Dynamic target real-time ranging tracking method and system based on multi-camera collaborative perception

Abstract

The invention discloses a multi-camera collaborative perception dynamic target real-time ranging tracking method and system, and relates to the technical field of target tracking, after a target value of each initial decision variable configuration is calculated based on a target function, all initial decision variable configurations are subjected to non-dominated sorting and are divided into different Pareto levels, and the target value of each initial decision variable configuration is calculated; and performing content randomization operation on all initial decision variable configurations, performing iterative optimization on newly generated decision variable configurations, outputting all Pareto front solution sets when convergence conditions are met, and dynamically selecting the Pareto front solution sets according to scene changes to be applied to corresponding scenes. The tracking system comprehensively considers multiple targets such as ranging precision, tracking robustness and real-time performance, generates a group of non-dominated solutions, can dynamically adapt to different scene requirements, realizes multi-dimensional balance optimization of performance, and remarkably improves the flexibility and adaptability of the system in combination with the specific requirements of real-time scenes.

Description

Technical Field

[0001] The present invention relates to the technical field of target tracking, and particularly to a method and system for real-time ranging and tracking of dynamic targets with multi-camera collaborative perception. Background Art

[0002] In the fields of intelligent transportation, driverless, security monitoring, and industrial automation, real-time ranging and tracking of dynamic targets is an important task for achieving precise perception. Traditional single-camera ranging and target tracking methods are limited by problems such as the field of view, occlusion, and insufficient depth information, and it is difficult to provide reliable perception results in complex dynamic scenarios. To solve these problems, multi-camera collaborative perception systems have emerged.

[0003] The existing technologies have the following deficiencies:

[0004] 1. Existing technologies usually focus on a single optimization goal, such as maximizing ranging accuracy or optimizing real-time performance, while ignoring the mutual constraints between multiple goals. In a system with ranging accuracy as the main focus, excessive pursuit of calculation details may lead to a decrease in system real-time performance and inability to meet the requirements of rapidly changing scenarios of dynamic targets. In a system with real-time performance as the main focus, to reduce latency, the robustness of target tracking is often sacrificed, resulting in unstable performance in occluded or complex scenarios. Single-target optimization cannot adapt to the changing actual scenario requirements, especially in applications where multi-target performance requires dynamic trade-offs (such as intelligent transportation, UAV formations, etc.), and it is difficult for the system to achieve the best performance.

[0005] 2. Existing technologies usually operate based on static configurations and preset parameters, lacking the ability to adjust according to dynamic scene changes. For example, when the number of targets increases or the occlusion situation worsens, the system cannot dynamically adjust the camera field of view and task allocation strategies, resulting in an increase in the tracking loss rate. In complex scenarios (such as intersections with dense vehicles or scenes where moving targets cross at high speeds), traditional methods cannot optimize weights and parameters in real time, and the system performance significantly deteriorates. The system operation is rigid and difficult to adapt to the real-time changing complex environment, resulting in unstable ranging or tracking effects.

[0006] Based on this, the present application proposes a method and system for real-time ranging and tracking of dynamic targets with multi-camera collaborative perception, comprehensively considering multiple goals such as ranging accuracy, tracking robustness, and real-time performance, generating a set of non-dominated solutions, being able to dynamically adapt to different scenario requirements, achieving multi-dimensional balanced optimization of performance, and significantly enhancing the flexibility and adaptability of the system in combination with the specific requirements of real-time scenarios. Summary of the Invention

[0007] The object of the present invention is to provide a method and system for real-time ranging and tracking of dynamic targets with multi-camera collaborative perception to solve the deficiencies in the background art.

[0008] To achieve the above object, the present invention provides the following technical solution: a dynamic target real-time ranging and tracking method for multi-camera collaborative perception, the tracking method comprising the following steps:

[0009] The tracking system generates multiple initial decision variable configurations through initialization. After calculating the objective value of each initial decision variable configuration based on the objective function, non-dominated sorting is performed on all initial decision variable configurations and divided into different Pareto levels;

[0010] After performing content randomization operations on all initial decision variable configurations, the newly generated decision variable configurations are iteratively optimized. When the convergence condition is met, all Pareto front solution sets are output, and the Pareto front solution set is dynamically selected according to the scene change and applied to the corresponding scene.

[0011] In a preferred embodiment, the objective value of each initial decision variable configuration is calculated based on the objective function, and the function expression is: , where is the objective value, is the robustness index, is the response delay index, , are the proportionality coefficients of the robustness index and the response delay index respectively, and , are both greater than 0.

[0012] In a preferred embodiment, non-dominated sorting is performed on all initial decision variable configurations and divided into different Pareto levels, including the following steps:

[0013] After selecting the current initial decision variable configuration, compare it one by one with the current initial decision variable configuration;

[0014] If the current initial decision variable configuration is not worse than the compared initial decision variable configuration in all parameters, and at least one parameter value of the current initial decision variable configuration is better than the compared initial decision variable configuration, then the current initial decision variable configuration dominates the compared initial decision variable configuration;

[0015] If the compared initial decision variable configuration is not worse than the current initial decision variable configuration in all parameters, and at least one parameter value of the compared initial decision variable configuration is better than the current initial decision variable configuration, then the compared initial decision variable configuration dominates the current initial decision variable configuration;

[0016] Otherwise, the current initial decision variable configuration does not dominate the compared initial decision variable configuration, or the compared initial decision variable configuration does not dominate the current initial decision variable configuration;

[0017] If there is no domination relationship among the initial decision variable configurations, it is recorded as a non-dominated solution, that is, there is no other initial decision variable configuration that dominates the current initial decision variable configuration solution;

[0018] Sort according to the domination relationship among the initial decision variable configurations;

[0019] The first Pareto front: contains the initial decision variable configurations that are not dominated by any other initial decision variable configurations;

[0020] The second Pareto front: contains the initial decision variable configurations that are dominated by the initial decision variable configurations of the first Pareto front but not dominated by other initial decision variable configurations.

[0021] In a preferred embodiment, in each Pareto front, calculate the crowding degree of the initial decision variable configuration, and the expression is: , where is the crowding degree of the current initial decision variable configuration, , are the objective values of the adjacent initial decision variable configurations of the current initial decision variable configuration, is the maximum objective value in the Pareto front, is the minimum objective value in the Pareto front;

[0022] In each Pareto front, sort all the initial decision variable configurations from largest to smallest according to the crowding degree.

[0023] In a preferred embodiment, after performing a content randomization operation on all the initial decision variable configurations, perform iterative optimization on the newly generated decision variable configurations. When the convergence condition is met, output all the Pareto front solution sets, including the following steps:

[0024] Based on the initial decision variable configuration, perform a randomization operation to generate a new initial decision variable configuration. The randomization operation includes random perturbation, probability-based change, and mutation operation;

[0025] Re-perform the objective value calculation, non-dominated sorting, and Pareto ranking steps on the initial decision variable configurations that have completed the content randomization operation to generate a new Pareto front;

[0026] When the number of iterations is equal to the number threshold, the change in the objective value is less than the objective value change threshold, or the change in the crowding degree is less than the crowding degree change threshold, it is determined that the convergence condition is met. When the convergence condition is met, output all the Pareto front solution sets.

[0027] In a preferred embodiment, the calculation logic of the robustness index is as follows: obtain the ranging error and tracking failure rate of the initial decision variable configuration, perform normalization processing on the ranging error and tracking failure rate to map the value ranges of the ranging error and tracking failure rate to between [0, 1], obtain the normalized ranging error value and the normalized tracking failure rate value, and calculate and obtain the robustness index. The expression is: , where is the robustness index, is the normalized ranging error value, is the normalized tracking failure rate value;

[0028] The calculation logic of the response delay index is as follows: obtain the algorithm delay and network transmission delay of the initial decision variable configuration, and sum the algorithm delay and network transmission delay to obtain the response delay index.

[0029] In a preferred embodiment, the content of the initial decision variable configuration includes the resolution, frame rate, viewing angle coverage range of each camera, weight allocation for multi-camera depth estimation, and camera task priority allocation.

[0030] The dynamic target real-time ranging and tracking system for multi-camera collaborative perception includes a configuration generation module, a non-dominated sorting module, and a scenario application module:

[0031] Configuration generation module: generate multiple initial decision variable configurations through initialization, and calculate the objective value of each initial decision variable configuration based on the objective function;

[0032] Non-dominated sorting module: perform non-dominated sorting on all initial decision variable configurations and divide them into different Pareto rank blocks;

[0033] Scenario application module: after performing content randomization operations on all initial decision variable configurations, perform iterative optimization on the newly generated decision variable configurations. When the convergence condition is met, output all Pareto front solution sets, and dynamically select a Pareto front solution set according to the scenario change and apply it to the corresponding scenario.

[0034] In the above technical solution, the technical effects and advantages provided by the present invention:

[0035] The present invention generates multiple initial decision variable configurations through initialization. After calculating the objective values of each initial decision variable configuration based on the objective function, non-dominated sorting is performed on all initial decision variable configurations and divided into different Pareto levels. After performing content randomization operations on all initial decision variable configurations, the newly generated decision variable configurations are iteratively optimized. When the convergence condition is met, all Pareto front solution sets are output, and the Pareto front solution sets are dynamically selected according to scene changes and applied to the corresponding scenes. The tracking system comprehensively considers multiple objectives such as ranging accuracy, tracking robustness, and real-time performance, generates a set of non-dominated solutions, can dynamically adapt to different scene requirements, realizes multi-dimensional balanced optimization of performance, and significantly improves the flexibility and adaptability of the system in combination with the specific requirements of the real-time scene. Description of the Drawings

[0036] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required to be used in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0037] Figure 1 It is a flowchart of the method of the present invention. Detailed Embodiments

[0038] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0039] Embodiment 1: Please refer to Figure 1 As shown, the dynamic target real-time ranging and tracking method for multi-camera collaborative perception in this embodiment includes the following steps:

[0040] The tracking system generates multiple initial decision variable configurations through initialization. The content of each initial decision variable configuration includes the resolution, frame rate, viewing angle coverage range of each camera, weight allocation for multi-camera depth estimation, and camera task priority allocation. After calculating the objective value of each initial decision variable configuration based on the objective function, non-dominated sorting is performed on all initial decision variable configurations and divided into different Pareto levels. After performing content randomization operations on all initial decision variable configurations, the newly generated decision variable configurations are iteratively optimized. When the convergence condition is met, all Pareto front solution sets are output, and the Pareto front solution sets are dynamically selected according to scene changes and applied to the corresponding scenes.

[0041] This application generates multiple initial decision variable configurations through initialization. After calculating the objective values of each initial decision variable configuration based on the objective function, non-dominated sorting is performed on all initial decision variable configurations and divided into different Pareto levels. After performing a content randomization operation on all initial decision variable configurations, the newly generated decision variable configurations are iteratively optimized. When the convergence condition is met, all Pareto front solution sets are output, and the Pareto front solution sets are dynamically selected according to the scenario changes and applied to the corresponding scenarios. The tracking system comprehensively considers multiple objectives such as ranging accuracy, tracking robustness, and real-time performance, generates a set of non-dominated solutions, can dynamically adapt to different scenario requirements, realizes multi-dimensional balanced optimization of performance, and significantly improves the flexibility and adaptability of the system in combination with the specific requirements of the real-time scenario.

[0042] Embodiment 2: The tracking system generates multiple initial decision variable configurations through initialization. The content of each initial decision variable configuration includes the resolution, frame rate, view coverage range of each camera, weight assignment for multi-camera depth estimation, and camera task priority assignment, including the following steps:

[0043] Generate diverse configuration combinations for each camera to cover different scenario requirements.

[0044] The decision variables include:

[0045] Resolution: The pixel resolution of the camera (such as 720p, 1080p, 4K).

[0046] Frame rate: The number of frames captured per second (such as 15FPS, 30FPS, 60FPS).

[0047] View coverage range: The horizontal and vertical viewing angles of the camera (such as 90°, 120°).

[0048] Depth estimation weight: The weight of different cameras' contribution to depth information in multi-view fusion (for example, a higher weight can be assigned to high-occlusion areas).

[0049] Task priority: The priority for different camera tasks (such as tracking in occlusion areas or global monitoring).

[0050] Use the parameter range (such as resolution [720p, 1080p, 4K]) for random sampling to generate the initial population. Improve the population diversity through orthogonal experimental design or Latin square design.

[0051] Assume the parameter range of the multi-camera system is as follows:

[0052] Resolution: [720p, 1080p, 4K];

[0053] Frame rate: [15FPS, 30FPS, 60FPS];

[0054] Viewing angle coverage: [90°, 120°, 150°];

[0055] Depth estimation weight: [0.1, 0.3, 0.5, 0.7, 0.9];

[0056] Task priority: [1, 2, 3, 4] (1 represents the highest priority and 4 is the lowest);

[0057] Method 1: Random sampling to generate the initial population

[0058] Randomly select values within the parameter range for combination to generate diverse configurations. For example:

[0059] Configuration of Camera A: Resolution: 1080p, Frame rate: 30FPS, Viewing angle coverage: 120°, Depth estimation weight: 0.7, Task priority: 2;

[0060] Configuration of Camera B: Resolution: 720p, Frame rate: 15FPS, Viewing angle coverage: 150°, Depth estimation weight: 0.5, Task priority: 1;

[0061] Configuration of Camera C: Resolution: 4K, Frame rate: 60FPS, Viewing angle coverage: 90°, Depth estimation weight: 0.9, Task priority: 3.

[0062] The generated population is: [A: 1080p, 30FPS, 120°, 0.7, 2], [B: 720p, 15FPS, 150°, 0.5, 1], [C: 4K, 60FPS, 90°, 0.9, 3].

[0063] Use the orthogonal array to evenly distribute the combinations of parameters to enhance the representativeness of the configuration combinations. Assume each parameter has 3 levels:

[0064] Resolution: 720p, 1080p, 4K;

[0065] Frame rate: 15FPS, 30FPS, 60FPS;

[0066] Viewing angle coverage: 90°, 120°, 150°;

[0067] The orthogonal array is shown in Table 1:

[0068] Allocation:

[0069] Group number Resolution Frame rate Viewing angle coverage 1 720p 15FPS 90° 2 720p 30FPS 120° 3 720p 60FPS 150° 4 1080p 15FPS 120° 5 1080p 30FPS 150° 6 1080p 60FPS 90° 7 4K 15FPS 150° 8 4K 30FPS 90° 9 4K 60FPS 120°

[0070] Table 1

[0071] The above configuration is further expanded by combining depth estimation weights and task priorities. By increasing the depth estimation weights and task priorities: Select 3 values from the depth estimation weights [0.1, 0.3, 0.5, 0.7, 0.9] for distribution (such as 0.3, 0.7, 0.9). Allocate from the task priorities [1, 2, 3]. For example, expand the first group of configurations in the orthogonal table to:

[0072] Configuration 1: Resolution: 720p, Frame rate: 15FPS, View coverage: 90°, Depth estimation weight: 0.3, Task priority: 1;

[0073] Configuration 2: Resolution: 720p, Frame rate: 15FPS, View coverage: 90°, Depth estimation weight: 0.7, Task priority: 2;

[0074] Configuration 3: Resolution: 720p, Frame rate: 15FPS, View coverage: 90°, Depth estimation weight: 0.9, Task priority: 3.

[0075] Through the above method, the following population can be generated:

[0076] [1:720p,15FPS,90°,0.3,1],[2:1080p,30FPS,120°,0.7,2],[3:4K,60FPS,150°,0.9,3],[4:720p,30FPS,120°,0.5,2],[5:1080p,15FPS,150°,0.7,1]. This population not only includes the diversity of parameter ranges but also takes into account the balanced distribution of experimental designs, providing a good foundation for subsequent optimization and scenario adaptation.

[0077] Perform non - dominated sorting on all initial decision variable configurations and divide them into different Pareto levels, including the following steps:

[0078] Calculate the objective value of each initial decision variable configuration based on the objective function. The function expression is:

[0079] , where, is the objective value, is the robustness index, is the response delay index, 、 are the proportionality coefficients of the robustness index and the response delay index respectively, and 、 are both greater than 0.

[0080] The calculation logic of the robustness index is as follows: Obtain the ranging error and tracking failure rate of the initial decision variable configuration, normalize the ranging error and tracking failure rate so that the value ranges of the ranging error and tracking failure rate are mapped to between [0, 1], obtain the normalized value of the ranging error and the normalized value of the tracking failure rate, and calculate the robustness index. The expression is: , where is the robustness index, is the normalized value of the ranging error, is the normalized value of the tracking failure rate. The larger the robustness index, the smaller the ranging error and tracking failure rate of the initial decision variable configuration, and the better the robustness;

[0081] The calculation expression of the ranging error is: , where is the ranging error, is the number of times the object is measured, the distance of the object measured at the the actual object distance at the the time. The larger the ranging error value, the lower the ranging accuracy.

[0082] The calculation logic of the response delay index is as follows: Obtain the algorithm delay and network transmission delay of the initial decision variable configuration, sum the algorithm delay and network transmission delay to obtain the response delay index. The larger the response delay index, the worse the overall response of the initial decision variable configuration.

[0083] Perform non-dominated sorting on all initial decision variable configurations and divide them into different Pareto levels, including the following steps:

[0084] After selecting the current initial decision variable configuration, compare it one by one with the current initial decision variable configuration;

[0085] If the current initial decision variable configuration is not worse than the compared initial decision variable configuration in all parameters, and the current initial decision variable configuration has at least one parameter value better than the compared initial decision variable configuration, then the current initial decision variable configuration dominates the compared initial decision variable configuration;

[0086] For example: Configuration A dominates Configuration B if, for all parameters in all objective functions, all parameter values in the objective function of A are no worse than those of B (e.g., the robustness index of A is greater than or equal to that of B, and the response delay index of A is less than or equal to that of B), and A is superior to B in all parameters of at least one objective function (e.g., the robustness index of A is greater than that of B, or the response delay index of A is less than that of B). In other words, Configuration A performs better than Configuration B in all parameters of the objective function, or is at least equal in all parameters of some objective functions. Then A dominates B.

[0087] If the comparison initial decision variable configuration is no worse than the current initial decision variable configuration in all parameters, and at least one parameter value of the comparison initial decision variable configuration is superior to the current initial decision variable configuration, then the comparison initial decision variable configuration dominates the current initial decision variable configuration;

[0088] Otherwise, the current initial decision variable configuration does not dominate the comparison initial decision variable configuration, or the comparison initial decision variable configuration does not dominate the current initial decision variable configuration;

[0089] If there is no dominance relationship between the initial decision variable configurations, it is recorded as a non-dominated solution, that is, a solution where no other initial decision variable configuration dominates the current initial decision variable configuration;

[0090] Sort according to the dominance relationship between the initial decision variable configurations;

[0091] The first Pareto front: contains the initial decision variable configurations that are not dominated by any other initial decision variable configurations;

[0092] The second Pareto front: contains the initial decision variable configurations that are dominated by the initial decision variable configurations in the first Pareto front but are not dominated by other initial decision variable configurations, and so on, to obtain all Pareto fronts.

[0093] In each Pareto front, calculate the crowding degree of the initial decision variable configuration, and the expression is:

[0094] , where is the crowding degree of the current initial decision variable configuration, 、 are the objective values of the initial decision variable configurations adjacent to the current initial decision variable configuration (i.e., before and after the current initial decision variable configuration), is the maximum objective value in the Pareto front, is the minimum objective value in the Pareto front;

[0095] In each Pareto front, all initial decision variable configurations are sorted in descending order of crowding distance. The larger the crowding distance of an initial decision variable configuration, the sparser it is in the objective space, and it may be more optimal.

[0096] After performing content randomization operations on all initial decision variable configurations, the newly generated decision variable configurations are iteratively optimized. When the convergence condition is met, all Pareto front solution sets are output, and the Pareto front solution sets are dynamically selected according to the scene changes and applied to the corresponding scenes, including the following steps:

[0097] Based on the initial decision variable configurations, randomization operations are performed to generate new initial decision variable configurations. The randomization operations aim to increase the diversity of the population and avoid falling into local optimal initial decision variable configurations. This process can be achieved through the following methods:

[0098] Random perturbation: A small random perturbation is applied to each decision variable. For example, minor adjustments are made to parameters such as resolution, frame rate, and depth estimation weights.

[0099] Probability-based change: Decision variables are randomly selected and updated according to a certain probability to avoid making the same changes each time.

[0100] Mutation operation: Using the mutation method in genetic algorithms, some decision variables are selected for random changes to generate new candidate initial decision variable configurations.

[0101] A certain proportion of the initial decision variable configurations are randomly selected for perturbation. The perturbed initial decision variable configurations are repaired (if there are constraint conditions) to ensure that the generated initial decision variable configurations are valid.

[0102] Recalculate the objective values, perform non-dominated sorting, and conduct Pareto ranking steps on the initial decision variable configurations that have completed the content randomization operation to generate new Pareto fronts;

[0103] When the number of iterations is equal to the iteration threshold, the change in the objective value is less than the objective value change threshold, or the change in the crowding distance is less than the crowding distance change threshold, it is determined that the convergence condition is met. When the convergence condition is met, all Pareto front solution sets are output;

[0104] Since the multi-camera collaborative perception dynamic target real-time ranging and tracking system needs to be adjusted according to the changes in the actual scene, it is first necessary to detect the scene changes and classify them. Different scenes may have different requirements for the system. For example, higher resolution, faster frame rate, or more accurate depth estimation may be required.

[0105] Scene Monitoring: Through the system's sensors, data acquisition modules, or external environment perception, the scene changes are monitored in real time. For example, factors such as the lighting conditions in the scene, the movement speed of the target, and the occlusion situation.

[0106] Scene Classification: According to the detected scene information, the scenes are classified into different types. For example:

[0107] Static Scenes: The targets change slowly and the background is relatively fixed.

[0108] Dynamic Scenes: The targets change rapidly and the background may also change significantly.

[0109] High Occlusion Scenes: The targets are partially or completely occluded.

[0110] High Precision Requirement Scenes: Higher precision ranging and target tracking are required.

[0111] According to the changes in the scene, it is necessary to dynamically select the appropriate Pareto front solution set to meet different requirements. Different scenes have different requirements for system performance, so it is necessary to select the solution that best suits the current scene from the output Pareto front solution set.

[0112] Scene Requirement Mapping: Define a set of performance requirements for each scene type. For example:

[0113] Static Scenes: May tend to have low power consumption and low latency configurations.

[0114] Dynamic Scenes: Require high frame rate and high resolution configurations.

[0115] High Precision Scenes: Need to prioritize the accuracy of depth estimation and low ranging error.

[0116] According to the scene requirements, select the most suitable solution from the current Pareto front solution set. Usually, the solution with the performance most in line with the scene requirements is selected, but other solutions can also be considered as alternative options to quickly adjust when the scene changes.

[0117] If the current scene is a static environment, the system can select a solution with a lower resolution but a smaller response delay to save computing resources. If it is a dynamic scene, the system will select a solution with a higher resolution and a higher frame rate to ensure tracking accuracy and real-time performance. For high occlusion scenes, the system will preferentially select those solutions with a higher weight for depth estimation to improve the target recognition and ranging accuracy in the case of occlusion.

[0118] Based on the selected Pareto front solution set, the system adjusts its various parameter configurations (such as camera resolution, frame rate, depth estimation weight, etc.) to adapt to the requirements of the current scenario. Adjust parameters such as camera resolution, frame rate, viewing angle, and task priority to ensure that the system can effectively adapt to the current scenario. For example, in high-precision scenarios, the system can increase the camera resolution and depth estimation weight, and in low-latency scenarios, the system can reduce the resolution and increase the frame rate. Dynamically optimize the allocation of system resources according to the selected solution set. For example, in low-resource scenarios, the system may reduce the number of tasks processed in parallel, while in high-load scenarios, it may increase the parallelism and computing power. Continuously adjust parameters based on the real-time feedback of the scenario to ensure the stability and performance of the system.

