A method for predicting trajectory of a maneuvering target based on probability statistics compression

Abstract

The application discloses a kind of based on probability statistics compression's maneuvering target trajectory prediction method, belong to autonomous navigation, traffic planning and remote sensing early warning technical field, main in solving the problem that maneuvering target trajectory prediction is difficult to deal with motion state change.The main technology includes input historical trajectory dataset, pre-processing and extracting maneuvering information, obtain the processed dataset;Create state grid coordinate system;The state of each frame in data set is discretized in grid coordinate system, and the corresponding maneuvering information and weight are stored in limited state grid point, to obtain table T;From table T, the optimal maneuvering decision is extracted, to obtain table T * ;The state of input frame to be predicted is inquired in table T * Neighborhood grid point maneuvering decision is calculated, and the optimal maneuvering decision corresponding to the state to be predicted is calculated;The state of frame to be predicted and optimal maneuvering decision are used in physical model, to predict the state of next time;Repeat execution until specified prediction step, realize multi-step prediction.

Description

Technical Field

[0001] This invention belongs to the fields of autonomous navigation, traffic planning and remote sensing early warning technology, and provides a method for predicting the trajectory of a maneuvering target based on probabilistic statistical compression. Background Technology

[0002] Trajectory prediction is the task of predicting the future position, velocity, direction, and other state information of a target given its past or current trajectory. It is an important component in many fields and has wide applications in autonomous navigation, traffic planning, and remote sensing early warning.

[0003] Based on the method of variable selection, trajectory prediction methods can be divided into three types: local regression, global regression, and autoregression. Local regression typically uses time as the independent variable, and each prediction uses only a segment of trajectory data near the point to be predicted. Local regression decomposes complex trajectories into several simpler segments, reducing the complexity of modeling, but its long-term prediction performance is poor. Currently, commonly used local regression methods in trajectory prediction include locally weighted linear regression and locally weighted multinomial regression.

[0004] Global regression models use macroscopic factors such as the planned route and environmental conditions as independent variables. Each prediction outputs a complete trajectory, thus avoiding the error accumulation problem associated with multi-step predictions and offering better long-term predictive capabilities. However, they cannot perform real-time predictions. Furthermore, they are difficult to represent with explicit mathematical models, thus placing higher demands on model structure and scale compared to local regression. Currently, prediction methods using global regression are relatively few and are typically based on artificial neural networks.

[0005] Autoregression is the most widely used method in trajectory prediction. It utilizes the dependencies (i.e., self-correlation) between the historical time series values ​​of the target at different periods to build a regression model for prediction. It differs from the two regression methods mentioned above in that it uses the same types of independent and dependent variables. Autoregressive models include state estimation methods and artificial neural network methods.

[0006] State estimation methods establish state transition equations based on the position, velocity, and acceleration of a maneuvering target, using observed states to continuously refine the estimation of the system state. However, this method has the following problems: it requires continuous input of the latest observed states during prediction, thus only allowing for single-step prediction; and its prediction of the maneuvering target's maneuvering is based on the state at past moments, making it unable to capture possible future maneuvering changes. Currently, common state estimation methods include Kalman filtering, particle filtering, and hidden Markov models.

[0007] Autoregressive models based on artificial neural networks typically use recurrent neural networks as their underlying structure. However, existing autoregressive models based on artificial neural networks do not perform well in terms of prediction, with their performance ceiling only approaching that of state estimation methods and local regression methods. Summary of the Invention

[0008] The purpose of this invention is to provide a method for predicting the trajectory of maneuvering targets based on probabilistic statistical compression, addressing the following problems in existing technologies: Existing local regression and autoregressive models struggle to capture changes in the future motion state of maneuvering targets, making them unsuitable for long-term prediction; while global regression models theoretically offer better long-term prediction performance, their characteristic of generating a complete trajectory at once prevents real-time prediction based on the current state, and they also consume significant space and have slow prediction speeds. This invention features: fast prediction speed, small storage space, and high accuracy in both short-term and long-term predictions.

[0009] To achieve the above objectives, the present invention adopts the following technical solution:

[0010] A method for predicting the trajectory of a maneuvering target based on probabilistic statistical compression includes the following steps:

[0011] Step 1: Input the historical trajectory dataset D, preprocess it and extract the maneuver information to obtain the processed dataset D*;

[0012] Step 2: Create the state grid coordinate system C;

[0013] Step 3: Transfer dataset D * The state information s of each frame n Discretize in the grid coordinate system C, and then assign the corresponding maneuver information u. n The weights are stored in a finite number of state grid points to obtain table T;

[0014] Step 4: Extract the optimal maneuver decision from the maneuver information and weights in table T, thus obtaining table T containing the optimal maneuver decision. * ;

[0015] Step 5: Input the state of the frame to be predicted, query the maneuver decisions of the neighborhood grid points in table T*, and calculate the optimal maneuver decision corresponding to the state to be predicted based on these decisions.

[0016] Step 6: Apply the state of the frame to be predicted and the optimal maneuver decision obtained in step 5 to the physical model G to predict the state at the next moment.

[0017] Step 7: Repeat steps 5 and 6 until the specified number of prediction steps is reached to achieve multi-step prediction.

[0018] