Multi-objective Trajectory Prediction Method and System Based on Improved Probabilistic Graph Model

By improving the probability graph model and using target motion information to build a multi-objective distribution probability model, the problem that traditional models cannot predict unknown dynamic target trajectories is solved, and high-precision trajectory prediction is achieved.

CN116300929BActive Publication Date: 2025-07-08Chinese People's Liberation Army Cyberspace Force Information Engineering University
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202310265840.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-03-17
Publication Date
2025-07-08
Estimated Expiration
2043-03-17

AI Technical Summary

Technical Problem

The traditional target probability graph model cannot fully utilize the target motion state information, resulting in the inability to effectively solve the trajectory prediction problem of unknown dynamic targets.

Method used

The improved probability graph model is adopted to construct a multi-objective distribution probability model using the target motion information structure, and the target motion trajectory is fitted through a two-dimensional normal probability distribution map to predict the trajectory of unknown dynamic targets.

Benefits of technology

It improves the accuracy of trajectory prediction and can effectively predict unknown dynamic single-objective and multi-objective motion trajectories, which is convenient for practical application.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116300929B_ABST
    Figure CN116300929B_ABST
Patent Text Reader

Abstract

The present invention relates to the technical field of trajectory planning, and particularly relates to a multi-objective trajectory prediction method and system based on an improved probability graph model. The target motion information structure is set according to the initial positions and initial velocities of each target, and a multi-objective distribution probability model within a preset area is constructed. The target motion information structure is used to solve each target distribution probability model, and the corresponding target trajectory distribution probability is predicted according to the solution result. By improving the target probability graph model, the present invention uses the target motion information to improve the prediction model of unknown dynamic targets, and realizes the fitting of the target motion trajectory by generating a directional two-dimensional normal probability distribution graph, thereby being able to realize the trajectory prediction of unknown dynamic single targets and / or multi-targets, improve the accuracy of trajectory prediction, and facilitate the application in actual scenarios.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of trajectory planning, and particularly to a multi-target trajectory prediction method and system based on an improved probability graph model. Background Technique

[0002] Target tracking technology refers to using the measurements obtained by sensors to estimate the state of a target, including information such as the number, position, speed, and acceleration of the target. Multi-target tracking (MTT) technology is widely used in military fields such as aerial reconnaissance and early warning, missile defense, and battlefield surveillance, as well as civilian fields such as robot vision, air navigation, and traffic control. The key to multi-target tracking technology lies in solving the data association problem between measurements and targets. The Target Probability Map (TPM) model can be used to solve the target tracking problem in a multi-sensor network. First, define the association variables between targets and measurements, use the variable nodes in the graph model to represent the association variables, and the factor nodes between the variable nodes represent the relationships between the association variables. Then construct a probability graph model for describing the joint distribution of all association variables, and solve the graph model through the belief propagation algorithm to obtain the marginal distribution or maximum a posteriori estimate of all association variables, thereby realizing multi-target tracking. The traditional TPM algorithm only focuses on the distribution of unknown static targets and cannot make full use of the target motion state information. Therefore, it cannot solve the trajectory prediction problem of unknown dynamic targets. Summary of the Invention

[0003] For this reason, the present invention provides a multi-target trajectory prediction method and system based on an improved probability graph model to solve the problem that it is difficult to make full use of the target motion state information, and uses a normal probability distribution graph to realize the fitting of multi-target motion trajectories.

[0004] According to the design scheme provided by the present invention, a multi-target trajectory prediction method based on an improved probability graph model is provided, including:

[0005] Set the target motion information structure according to the initial positions and initial velocities of each target, and construct a multi-target distribution probability model within a preset area;

[0006] Use the target motion information structure to solve each target distribution probability model, and predict the corresponding target trajectory distribution probability according to the solution result.

[0007] As the multi-target trajectory prediction method based on the improved probability graph model of the present invention, further, the target motion information structure is expressed as: where, (x0, y0) represents the initial position of the target, represents the initial velocity vector of the target.

[0008] As the multi - target trajectory prediction method based on the improved probability graph model of the present invention, further, a multi - target distribution probability model within a preset area is constructed, including the following contents:

[0009] First, obtain the coordinate range of the preset area and the initial positions of each target;

[0010] Next, according to the two - dimensional normal distribution algorithm and using the initial positions of the targets and the coordinate range of the preset area, construct the distribution probability model of each target;

[0011] Then, obtain the multi - target distribution probability model by performing normalization fusion processing on the distribution probability models of each target.

[0012] As the multi - target trajectory prediction method based on the improved probability graph model of the present invention, further, use the target motion information structure to solve the distribution probability model of each target, including the following contents:

[0013] First, obtain the predicted positions of the targets after a preset time interval according to the target motion information structure, and construct the ratio relationship between the horizontal axis variance and the vertical axis variance at the predicted positions of the targets according to the variance theorem;

[0014] Next, construct the sum - type relationship between the target motion step length within the preset time interval and the sum of the horizontal axis variance and the vertical axis variance of the target predicted positions according to the Pauta criterion and the Pythagorean theorem, and use the ratio relationship between the horizontal axis variance and the vertical axis variance at the predicted positions of the targets and the tangent value of the target velocity direction to obtain the horizontal axis variance and the vertical axis variance at the predicted positions of the targets;

[0015] Then, obtain the covariance - related matrix according to the two - dimensional normal probability distribution theorem, and obtain the corresponding target trajectory distribution probability by substituting the covariance - related matrix into the target distribution probability model.

[0016] Further, based on the above - mentioned method, the present invention also provides a multi - target trajectory prediction system based on the improved probability graph model, including: a multi - target model construction module and a multi - target model solution module, where,

[0017] The multi - target model construction module is used to set the target motion information structure according to the initial positions and initial velocities of each target, and construct a multi - target distribution probability model within a preset area;

[0018] The multi - target model solution module uses the target motion information structure to solve the distribution probability model of each target, and predicts the corresponding target trajectory distribution probability according to the solution results.

[0019] The beneficial effects of the present invention:

[0020] The present invention improves the target probability graph model, uses target motion information to improve the prediction model of unknown dynamic targets, and realizes the fitting of target motion trajectories by generating a directional two-dimensional normal probability distribution graph, thereby enabling the trajectory prediction of unknown dynamic single targets and / or multi-targets, improving the accuracy of trajectory prediction, and facilitating applications in actual scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 Schematic diagram of the multi-target trajectory prediction process based on the improved probability graph model in the embodiment;

[0022] Figure 2 Schematic diagram of the principle of the multi-target trajectory prediction algorithm in the embodiment;

[0023] Figure 3 Schematic diagram of the multi-target probability distribution graph in the embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0024] To make the objectives, technical solutions, and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and technical solutions.

[0025] In the embodiment of the present invention, as shown in Figure 1 a multi-target trajectory prediction method based on an improved probability graph model is provided, including:

[0026] S101. Set the target motion information structure according to the initial positions and initial velocities of each target, and construct a multi-target distribution probability model within a preset area.

[0027] In specific applications, p(x, y) can be defined as the distribution probability of the target at the position (x, y), and the target motion information structure can be constructed according to the following expression

[0028]

[0029] where (x0, y0) represents the initial position of the target, represents the initial velocity vector of the target.

[0030] Furthermore, in constructing the multi-target distribution probability model within the preset area, it can be designed to include the following content:

[0031] First, obtain the coordinate range of the preset area and the initial positions of each target;

[0032] Next, construct the distribution probability model of each target according to the two-dimensional normal distribution algorithm and using the initial position of the target and the coordinate range of the preset area;

[0033] Then, obtain the multi-target distribution probability model through normalization and fusion processing of the distribution probability models of each target.

[0034] Set as the distribution probability model of the target within the L X ×L Y region. According to the two-dimensional normal distribution algorithm, the distribution probability model corresponding to the target can be expressed as:

[0035]

[0036]

[0037] Among them, σ x and σ y respectively represent the horizontal axis variance and the vertical axis variance, and ρ is a constant parameter representing the linear correlation coefficient.

[0038] According to the motion information of multiple targets, generate multiple target distribution probabilities, denoted as Normalize and fuse to obtain the multi-target probability distribution model P MTPM , which is expressed as:

[0039]

[0040] Among them, Sigmoid(g) represents the normalization function, and I represents the number of targets.

[0041] S102. Use the target motion information structure to solve the distribution probability model of each target, and predict the trajectory distribution probability of the corresponding target based on the solution result.

[0042] Furthermore, using the target motion information structure to solve the distribution probability model of each target can be designed to include the following content:

[0043] First, obtain the predicted position of the target after a preset time interval based on the target motion information structure, and construct the ratio relationship between the horizontal axis variance and the vertical axis variance at the predicted position of the target according to the variance theorem;

[0044] Next, construct the sum formula relationship between the target motion step length within the preset time interval and the sum of the horizontal axis variance and the vertical axis variance at the predicted position of the target according to the Pauta criterion and the Pythagorean theorem, and use the ratio relationship between the horizontal axis variance and the vertical axis variance at the predicted position of the target and the tangent value of the target velocity direction to obtain the horizontal axis variance and the vertical axis variance at the predicted position of the target;

[0045] Then, obtain the covariance correlation matrix according to the two-dimensional normal probability distribution theorem, and obtain the trajectory distribution probability of the corresponding target by substituting the covariance correlation matrix into the target distribution probability model.

[0046] Set the time interval as τ. According to the current motion state of the target, the target position (x τ , y τ)。According to the variance theorem, the horizontal axis variance σ is established x and the vertical axis variance σ y The ratio relationship between them can be expressed as follows:

[0047]

[0048]

[0049]

[0050] where and respectively represent The vertical and horizontal axis components of.

[0051] According to the Pauta criterion and the Pythagorean theorem, the movement step of the target within τ time can be related to the horizontal axis variance σ x and the vertical axis variance σ y A summation relationship is established and constructed according to the following expression

[0052]

[0053] Substituting the result of the ratio relationship, the horizontal axis variance σ x and the vertical axis variance σ y can be solved and expressed as follows:

[0054]

[0055]

[0056] where represents the tangent value of the target velocity vector.

[0057] According to the two-dimensional normal probability distribution theorem, the covariance matrix Φ and the distribution probability p(x, y) of the target in the established area can be obtained as follows:

[0058]

[0059] p(x, y)~N(x0, y0, Φ)

[0060] where ρ is a constant parameter representing the linear correlation coefficient.

[0061] Furthermore, based on the above method, an embodiment of the present invention also provides a multi-target trajectory prediction system based on an improved probability graph model, including: a multi-target model construction module and a multi-target model solving module, where,

[0062] The multi-target model construction module is used to set the target motion information structure according to the initial positions and initial velocities of each target, and construct a multi-target distribution probability model within a preset area;

[0063] The multi-objective model solving module uses the target motion information structure to solve the distribution probability models of each target, and predicts the distribution probability of the corresponding target trajectory based on the solution results.

[0064] To verify the effectiveness of the solution of this case, the following further explanations will be made in combination with experimental data:

[0065] The algorithm used in the experiment is as Figure 2 shown. According to the target motion information, a distribution probability model of the target in a given area is constructed. (2) According to the target motion information, predict the motion trajectory of the target. According to the variance theorem, establish the ratio relationship between the horizontal axis variance σ x and the vertical axis variance σ y ; (3) According to the Pauta criterion and the Pythagorean theorem, establish the sum relationship between the horizontal axis variance σ x and the vertical axis variance σ y , and solve for σ x and σ y ; (4) According to the two-dimensional normal probability distribution theorem, obtain the correlation matrix Φ, and obtain the distribution probability p(x, y) of the target in the given area; (5) By repeating steps (1) to (4), a distribution probability model of multiple targets can be generated. After normalization and fusion, a multi-target probability distribution model P MTPM is obtained, and the distribution probability of each target trajectory is obtained by solving.

[0066] Using the Monte Carlo method, set the random sampling I = 10 dynamic target motion state information in the experiment, and set the parameters ρ = 0.5, τ = 30, L X = L Y = 1000. Use the algorithm of this case to generate a multi-target probability distribution model P MTPM as Figure 3 shown. (a) and (b) are the simulation results of the three-dimensional and two-dimensional correlation matrix distributions respectively. It can be seen from the figure that the solution of this case uses the target motion state information and measurement data for association to realize the probability prediction of the trajectory distribution of multiple dynamic targets, and can be used as a data processing solution for platforms such as UAV swarms to perform search tasks, and can effectively improve the search effect for dynamic targets.

[0067] Unless otherwise specifically defined, the relative steps, numerical expressions, and numerical values of the components and steps set forth in these embodiments do not limit the scope of the present invention.

[0068] In the present specification, the various embodiments are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. For the same or similar parts among the embodiments, reference can be made to each other. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple. For related parts, reference can be made to the description in the method section.

[0069] The units and method steps of the examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. To clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described according to functions in the above description. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Those of ordinary skill in the art can use different methods to implement the described functions for each specific application, but such implementation is not considered to exceed the scope of the present invention.

[0070] Those of ordinary skill in the art can understand that all or part of the steps in the above methods can be completed by instructing relevant hardware through a program. The program can be stored in a computer-readable storage medium, such as a read-only memory, a magnetic disk, or an optical disc, etc. Optionally, all or part of the steps of the above embodiments can also be implemented using one or more integrated circuits. Correspondingly, the various modules / units in the above embodiments can be implemented in the form of hardware or in the form of software function modules. The present invention is not limited to any specific form of the combination of hardware and software.

[0071] Finally, it should be noted that the above-described embodiments are only specific implementation manners of the present invention, used to illustrate the technical solutions of the present invention, rather than to limit it. The protection scope of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that any person skilled in the art within the technical scope disclosed by the present invention can still modify the technical solutions recorded in the foregoing embodiments, or can easily think of changes, or make equivalent replacements for some of the technical features; and these modifications, changes, or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A multi-objective trajectory prediction method based on an improved probabilistic graph model, characterized in that Including: Set the target motion information structure according to the initial positions and initial velocities of the targets, obtain the coordinate range of the preset area and the initial positions of the targets, construct the distribution probability models of the targets according to the two-dimensional normal distribution algorithm using the initial positions of the targets and the coordinate range of the preset area, and construct the multi-target distribution probability model within the preset area through the normalization and fusion processing of the distribution probability models of the targets. Among them, the distribution probability model of each target is expressed as: L X ×L Y represents the preset regional coordinate range, σ x and σ y respectively represent the horizontal axis variance and the vertical axis variance at the predicted target motion trajectory position (x, y), ρ represents the linear correlation coefficient, and (x0, y0) represents the target initial position; Obtain the predicted positions of the targets after a preset time interval according to the target motion information structure, construct the ratio relationship between the horizontal axis variance and the vertical axis variance at the predicted positions of the targets according to the variance theorem, construct the sum relationship between the motion step length of the targets within the preset time interval and the sum of the horizontal axis variance and the vertical axis variance at the predicted positions of the targets according to the Pauta criterion and the Pythagorean theorem, and use the ratio relationship between the horizontal axis variance and the vertical axis variance at the predicted positions of the targets and the tangent value of the target velocity direction to obtain the horizontal axis variance and the vertical axis variance at the predicted positions of the targets, obtain the covariance correlation matrix according to the two-dimensional normal probability distribution theorem, and obtain the corresponding target trajectory distribution probability by substituting the covariance correlation matrix into the target distribution probability model, so as to solve the distribution probability models of the targets by using the target motion information structure and predict the corresponding target trajectory distribution probability according to the solution results.

2. The multi-objective trajectory prediction method based on an improved probabilistic graph model according to claim 1, wherein The described target motion information structure is represented as: where (x0, y0) represents the initial position of the target, represents the initial velocity vector of the target.

3. The multi-objective trajectory prediction method based on an improved probabilistic graph model according to claim 1, wherein The multi-object distribution probability model is expressed as: where Sigmoid(·) represents the normalization function, I represents the number of objects, represents the distribution probability model of the i-th object, L X ×L Y represents the preset regional coordinate range.

4. The multi-objective trajectory prediction method based on an improved probabilistic graph model according to claim 1, wherein The ratio relationship between the horizontal axis variance and the vertical axis variance at the target prediction position is expressed as: Among them, and respectively represent the vertical axis and horizontal axis components of the target initial velocity vector .

5. The multi-objective trajectory prediction method based on an improved probabilistic graph model according to claim 4, characterized in that The summation relationship between the target movement step and the variances of the horizontal and vertical axes of the target predicted position within a preset time interval is expressed as: where τ represents the time interval.

6. The multi-objective trajectory prediction method based on the improved probabilistic graph model according to claim 5, wherein The covariance correlation matrix is expressed as: where represents the tangent value of the target velocity vector.

7. A multi-objective trajectory prediction system based on an improved probabilistic graph model, characterized in that, Implemented based on the method described in claim 1, including: a multi-target model construction module and a multi-target model solution module, where The multi-target model construction module is used to set the target motion information structure according to the initial positions and initial velocities of the targets and construct the multi-target distribution probability model within the preset area; The multi-target model solution module uses the target motion information structure to solve the distribution probability models of the targets and predicts the corresponding target trajectory distribution probability according to the solution results.

Citation Information

Patent Citations

  • Robot map building and online planning method under emergency

    CN113253730A

  • Robot trajectory prediction method based on model confidence and Gaussian process

    CN113771034A