[0119] When the scenario changes, the system should be able to quickly adapt and update the selected solution set. For example, when the scenario changes from a static environment to a dynamic environment, the system should re-select a solution set suitable for dynamic target tracking and perform parameter adjustment.

[0120] Monitor the dynamic changes of the scenario, such as target speed, occlusion situation, etc. Re-select the Pareto front solution set that is most suitable for the current scenario according to the new scenario information. Perform online optimization or re-iteration according to the scenario change so that the system can maintain the best performance in the new scenario. According to the scenario requirements, the finally selected solution set will be applied to the specific system settings for real-time dynamic target tracking, ranging, and feedback. The system will continuously monitor the scenario to ensure that the optimal performance configuration is maintained throughout the operation process.

[0121] Embodiment 3: The dynamic target real-time ranging and tracking system for multi-camera collaborative perception described in this embodiment includes a configuration generation module, a non-dominated sorting module, and a scenario application module:

[0122] Configuration generation module: Generate multiple initial decision variable configurations through initialization. The content of each initial decision variable configuration includes the resolution, frame rate, viewing angle coverage range of each camera, weight allocation for multi-camera depth estimation, and camera task priority allocation. After calculating the objective value of each initial decision variable configuration based on the objective function, send the initial decision variable configuration and the objective value to the non-dominated sorting module;

[0123] Non-dominated sorting module: Perform non-dominated sorting on all initial decision variable configurations and divide them into different Pareto levels. Send the non-dominated sorting and Pareto level division results to the scenario application module;

[0124] Scenario application module: After performing a content randomization operation on all initial decision variable configurations, perform iterative optimization on the newly generated decision variable configurations. When the convergence condition is met, output all Pareto front solution sets and dynamically select a Pareto front solution set according to the scenario change and apply it to the corresponding scenario.

[0125] The above formulas are all dimensionless and only take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to get a formula that is closest to the actual situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0126] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. Here, A and B can be singular or plural. In addition, the character " / " in this article generally represents an "or" relationship between the associated objects before and after, but it may also represent an "and / or" relationship. Specifically, it can be understood by referring to the context before and after.

[0127] It should be understood that in various embodiments of the present application, the magnitudes of the serial numbers of the above processes do not mean the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0128] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed in this article can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0129] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in the present application, and all should be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A real-time ranging and tracking method for dynamic targets with multi-camera collaborative perception, characterized in that: The tracking method includes the following steps: The tracking system generates multiple initial decision variable configurations through initialization. After calculating the objective values of each initial decision variable configuration based on the objective function, non-dominated sorting is performed on all initial decision variable configurations and divided into different Pareto levels; After performing a content randomization operation on all initial decision variable configurations, the newly generated decision variable configurations are iteratively optimized. When the convergence condition is met, all Pareto front solutions are output, and the Pareto front solutions are dynamically selected according to the scenario change and applied to the corresponding scenario.

2. The real-time ranging and tracking method for dynamic targets with multi-camera collaborative perception according to claim 1, wherein: Calculate the objective value for each initial decision variable configuration based on the objective function, and the function expression is: , where is the objective value,[[]] is the robustness index,[[]] is the response delay index,[[]] , are the proportionality coefficients of the robustness index and the response delay index respectively, and , are both greater than 0.

