide area perception system provided by the embodiment of the present application is applicable to the situation where the target road section is relatively long. In a relatively long target road section, due to the changing environment of the target road section and different perception requirements, there will also be differences in the data acquisition devices set at various positions on the target road section. Therefore, according to the perception requirements of the target road section and other situations, the target road section is divided into multiple sub-road sections, and each sub-road section corresponds to a type of data acquisition device. The data acquisition device is used to collect the perception data of the corresponding road section, and the data acquisition device includes various types, which can be lidar, millimeter wave radar or bayonet camera.

[0044] In one embodiment, as Figure 2 shown, the total length of the target road section is 12 kilometers, including a 6-kilometer No. 1 tunnel, a service area and a 6-kilometer No. 2 tunnel. Then, it is set that every 3 kilometers is a section, and the range between every two sections is used as a sub-road section. Therefore, Figure 2The target road section shown is divided into 4 sub-road sections. Among them, the data acquisition devices for the first sub-road section 201 and the third sub-road section 203 can be millimeter-wave radars, and the data acquisition devices for the second sub-road section 202 and the fourth sub-road section can be lidars.

[0045] Each sub-universe 2 corresponds to a sub-road section. The sub-universe is used to process the perception data of the corresponding sub-road section, perceive the perception targets that appear in the sub-road section, and obtain the perception results of each perception target that appears in the sub-road section. Among them, the perception results include the information of the detected perception targets, and the perception results include information such as license plate, position, size, vehicle type, appearance, speed, and acceleration. It should be noted that considering the distance of each sub-road section and the perception range of the data acquisition device installed in this sub-road section, there may be a situation where multiple data acquisition devices are set in a sub-road section. At this time, the sub-universe not only needs to sort out the perception data obtained by each data acquisition device to obtain the perception results of each data acquisition device for the perception target, but also needs to integrate the multiple perception results of the same perception target determined by multiple data acquisition devices to obtain the perception result of the perception target in the sub-road section.

[0046] The cloud universe 3 receives the perception results sent by all sub-universes, and splices all the perception results from a macroscopic perspective. When splicing, it is spliced based on the connection order of all sub-road sections. It should be noted that the cloud universe can receive the perception results corresponding to each type of data acquisition device and process them. Since the multiple perception results are determined according to the perception data obtained by different types of data acquisition devices, there may be slight differences between the multiple perception results, such as data accuracy, data error size, etc. Therefore, when the cloud universe performs integration, it will first uniformly process the data, determine the corresponding relationship between the data, and then perform integration.

[0047] Specifically, for example, in the target road section, there is a single-base millimeter-wave radar and multiple lidars. The single-base millimeter-wave radar directly determines the perception result of the perception target in this sub-road section through the obtained perception data. The chromed millimeter-wave radar processes the perception data through the sub-universe corresponding to its sub-road section to obtain the perception result of the perception target in this sub-road section. When the position information is included in the perception result, the perception accuracies of the millimeter-wave radar and the lidar are different, and there may be a certain error in the obtained positions. At this time, the cloud universe can match the position information by setting an error range, and when the matching is successful, integrate the two perception results.

[0048] The system provided by the above embodiment is used to sense a target road section, and the target road section is divided into at least two sub-road sections; the system includes: a data acquisition device respectively arranged for each sub-road section, which is used to acquire the sensing data of the corresponding sub-road section; wherein, there are at least two sub-road sections with different types of data acquisition devices; at least two sub-universes, each sub-universe corresponds to a sub-road section, and the sub-universe is used to acquire the sensing data of the corresponding sub-road section, and perform sensing target sensing based on the sensing data to obtain the sensing result of the sensing target in the sub-road section; a cloud universe, which is used to acquire the sensing result corresponding to each sub-universe, and integrate the sensing result corresponding to each sub-universe to obtain the sensing result of the sensing target in the target road section. The cloud universe sensing system can be compatible with the sensing results of various sensors, and can splice the sensing results of multiple road sections at the same time, improving the tracking rate of sensing targets in a large range.

[0049] In one of the embodiments, as Figure 3 shown, the data acquisition device 1 includes an image acquisition module 101 and a point cloud data acquisition module 102;

[0050] The image acquisition module 101 is used to acquire the image data of the cross-section of the corresponding sub-road section;

[0051] The point cloud data acquisition module 102 is used to acquire the point cloud data within the corresponding sub-road section;

[0052] The image data is used to determine the sensing target in the corresponding sub-road section, and the point cloud data is used to determine the sensing result of the sensing target in the corresponding sub-road section.

[0053] Among them, the cross-section of the sub-road section refers to the position where the sensing target enters or leaves the sub-road section selected according to the moving direction of the sensing target in the sub-road section. By setting an image acquisition device at the cross-section of the sub-road section, the appearance information of the sensing target entering or leaving the sub-road section can be acquired. For example, when the sensing target is a vehicle, information such as the license plate, color, and size of the vehicle can be acquired through the image data. It can be known that in the above example, the license plate information of the vehicle is the unique information of the vehicle, so the sensing target can be uniquely determined through the image data. Specifically, the image acquisition module 101 may include a bayonet camera.

[0054] The point cloud data acquired by the point cloud data acquisition module 102 arranged in each sub-road section can identify the sensing target appearing in the sub-road section and the sensing result of the sensing target, but the identity of the sensing target cannot be determined. Based on the actual setting situation, there is a corresponding relationship between the image acquisition module 101 and the point cloud data acquisition module 102 of each sub-road section. Through this corresponding relationship, the image data within the corresponding sub-road section, and the point cloud data within the corresponding sub-road section, the identity of each sensing target in the sub-road section can be determined, and the sensing result of the sensing target in the sub-road section can be improved.

[0055] In the system provided by the above embodiment, by setting an image acquisition module in the cross section of each sub-road section, the identity of the perceived target can be determined, thereby determining the perception result of the perceived target.

[0056] In one of the embodiments, the system further includes a main control module;

[0057] The main control module is used to obtain image data, determine the corresponding relationship between the image acquisition module and the point cloud data acquisition module according to the image data, and send the corresponding relationship to the cloud domain;

[0058] The cloud global domain is also used to determine the perception results of the perception target in each sub-road section according to the corresponding relationship, frame the perception results and send them to the main control module.

[0059] The image acquisition module 101 sends the collected image data to the main control module, and the image data sent also includes the identifier of the image acquisition module 101 and the identifier of the corresponding point cloud data acquisition module 102. After receiving the image data, the main control module processes the image data to identify the perceived target in the image and obtain the identity information of the perceived target. Then, the identity information of the perceived target, the identifier of the image acquisition module 101 that obtains the identity information, and the identifier of the corresponding point cloud data acquisition module 102 are sent to the cloud global domain 3. The sub-global domain 2 where the point cloud data acquisition module 102 is located is determined by the cloud global domain, and the identity information of the perceived target is added to the perception result of the perceived target in the corresponding sub-road section of the sub-global domain 2 to enrich the perception result of the perceived target.

[0060] After determining the perception results of the perception target in the target road section, the cloud global domain 3 frames the perception results and sends them to the main control module. Specifically, the perception results of a single perception target in the target road section can be framed, and the perception results of all perception targets in the target road section that appear within a period of time can be uniformly framed and sent to the main control module.

[0061] In the system provided by the above embodiment, the image data is processed by the main control module, which reduces the workload of the entire cloud domain, improves the work efficiency of the entire cloud domain, and ensures that the perception results are obtained in a timely manner.

[0062] In one embodiment, if Figure 4 As shown, the corresponding sub-sections of sub-domain 2 are provided with a plurality of data acquisition devices [01, 02, ..., 16]; accordingly, sub-domain 2 is specifically used to process the perception data collected by each data acquisition device according to the installation position connection sequence of the data acquisition devices in the corresponding sub-sections, and obtain the perception results of the perception target in the corresponding sub-sections.

[0063] Among them, the installation position connection order refers to the order of multiple data acquisition devices installed according to the moving direction of the perceived target in the sub-section. In Figure 4 , the installation position connection order of 16 data acquisition devices is [01, 02,..., 16]. When the sub-universe processes the perception data collected by multiple data acquisition devices, it processes the perception data according to the installation position connection order to obtain the perception result of the perceived target in the corresponding sub-section. Specifically, the perception result of each data acquisition device on the perceived target is determined by the perception data collected by each data acquisition device, and then according to the installation position connection order, the perception results of multiple data acquisition devices on the perceived target are spliced together to form the perception result of the perceived target in this sub-section.

[0064] In the system provided by the above embodiment, based on the installation position connection order, the perception results of multiple data acquisition devices on the perceived target are spliced, ensuring that the perception results sent by the sub-universe to the cloud universe are concise and orderly, and can improve the processing efficiency of the cloud universe.

[0065] In one of the embodiments, as Figure 5 shown, the perception result includes the tracking result and the event detection result of the perceived target; correspondingly, the cloud universe 3 includes a splicing module 301 and an event deduplication module 302;

[0066] The splicing module 301 is used to splice the tracking results of the perceived target in at least two sub-universes according to the connection order of the corresponding sub-sections of at least two sub-universes to obtain the tracking result of the same perceived target in the target section;

[0067] The event deduplication module 302 is used to perform deduplication processing on all event detection results according to the perceived target corresponding to each event detection result.

[0068] Among them, the connection order refers to the order of all sub-sections determined according to the moving direction of the perceived target in the target section. For example, in Figure 2 , the connection order of 4 sub-sections is the first sub-section 201, the second sub-section 202, the third sub-section 203, and the fourth sub-section 204. Based on the connection order, the tracking results of the perceived target in at least two sub-universes are spliced to obtain the tracking result of the same perceived target in the target section. Specifically, when each sub-universe determines the tracking result of the perceived target, it will generate a universe identifier for each perceived target in this sub-universe, and the universe identifier carries the identifier of this sub-universe. For example, the universe identifier generated by sub-universe 01 for the perceived target carries "01", and the universe identifier generated by sub-universe 02 for the perceived target carries "02". Based on the connection order, the identifier order of all sub-universes is determined, and the splicing is performed according to the order during splicing.

[0069] Under normal circumstances, the identification of a sub-universe is set by the connection order of the corresponding sub-sections. In addition, the splicing module 301 can also directly splice the perception results, that is, not only splice the tracking results, but also splice the event detection results of each perception target to obtain the event detection results of the perception target in the target section.

[0070] After splicing the event detection results of each perception target, for the same type of event of the same perception target, multiple sub-universes may have multiple result records. At this time, deduplicate all the event detection results according to the perception target corresponding to each event detection result. For example, for the perceived perception target A, the event detection result in the first sub-section shows a speeding event, recorded as [A, first sub-section identifier, speeding event identifier]; the event detection result in the second sub-section shows a speeding event, recorded as [A, second sub-section identifier, speeding event identifier]; at this time, the two records can be merged into one, as [A, second sub-section identifier + second sub-section identifier, speeding event identifier], and one of the records can be deleted.

[0071] In the system provided by the above embodiment, the cloud universe can splice and process the perception results of the perception target in multiple sub-sections to obtain the accurate perception result of each perception target, and ensure the tracking rate of the perception target within the entire range of the target section.

[0072] In one of the embodiments, the cloud universe further includes a trajectory processing module 303;

[0073] The trajectory processing module 303 is used to perform trajectory simulation based on the tracking results of the same perception target in the target section to obtain the motion trajectory of the same perception target.

[0074] Among them, the trajectory processing module is used to determine the motion trajectory of the perception target in the target section according to the tracking results of the perception target in the target section. Through the trajectory processing module, the motion visualization of the perception target can be realized.

[0075] In one of the embodiments, the trajectory processing module 303 is further used to predict the tracking result of any perception target in any sub-section according to the tracking results of any perception target in other sub-sections when the perception data of any perception target is missing in any sub-section.

[0076] The perception results between two adjacent sub-road segments may have a certain overlap or omission due to the installation location of the data acquisition device. Or, due to the failure of the data acquisition device, there may be an omission of the perception results within the entire range or a certain range of the corresponding sub-road segment. For the above situations, in the case of missing tracking results, the trajectory processing module is also used to predict the missing tracking results based on the tracking results of the perception target in other sub-road segments. Specifically, the tracking results of the perception target in the missing sub-road segment can be predicted based on the tracking results in the two adjacent sub-road segments of the missing sub-road segment; or the tracking results of the perception target in the missing sub-road segment can be directly predicted based on the known tracking results in all sub-road segments. The perception results of the perception target in the target road segment are obtained through trajectory prediction, improving the perception effect within the large-scale overall area of the target road segment.

[0077] In one embodiment, the cloud-wide domain further includes a trajectory smoothing module 304;

[0078] The trajectory smoothing module 304 is used to smooth the motion trajectories of the same perception target.

[0079] In the case where there is an overlap or a small amount of omission in the perception results between two adjacent road segments, the trajectories corresponding to the perception results of the same perception target in the two adjacent sub-road segments are smoothed to obtain a smoother motion trajectory of the perception target. When performing trajectory smoothing, partial smoothing within partial sub-road segments of the in-domain motion trajectories belonging to the same perception target can be performed; and / or, overall smoothing of the motion trajectories of the same perception target in the target road segment can be performed.

