Multi-target association method and device of vehicle, vehicle and storage medium

Abstract

The application relates to the technical field of automatic driving perception targets, in particular to a multi-target association method and device of a vehicle, a vehicle and a storage medium, wherein the method comprises the following steps: acquiring road information around the vehicle and tracking information of a tracking target collected by multiple sensors; target information of the tracking target in the road information collected by each sensor is extracted, a cost matrix is constructed based on the tracking information and / or the target information; an optimal matching pair is determined by using the cost matrix and preset association pairing information; the multiple sensors are subjected to synchronous target association or asynchronous target association by using the optimal matching pair, so that synchronous tracking or asynchronous tracking of the tracking target is realized. Therefore, the problems in the prior art that target information is lost or deviated due to the environment and hardware limitations of different sensors, and it is difficult to realize multi-source heterogeneous sensor target association are solved.

Description

Technical Field

[0001] This application relates to the field of autonomous driving perception target technology, and in particular to a method, device, vehicle, and storage medium for multi-target association in a vehicle. Background Technology

[0002] With the advent of the intelligent era, technologies across all industries are flourishing. Consumers' hopes and demands for intelligent tea products are also increasing. However, for the automotive industry, intelligent technology is not only about meeting customer needs but also an effective way to enhance brand image. Therefore, autonomous driving, under the backdrop of the new era and new technologies, is gradually becoming a new social trend. Given the limitations of underlying technology, structured roads are the main direction in the field of autonomous driving, but two major technical challenges remain: 1. Target association from multiple sensor sources; 2. Target attribute fusion from multiple sensor sources. The complex real-world environment dictates that a single sensor is insufficient for display and representation. Targets from multiple heterogeneous sensor sources become essential information for autonomous driving, playing a crucial role in vehicle driving control. However, due to limitations imposed by the environment and hardware conditions, different sensors will experience target falsehoods, loss, and fluctuations and errors in key information such as position and speed. Summary of the Invention

[0003] This application provides a method, apparatus, vehicle, and storage medium for multi-target association in vehicles, in order to solve the problems in related technologies, such as the loss or deviation of target information acquired by different sensors due to environmental and hardware limitations, making it difficult to achieve target association of multi-source heterogeneous sensors.

[0004] The first aspect of this application provides a method for multi-target association of a vehicle, comprising the following steps: acquiring road information around the vehicle and tracking information of a target from multiple sensors; extracting target information of the target from the road information acquired by each sensor, and constructing a cost matrix based on the tracking information and / or the target information; determining an optimal matching pair using the cost matrix and preset association pairing information, and using the optimal matching pair to perform synchronous target association or asynchronous target association on the multiple sensors to achieve synchronous or asynchronous tracking of the target.

[0005] Based on the above technical means, the embodiments of this application can construct a cost matrix by using the tracking information and / or target information of the vehicle tracking target, and use the cost matrix to find the optimal path to solve the target association problem. This can not only achieve synchronous target matching and tracking, but also asynchronous target matching and tracking, and can more realistically restore the real scene.

[0006] Optionally, determining the optimal matching pair using the cost matrix and preset association pairing information includes: calculating the cost matrix using the Hungarian algorithm to determine the matching pair corresponding to the minimum matching cost, wherein the optimal matching pair is the matching pair corresponding to the minimum matching cost.

[0007] Based on the above technical means, the embodiments of this application can use the Hungarian algorithm to calculate the cost matrix, find the optimal matching pair, and thus achieve target association.

[0008] Optionally, the step of using the optimal matching pair to perform synchronous target association or asynchronous target association on the multiple sensors includes: forming a tracking list based on matching the tracking target with the observed target of each sensor, using the tracking list to link all target information of the tracking target to achieve synchronous target association of the multiple sensors; and associating the tracking target with the observed target of each sensor to achieve asynchronous target association of the multiple sensors.

[0009] Based on the above technical means, the embodiments of this application, for synchronous target association, can form a tracking list based on the matching of tracking target information and observation target, linking target context information, increasing the dynamic attributes of single sensor targets, and improving the reliability of targets; for asynchronous target association, it can associate tracking targets with observation targets of each sensor, improving the stability of real-time scene perception targets and reducing the resource consumption of environmental cognition.

[0010] Optionally, constructing the cost matrix based on the tracking information and / or the target information includes: if synchronous target association is performed on the multiple sensors, then constructing the cost matrix based on the tracking information and the target information; if asynchronous target association is performed on the multiple sensors, then constructing the cost matrix based on the tracking information or the target information.

[0011] Based on the above technical means, the embodiments of this application construct a cost matrix using different information for synchronous target association and asynchronous target association, and perform preprocessing for calculating the optimal association matching.

[0012] Optionally, if the multiple sensors are synchronously associated with a target, then constructing a cost matrix based on the tracking information and the target information includes: constructing a cost matrix for each sensor based on the tracking information and the target information, and using the cost matrix of each sensor to achieve single-sensor target association; after single-sensor target association, fusing the target information of each sensor to obtain fused information, and constructing a cost matrix based on the fused information and the target information of each sensor.

[0013] Based on the above technical means, in the embodiments of this application, when performing synchronous target association, a cost matrix can be constructed based on the target information of a single sensor and the fused target information.

[0014] Optionally, if asynchronous target association is performed on the multiple sensors, a cost matrix is ​​constructed based on the tracking information or the target information, including: constructing a cost matrix based on the tracking information of each sensor or the fusion information of the target information of each sensor.

[0015] Based on the above technical means, in the embodiments of this application, when performing asynchronous target association, a cost matrix can be constructed based on single sensor tracking information or multi-sensor fusion information.

[0016] Optionally, before constructing the cost matrix based on the tracking information and / or the target information, the method includes: acquiring sensor measurement features and sensor tracking features for each sensor; performing time compensation and spatial synchronization on the target information based on the sensor measurement features; and performing time compensation and spatial synchronization on the tracking information based on the sensor tracking features.

[0017] Based on the above technical means, before constructing the cost matrix, this application embodiment needs to perform time compensation and spatial synchronization on the target information and tracking information according to the measurement characteristics and tracking characteristics of the sensor, so as to accurately obtain the target information and tracking information and ensure the reliability of the subsequent construction of the cost matrix.

[0018] A second aspect of this application provides a multi-target association device for a vehicle, comprising: an acquisition module for acquiring road information around the vehicle and tracking information of a target collected by multiple sensors; a construction module for extracting target information of the target from the road information collected by each sensor, and constructing a cost matrix based on the tracking information and / or the target information; and a matching module for determining an optimal matching pair using the cost matrix and preset association pairing information, and using the optimal matching pair to perform synchronous target association or asynchronous target association on the multiple sensors to achieve synchronous or asynchronous tracking of the target.

[0019] A third aspect of this application provides a vehicle, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the multi-target association method for the vehicle as described in the above embodiments.

[0020] A fourth aspect of this application provides a computer-readable storage medium having a computer program stored thereon, which is executed by a processor to implement the multi-target association method for vehicles as described in the above embodiments.

[0021] Therefore, this application has at least the following beneficial effects:

[0022] (1) The embodiments of this application can construct a cost matrix by tracking information and / or target information of vehicle tracking targets, and use the cost matrix to find the optimal path to solve the target association problem. It can not only realize synchronous target matching and tracking, but also asynchronous target matching and tracking, and can more realistically restore the real scene.

[0023] (2) The embodiments of this application can use the Hungarian algorithm to calculate the cost matrix and find the optimal matching pair, thereby achieving target association.

[0024] (3) In the embodiments of this application, for synchronous target association, a tracking chain can be formed based on the matching of tracking target information and observation target, and the target context information is linked together, which increases the dynamic attributes of single sensor target and improves the reliability of target; for asynchronous target association, the tracking target can be associated with the observation target of each sensor, which improves the stability of real-time scene perception target and reduces the resource consumption of environmental cognition.

[0025] (4) In this embodiment of the application, different information is used to construct a cost matrix for synchronous target association and asynchronous target association, so as to preprocess the calculation of the optimal association matching.

[0026] (5) In the embodiments of this application, when performing synchronous target association, a cost matrix can be constructed based on the target information of a single sensor and the fused target information.

[0027] (6) In the embodiments of this application, when performing asynchronous target association, a cost matrix can be constructed based on single sensor tracking information or multi-sensor fusion information.

[0028] (7) In this embodiment of the application, before constructing the cost matrix, the target information and tracking information can be time-compensated and spatially synchronized according to the measurement characteristics and tracking characteristics of the sensor, so as to accurately obtain the target information and tracking information and ensure the reliability of the subsequent construction of the cost matrix.

[0029] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0030] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0031] Figure 1 This is a flowchart of a multi-target association method for vehicles provided according to an embodiment of this application;

[0032] Figure 2This is a flowchart of asynchronous target association matching based on an embodiment of this application;

[0033] Figure 3 This is a flowchart of the heterogeneous sensor synchronization target association matching process provided according to an embodiment of this application;

[0034] Figure 4 This is an example diagram of a multi-target association device for a vehicle provided according to an embodiment of this application;

[0035] Figure 5 This is a structural schematic diagram of a vehicle provided according to an embodiment of this application. Detailed Implementation

[0036] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0037] The following description, with reference to the accompanying drawings, outlines a method, apparatus, vehicle, and storage medium for multi-target association of vehicles according to embodiments of this application. Addressing the limitations of underlying technology mentioned in the background section, structured roads are a major direction in the field of autonomous driving, but problems still exist regarding multi-source sensor target association and multi-source sensor target attribute fusion. This application provides a method for multi-target association of vehicles, which employs the idea of ​​finding the optimal path using a cost matrix to solve the target association problem. This resolves the problems in related technologies where the loss or deviation of target information acquired by different sensors is due to environmental and hardware limitations, making it difficult to achieve multi-source heterogeneous sensor target association.

[0038] Specifically, Figure 1 This is a flowchart illustrating a multi-target association method for vehicles provided in an embodiment of this application.

[0039] like Figure 1 As shown, the multi-target association method for this vehicle includes the following steps:

[0040] In step S101, multiple sensors acquire road information around the vehicle and tracking information of the target being tracked.

[0041] Among them, multiple sensors may include, but are not limited to, forward-looking cameras, panoramic cameras, millimeter-wave radar, and lidar; road information includes basic information such as lanes, lane lines, vehicle targets, pedestrian targets, road traffic signs, traffic flow, traffic lights, and intersection instructions. For example, cameras use deep learning to detect and extract targets from lane images, while millimeter-wave and lidar use point cloud data clustering to obtain the number of lane targets; tracking information may include basic information such as the ID, type, movement trend, and speed of the tracked target.

[0042] In step S102, target information of the tracked target is extracted from the road information collected by each sensor, and a cost matrix is ​​constructed based on the tracking information and / or target information.

[0043] The target information includes basic information such as target tracking ID, position, velocity, acceleration, heading angle, and motion state; the cost matrix includes the Euclidean distance between the attribute of the feature target and the target to be associated. The feature targets include single-sensor tracked targets and multi-sensor fusion targets.

[0044] In this embodiment of the application, before constructing the cost matrix based on tracking information and / or target information, the process includes: acquiring sensor measurement features and sensor tracking features of each sensor; performing time compensation and spatial synchronization on the target information based on the sensor measurement features; and performing time compensation and spatial synchronization on the tracking information based on the sensor tracking features.

[0045] It should be noted that the embodiments of this application can perform time compensation and spatial synchronization based on the measurement and tracking characteristics of the sensor. Since there is an algorithm delay between the sensor's perception of the environment and its output to the target, it is necessary to compensate for the time difference based on the delay characteristics, and also to perform spatial transformation synchronization of the delay time.

[0046] In this embodiment of the application, constructing a cost matrix based on tracking information and / or target information includes: if multiple sensors are synchronously associated with a target, then a cost matrix is ​​constructed based on tracking information and target information; if multiple sensors are asynchronously associated with a target, then a cost matrix is ​​constructed based on tracking information or target information.

[0047] It is understandable that when multiple sensors are synchronously associated with a target, a cost matrix is ​​constructed based on tracking information and target information; when multiple sensors are asynchronously associated with a target, a cost matrix is ​​constructed based on tracking information or target information.

[0048] In this embodiment of the application, if multiple sensors are synchronously associated with a target, a cost matrix is ​​constructed based on tracking information and target information, including: constructing a cost matrix for each sensor based on tracking information and target information, and using the cost matrix of each sensor to achieve single-sensor target association; after single-sensor target association, the target information of each sensor is fused to obtain fused information, and a cost matrix is ​​constructed based on the fused information and the target information of each sensor.

[0049] It is understandable that if multiple sensors are used for synchronous target association, firstly, a cost matrix for each sensor is constructed based on tracking information and target information to achieve single-sensor target association. Then, the target information of each sensor is fused to obtain fused information. Finally, a cost matrix is ​​constructed based on the fused information, target information, and other attributes.

[0050] In this embodiment of the application, if multiple sensors are asynchronously associated with a target, a cost matrix is ​​constructed based on tracking information or target information, including: constructing a cost matrix based on the fusion information of tracking information of each sensor or target information of each sensor.

[0051] It is understandable that if multiple sensors are asynchronously associated with a target, a cost matrix can be constructed based on single-sensor tracking information or multi-sensor fusion information.

[0052] In step S103, the optimal matching pair is determined using the cost matrix and preset association pairing information. The optimal matching pair is then used to perform synchronous or asynchronous target association on multiple sensors to achieve synchronous or asynchronous tracking of the target.

[0053] Among them, the preset association pairing information refers to the information of the target association that has been completed.

[0054] It is understood that the embodiments of this application can combine the cost matrix and preset association pairing information to obtain the optimal matching pair and achieve the target association.

[0055] In this embodiment of the application, the optimal matching pair is determined using the cost matrix and preset association pairing information, including: using the Hungarian algorithm to calculate the cost matrix, determining the matching pair corresponding to the minimum matching cost, and the optimal matching pair is the matching pair corresponding to the minimum matching cost.

[0056] It is understood that the embodiments of this application may use the Hungarian algorithm to find the optimal perfect match, that is, to use the Hungarian algorithm to calculate the minimum matching cost on the cost matrix and find the optimal matching pair.

[0057] In this embodiment of the application, the optimal matching pair is used to perform synchronous target association or asynchronous target association for multiple sensors, including: forming a tracking list based on matching the tracking target with the observed target of each sensor, using the tracking list to link all target information of the tracking target to achieve synchronous target association of multiple sensors; and associating the tracking target with the observed target of each sensor to achieve asynchronous target association of multiple sensors.

[0058] It is understood that the embodiments of this application, for synchronous target association of multiple sensors, can form a tracking list based on the matching of tracking target information and observed targets, and link target context information, which not only increases the dynamic attributes of single-sensor targets, but also improves the reliability of targets; for asynchronous target association of multiple sensors, it can associate the tracking target with the observed targets of each sensor, improve the stability of real-time scene perception targets, and reduce the resource consumption of environmental cognition.

[0059] Specifically, a multi-objective association system for vehicles proposed in this application includes: a road information acquisition unit, a road information caching unit, a cost matrix construction unit, an optimal matching pair calculation unit, and an association rationality evaluation unit.

[0060] The system includes: a road information acquisition unit for receiving road environment information reported by various sensors; a road information caching unit for caching road data perceived by various sensors; a cost matrix construction unit for constructing a target cost matrix using Euclidean distance with feature target attributes as the parameter set and the target attributes to be associated as the parameter set; an optimal matching pair calculation unit for finding the optimal match of the cost matrix using the Hungarian algorithm to achieve pairing of the observed target and the tracked target; and an association rationality evaluation unit for calculating the rationality of the current matching pair by combining historical context information.

[0061] The road information includes basic information such as lanes, lane lines, vehicle targets, pedestrian targets, road traffic signs, traffic flow, traffic lights, and intersection signs; the target information includes basic information such as target tracking ID, position, speed, acceleration, heading angle, and motion state; the cost matrix includes the Euclidean distance between the attribute of the feature target and the target to be associated; the feature targets include single-sensor tracked targets and multi-sensor fusion targets.

[0062] Based on the multi-target association method for vehicles described in the above embodiments, it can be understood that the implementation of the multi-target association method for vehicles mainly includes the following steps:

[0063] Step 1: Obtain road information using sensors such as forward-looking cameras, panoramic cameras, millimeter-wave radar, and lidar;

[0064] Step 2: Receive road environment information reported by different sensors: target data, lane line data, road sign data, etc., and cache them to form historical perception data;

[0065] Step 3: Combine single-sensor tracking information or multi-sensor fusion data with the sensing data to be associated to construct a cost matrix, which is a preprocessing step for calculating the optimal association matching;

[0066] Step 4: Combining the cost matrix and historical completed pairing information, redundant information in the cost matrix is ​​removed. For the remaining targets that have not yet been associated, the Hungarian algorithm is used to find the minimum cost matching pair.

[0067] Step 5: Combine historical matching information and the similarity of associated features to evaluate the credibility of the minimum cost matching pair, and output the optimal matching pair that is closest to the real environment based on the confidence level.

[0068] In summary, the embodiments of this application, for the association of targets from the same source sensor, can form a tracking list based on the matching of tracking target information and observed target, and link target context information, which not only increases the dynamic attributes of single sensor targets, but also improves the reliability of targets; for the association of targets from multiple source sensors, it can associate heterogeneous tracking targets with the observed targets of each sensor, improve the stability of real-time scene perception targets, and reduce the resource consumption of environmental cognition.

[0069] The following sections will elaborate on the multi-target association methods for vehicles, specifically focusing on asynchronous target association matching from the same-source sensor and synchronous target association matching from heterogeneous sensors.

[0070] like Figure 2 As shown, an asynchronous target association and matching method using sensors from the same source includes the following steps:

[0071] Step 1: Acquire road target information through sensing devices. Road target information includes target tracking ID, position, speed, acceleration in the vehicle coordinate system, target type, motion trend, heading angle, etc. Information can be acquired from sensors such as cameras, millimeter-wave radar, and lidar.

[0072] Step 2: Specify preprocessing data cleaning rules based on the characteristics of different sensors and prior knowledge, and filter out invalid and obviously abnormal targets;

[0073] Step 3: Perform time compensation and spatial synchronization based on the sensor's measurement characteristics. Since there is an algorithm delay between the sensor's perception of the environment and its output to the target, it is necessary to compensate for the time difference based on the delay characteristics, and also to spatially transform and synchronize the delay time.

[0074] Step 4: Perform time compensation and spatial synchronization based on the sensor tracking characteristics. Since there is a time difference between the tracked target and the current target, time compensation and spatial synchronization are necessary based on these characteristics.

[0075] Step 5: Combine single-sensor tracking information or multi-sensor fusion data with the sensing data to be associated to construct a cost matrix, which is a preprocessing step for calculating the optimal association matching;

[0076] Step 7: Use the Hungarian algorithm (or a method not limited to this) to find the optimal perfect match. Calculate the minimum matching cost using the Hungarian algorithm on the cost matrix to find the optimal matching pair;

[0077] Step 8: Evaluate the optimal perfect match by combining tracking information. Utilize the feature information extracted during tracking to evaluate the confidence level of the current matching pair, thereby enhancing the reliability of the output target.

[0078] like Figure 3 As shown, a method for synchronous target association matching of heterogeneous sensors includes the following steps:

[0079] Step 1: Acquisition of target information from a single sensor:

[0080] (1) Different sensor devices use different methods to acquire lane target information. For example, cameras use deep learning to detect and extract targets from lane images; millimeter-wave and lidar use point cloud data clustering to obtain lane target data;

[0081] (2) Based on the characteristics of different sensors and prior knowledge, specify preprocessing data cleaning rules to remove obvious abnormal data. For example, the continuity of target position and velocity, and the continuity of motion state, etc.

[0082] (3) Perform time compensation and spatial synchronization of the target based on sensor characteristics.

[0083] Step 2, Single sensor target association matching:

[0084] (1) Construct a cost matrix using measurement target information and tracking target information. Construct a cost matrix using the location information of the measurement target and the tracking target, as well as other attributes;

[0085] (2) The Hungarian algorithm is used to find the optimal perfect match of the cost matrix.

[0086] Step 3, Single Sensor Matching Evaluation: Combine single sensor features and tracking information to evaluate the optimal perfect match pair, thereby enhancing the reliability of the target.

[0087] Step 4: Multi-sensor target association and matching:

[0088] (1) Construct a cost matrix using single-sensor target information and fused target information. Construct the cost matrix using single-sensor target and fused target location information, as well as other attributes;

[0089] (2) The Hungarian algorithm is used to find the optimal perfect match of the cost matrix.

[0090] Step 5: Target matching evaluation: Evaluate the optimal perfect match pair by combining fusion characteristics to enhance the reliability of the fusion target output.

[0091] The multi-target association method for vehicles proposed in this application can construct a cost matrix using the tracking information and / or target information of the vehicle tracking target. The target association problem is solved by using the cost matrix to find the optimal path. This method can achieve both synchronous and asynchronous target matching and tracking, enabling a more realistic reconstruction of real-world scenarios. The Hungarian algorithm can be used to calculate the cost matrix and find the optimal matching pair, thereby achieving target association. For synchronous target association, a tracking chain can be formed based on the matching of the tracking target information and the observed target, linking the target context information, increasing the dynamic attributes of the single-sensor target, and improving the reliability of the target. For asynchronous target association, it enables the tracking target to be matched with the observations of each sensor. Target association improves the stability of real-time scene perception targets and reduces the resource consumption of environmental cognition. For synchronous and asynchronous target association, different information is used to construct cost matrices, which serves as preprocessing for calculating the optimal association match. When performing synchronous target association, the cost matrix can be constructed based on target information from a single sensor and fused target information. When performing asynchronous target association, the cost matrix can be constructed based on tracking information from a single sensor or fused information from multiple sensors. Before constructing the cost matrix, time compensation and spatial synchronization can be performed on target information and tracking information based on the measurement and tracking characteristics of the sensors, respectively, to accurately obtain target information and tracking information and ensure the reliability of the subsequent construction of the cost matrix.

[0092] Next, the multi-target association device for vehicles according to embodiments of this application is described with reference to the accompanying drawings.

[0093] Figure 4 This is a block diagram of a vehicle multi-target association device according to an embodiment of this application.

[0094] like Figure 4 As shown, the multi-target association device 10 of the vehicle includes: an acquisition module 100, a construction module 200, and a matching module 300.

[0095] The acquisition module 100 is used to acquire road information around the vehicle and tracking information of the target collected by multiple sensors; the construction module 200 is used to extract target information of the target collected by each sensor from the road information, and construct a cost matrix based on the tracking information and / or target information; the matching module 300 is used to determine the optimal matching pair using the cost matrix and preset association pairing information, and use the optimal matching pair to perform synchronous target association or asynchronous target association on multiple sensors to achieve synchronous or asynchronous tracking of the target.

[0096] It should be noted that the foregoing explanation of the multi-target association method for vehicles also applies to the multi-target association device for vehicles in this embodiment, and will not be repeated here.

[0097] The multi-target association device for vehicles proposed in this application can construct a cost matrix using the tracking information and / or target information of the vehicle tracking target. The target association problem is solved by using the cost matrix to find the optimal path. This method can achieve both synchronous and asynchronous target matching and tracking, enabling a more realistic reconstruction of real-world scenarios. The Hungarian algorithm can be used to calculate the cost matrix and find the optimal matching pair, thereby achieving target association. For synchronous target association, a tracking chain can be formed based on the matching of the tracking target information and the observed target, linking the target context information, increasing the dynamic attributes of the single-sensor target, and improving the reliability of the target. For asynchronous target association, it enables the tracking target to be matched with the observations of each sensor. Target association improves the stability of real-time scene perception targets and reduces the resource consumption of environmental cognition. For synchronous and asynchronous target association, different information is used to construct cost matrices, which serves as preprocessing for calculating the optimal association match. When performing synchronous target association, the cost matrix can be constructed based on target information from a single sensor and fused target information. When performing asynchronous target association, the cost matrix can be constructed based on tracking information from a single sensor or fused information from multiple sensors. Before constructing the cost matrix, time compensation and spatial synchronization can be performed on target information and tracking information based on the measurement and tracking characteristics of the sensors, respectively, to accurately obtain target information and tracking information and ensure the reliability of the subsequent construction of the cost matrix.

[0098] Figure 5 A schematic diagram of the structure of a vehicle provided in an embodiment of this application. The vehicle may include:

[0099] The memory 501, the processor 502, and the computer program stored on the memory 501 and capable of running on the processor 502.

[0100] When processor 502 executes the program, it implements the multi-target association method for vehicles provided in the above embodiments.

[0101] Furthermore, the vehicle also includes:

[0102] Communication interface 503 is used for communication between memory 501 and processor 502.

[0103] The memory 501 is used to store computer programs that can run on the processor 502.

[0104] The memory 501 may include high-speed RAM (Random Access Memory) memory, and may also include non-volatile memory, such as at least one disk storage.

[0105] If the memory 501, processor 502, and communication interface 503 are implemented independently, then the communication interface 503, memory 501, and processor 502 can be interconnected via a bus to complete communication between them. The bus can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 5 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0106] Optionally, in a specific implementation, if the memory 501, processor 502, and communication interface 503 are integrated on a single chip, then the memory 501, processor 502, and communication interface 503 can communicate with each other through an internal interface.

[0107] Processor 502 may be a CPU (Central Processing Unit), an ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement embodiments of this application.

[0108] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described multi-target association method for vehicles.

[0109] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0110] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0111] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0112] It should be understood that the various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (FPGAs), field-programmable gate arrays (FPGAs), etc.

[0113] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0114] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A method for multi-target association of vehicles, characterized in that, Includes the following steps: Acquire road information around the vehicle and tracking information of the target from multiple sensors; Extract the target information of the tracked target from the road information collected by each sensor, and construct a cost matrix based on the tracking information and / or the target information; The optimal matching pair is determined using the cost matrix and preset association pairing information. The optimal matching pair is then used to perform synchronous target association or asynchronous target association on the multiple sensors to achieve synchronous or asynchronous tracking of the target. The construction of the cost matrix based on the tracking information and / or the target information includes: If the multiple sensors are synchronized and associated with a target, a cost matrix is ​​constructed based on the tracking information and the target information; If asynchronous target association is performed on the multiple sensors, a cost matrix is ​​constructed based on the tracking information or the target information; If the multiple sensors are synchronized for target association, a cost matrix is ​​constructed based on the tracking information and the target information, including: Based on the tracking information and the target information, a cost matrix for each sensor is constructed, and the cost matrix of each sensor is used to achieve single-sensor target association. After single-sensor target association, the target information of each sensor is fused to obtain fused information, and a cost matrix is ​​constructed based on the fused information and the target information of each sensor.

2. The method according to claim 1, characterized in that, The step of determining the optimal matching pair using the cost matrix and preset association pairing information includes: The cost matrix is ​​calculated using the Hungarian algorithm to determine the matching pair corresponding to the minimum matching cost, wherein the optimal matching pair is the matching pair corresponding to the minimum matching cost.

3. The method according to claim 1, characterized in that, The step of using the optimal matching pair to perform synchronous or asynchronous target association on the multiple sensors includes: A tracking list is formed by matching the tracking target with the observed target of each sensor. All target information of the tracking target is linked together using the tracking list to achieve synchronous target association of the multiple sensors. The tracking target is associated with the observation target of each sensor, thereby realizing asynchronous target association among the multiple sensors.

4. The method according to claim 1, characterized in that, If asynchronous target association is performed on the multiple sensors, a cost matrix is ​​constructed based on the tracking information or the target information, including: A cost matrix is ​​constructed based on the tracking information from each sensor or the fusion information of the target information from each sensor.

5. The method according to claim 1, characterized in that, Before constructing the cost matrix based on the tracking information and / or the target information, the process includes: Obtain the sensor measurement characteristics and sensor tracking characteristics for each sensor; The target information is time-compensated and spatially synchronized based on the sensor measurement characteristics. The tracking information is time-compensated and spatially synchronized based on the sensor tracking characteristics.

6. A multi-target association device for vehicles, characterized in that, A method for implementing the multi-target association of vehicles as described in any one of claims 1-5 includes: The acquisition module is used to acquire road information around the vehicle and tracking information of the target being tracked from multiple sensors; The construction module is used to extract the target information of the tracked target from the road information collected by each sensor, and construct a cost matrix based on the tracking information and / or the target information; The matching module is used to determine the optimal matching pair using the cost matrix and preset association pairing information, and to perform synchronous target association or asynchronous target association on the multiple sensors using the optimal matching pair, so as to realize synchronous tracking or asynchronous tracking of the tracking target.

7. A vehicle, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the program to implement the multi-target association method for vehicles as described in any one of claims 1-5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by the processor to implement the multi-target association method for vehicles as described in any one of claims 1-5.

Images