3. The real-time ranging and tracking method for dynamic targets with multi-camera collaborative perception according to claim 2, characterized in that: Performing non-dominated sorting on all initial decision variable configurations and dividing them into different Pareto levels includes the following steps: After selecting the current initial decision variable configuration, compare it one by one with the current initial decision variable configuration; If the current initial decision variable configuration is not worse than the compared initial decision variable configuration in all parameters, and at least one parameter value of the current initial decision variable configuration is better than the compared initial decision variable configuration, then the current initial decision variable configuration dominates the compared initial decision variable configuration; If the compared initial decision variable configuration is not worse than the current initial decision variable configuration in all parameters, and at least one parameter value of the compared initial decision variable configuration is better than the current initial decision variable configuration, then the compared initial decision variable configuration dominates the current initial decision variable configuration; Otherwise, the current initial decision variable configuration does not dominate the compared initial decision variable configuration, or the compared initial decision variable configuration does not dominate the current initial decision variable configuration; If there is no dominance relationship between the initial decision variable configurations, it is recorded as a non-dominated solution, that is, there is no solution in which any other initial decision variable configuration dominates the current initial decision variable configuration; Sort according to the dominance relationship between the initial decision variable configurations; The first Pareto front: contains the initial decision variable configurations that are not dominated by any other initial decision variable configurations; The second Pareto front: contains the initial decision variable configurations that are dominated by the initial decision variable configurations of the first Pareto front but are not dominated by other initial decision variable configurations.

4. The real-time ranging and tracking method for dynamic targets with multi-camera collaborative perception according to claim 3, wherein: In each Pareto front, calculate the crowding degree of the initial decision variable configuration, and the expression is: , where is the crowding degree of the current initial decision variable configuration, , are the objective values of the adjacent initial decision variable configurations of the current initial decision variable configuration, is the maximum objective value in the Pareto front, is the minimum objective value in the Pareto front; In each Pareto front, all initial decision variable configurations are sorted from largest to smallest according to the crowding degree.

5. The real-time ranging and tracking method for dynamic targets with multi-camera collaborative perception according to claim 4, characterized in that: After performing a content randomization operation on all initial decision variable configurations, the newly generated decision variable configurations are iteratively optimized. When the convergence condition is met, outputting all Pareto front solutions includes the following steps: Based on the initial decision variable configuration, a randomization operation is performed to generate a new initial decision variable configuration. The randomization operation includes random perturbation, probability-based change, and mutation operation; Recalculate the objective value, perform non-dominated sorting, and perform Pareto level division steps on the initial decision variable configuration that has completed the content randomization operation to generate a new Pareto front; When the number of iterations is equal to the number threshold, the change in the objective value is less than the objective value change threshold, or the change in the crowding degree is less than the crowding degree change threshold, it is determined that the convergence condition is met. When the convergence condition is met, all Pareto front solutions are output.

6. The real-time ranging and tracking method for dynamic targets with multi-camera collaborative perception according to claim 5, characterized in that: The calculation logic of the robustness index is as follows: Obtain the ranging error and tracking failure rate of the initial decision variable configuration, perform normalization processing on the ranging error and tracking failure rate to map the value ranges of the ranging error and tracking failure rate to between [0, 1], obtain the ranging error normalization value and the tracking failure rate normalization value, and calculate the robustness index. The expression is: , where is the robustness index, is the ranging error normalization value, is the tracking failure rate normalization value; The calculation logic of the response delay index is as follows: obtain the algorithm delay and network transmission delay of the initial decision variable configuration, and sum the algorithm delay and network transmission delay to obtain the response delay index.

7. The real-time ranging and tracking method for dynamic targets with multi-camera collaborative perception according to claim 6, characterized in that: The content of the initial decision variable configuration includes the resolution, frame rate, view coverage range of each camera, weight allocation for multi-camera depth estimation, and camera task priority allocation.

8. A dynamic target real-time ranging and tracking system with multi-camera collaborative perception, which is used to implement the tracking method described in any one of claims 1-7, and is characterized in that: It includes a configuration generation module, a non-dominated sorting module, and a scenario application module: Configuration generation module: Generate multiple initial decision variable configurations through initialization, and calculate the objective value of each initial decision variable configuration based on the objective function; Non-dominated sorting module: Perform non-dominated sorting on all initial decision variable configurations and divide them into different Pareto rank blocks; Scenario application module: After performing content randomization operations on all initial decision variable configurations, perform iterative optimization on the newly generated decision variable configurations. When the convergence condition is met, output all Pareto front solution sets, and dynamically select the Pareto front solution set according to the scenario change and apply it to the corresponding scenario.

Images