[0080] In the system provided by the above embodiment, the splicing area is smoothed to avoid a large-scale distortion of the perception results of the perception target.

[0081] In one embodiment, the cloud-wide domain further includes a preset target detection module 305;

[0082] The preset target detection module is used to determine the corresponding perception target as the preset target type when there is a perception target in the perception results of the target road segment that meets the preset target type condition.

[0083] The preset target type condition is used to screen the perception targets. In one embodiment, the preset target type can be set according to "two types of passenger vehicles and one type of dangerous goods vehicle" for determining vehicles of the "two types of passenger vehicles and one type of dangerous goods vehicle" type. Specifically, the perception targets can be judged based on the information in the perception results of each perception target. For example, by summarizing the external characteristics of "two types of passenger vehicles and one type of dangerous goods vehicle" and setting the preset target type condition according to this specific setting, the vehicles of "two types of passenger vehicles and one type of dangerous goods vehicle" are screened out.

[0084] In the above example, the existing "two passengers and one dangerous goods" detection devices exist independently. These devices only perform a small amount of detection on the front-end pictures, resulting in a waste of resources. By integrating the "two passengers and one dangerous goods" into the cloud global domain, setting preset target type conditions according to the "two passengers and one dangerous goods", receiving the perception results of multiple downstream sub-global domains through the cloud global domain, and uniformly interacting with the "two passengers and one dangerous goods", it also avoids the interaction between the "two passengers and one dangerous goods" detection devices and multiple sub-global domains.

[0085] In the system provided by the above embodiment, by integrating a preset target detection module in the cloud global domain, the screening of specific perception targets is realized, and there is no need to additionally add entity detection devices in the perception system.

[0086] In one of the embodiments, the cloud global domain further includes an early warning module 306;

[0087] The early warning module is used to send an early warning message to the corresponding perception target when the event detection result of the perception target in the target section is a preset event type.

[0088] The perception result determined by the cloud global domain in the target section has global macroscopicity. Based on this perception result, the motion state data of the perception target and the event detection result are obtained. When the event detection result of the perception target is a preset event type, an early warning message is sent to the corresponding perception target. For example, by the real-time driving speed of the perception target or the driving acceleration of the perception target, etc., or by identifying the driving path of the perception target, it is judged whether the perception target is engaged in dangerous driving. When it is determined that there is dangerous driving, there must be a dangerous driving event identifier in its event detection result.

[0089] The system provided by the above embodiment can perform real-time safety monitoring on the moving perception targets, provide accompanying information services, and give driving safety early warning messages.

[0090] In one of the embodiments, the sub-global domain is further used to assign a global identifier to the perception target in the corresponding sub-section;

[0091] The cloud global domain is further used to assign a global identifier to the corresponding perception target when there is a perception target without a global identifier.

[0092] Since the perception result determined by the cloud global domain in the target section has global macroscopicity, the cloud global domain splices the perception results of the sub-global domains from a macroscopic perspective, can record the perception results of the splicing failures, and the perception targets that exist in the section but do not have a global ID. For such perception results, global identifiers are assigned to the corresponding perception targets to ensure that all the perception results of the perceived perception targets are recorded, which to a certain extent guarantees the tracking rate within the entire global range of the target section.

[0093] In one of the embodiments, such asFigure 6 As shown, a cloud-wide system is provided. The system is used to sense a target road section, and the target road section is divided into sub-road sections 6011 - 6014. The system includes:

[0094] The sub-road section 6011 obtains sensing data through single-base millimeter-wave radars 01 - 03, and the base station uses it as a sub-region to obtain a sensing result. The sub-road section 6012 obtains sensing data through lidars 04 - 17. The sub-region 01 processes the sensing data to obtain a sensing result, and sends the sensing result to the cloud-wide region. The sub-road section 6013 obtains sensing data through a single-base millimeter-wave radar 18, and the base station uses it as a sub-region to obtain a sensing result. The sub-road section 6014 obtains sensing data through lidars 19 - N. The sub-region 02 processes the sensing data to obtain a sensing result, and sends the sensing result to the cloud-wide region.

[0095] A bayonet camera, which is used to obtain image data of the cross-section of each sub-road section.

[0096] The cloud-wide region includes an event deduplication module, a splicing module, a trajectory smoothing module, a trajectory processing module, a preset target detection module, and an early warning module. It is used to obtain the corresponding sensing results and image data of each sub-region, integrate the corresponding sensing results of each sub-region, obtain the sensing result of the sensing target in the target road section, frame the sensing result, and send it to the main control.

[0097] The cloud-wide system provided in the above embodiment can be compatible with the sensing results of various sensors, unify the splicing of multi-section sensing results, and at the same time perform an online trajectory smoothing function from a macroscopic perspective; the data flow is clear; it is compatible with non-sensed road sections, and through the bayonet cameras of non-sensed road sections and the upstream whole-region sensing, multi-target simulation trajectory prediction of non-sensed road sections is carried out.

Claims

1. A cloud-wide perception system, characterized in that, The system is used to sense a target road section, and the target road section is divided into at least two sub-road sections; the system includes: A data acquisition device correspondingly set for each sub-road section, which is used to acquire the sensing data of the corresponding sub-road section; wherein, there are at least two sub-road sections with different types of data acquisition devices correspondingly set; At least two sub-universes, each sub-universe corresponding to a sub-road section, and the sub-universe is used to acquire the sensing data of the corresponding sub-road section and perform sensing target sensing based on the sensing data to obtain the sensing result of the sensing target in the sub-road section; A cloud universe, which is used to acquire the sensing result corresponding to each sub-universe, integrate the sensing results corresponding to each sub-universe, and obtain the sensing result of the sensing target in the target road section.

2. The system according to claim 1, wherein The data acquisition device includes an image acquisition module and a point cloud data acquisition module; The image acquisition module is used to acquire the image data of the cross-section of the corresponding sub-road section; The point cloud data acquisition module is used to acquire the point cloud data within the corresponding sub-road section; The image data is used to determine the sensing target in the corresponding sub-road section, and the point cloud data is used to determine the sensing result of the sensing target in the corresponding sub-road section.

3. The system according to claim 2, wherein The system further includes a main control module; The main control module is used to acquire the image data, determine the corresponding relationship between the image acquisition module and the point cloud data acquisition module according to the image data, and send the corresponding relationship to the cloud universe; The cloud universe is further used to determine the sensing result of the sensing target in each sub-road section according to the corresponding relationship, frame the sensing results and send them to the main control module.

4. The system according to claim 1, characterized in that, A plurality of data acquisition devices are set for the sub-road section corresponding to the sub-universe; correspondingly, the sub-universe is specifically used to process the sensing data collected by each data acquisition device according to the installation position connection sequence of the data acquisition devices in the corresponding sub-road section to obtain the sensing result of the sensing target in the corresponding sub-road section.

5. The system according to claim 1, characterized in that, The sensing result includes the tracking result and event detection result of the sensing target; correspondingly, the cloud universe includes a splicing module and an event deduplication module; The splicing module is used to splice the tracking results of the sensing target in at least two sub-universes according to the connection sequence of the sub-road sections corresponding to the at least two sub-universes to obtain the tracking result of the same sensing target in the target road section; The event deduplication module is used to perform deduplication processing on all event detection results according to the sensing target corresponding to each event detection result.

6. The system according to claim 5, characterized in that The cloud universe further includes a trajectory processing module; The trajectory processing module is used to perform trajectory simulation according to the tracking result of the same sensing target in the target road section to obtain the motion trajectory of the same sensing target.

7. The system according to claim 6, wherein The trajectory processing module is further used to predict the tracking result of any sensing target in any sub-road section according to the tracking result of the any sensing target in other sub-road sections when the sensing data of the any sensing target is missing in any sub-road section.

8. The system according to claim 6, characterized in that The cloud universe further includes a trajectory smoothing module; The trajectory smoothing module is used to perform smoothing processing on the motion trajectory of the same sensing target.

9. The system according to claim 5, wherein The cloud universe further includes a preset target detection module; The preset target detection module is used to determine that a corresponding sensed target is of a preset target type when there is a sensed target in the sensing result of the target road section that meets the conditions of the preset target type.

10. The system according to claim 5, characterized in that, The cloud-wide area further includes an early warning module; The early warning module is used to send an early warning message to a corresponding sensed target when the event detection result of the sensed target in the target road section is a preset event type.

11. The system according to claim 1, characterized in that, The sub-wide area is further used to assign a global identifier to the sensed target in the corresponding sub-road section; The cloud-wide area is further used to assign a global identifier to a corresponding sensed target when there is a sensed target without a global identifier.

Images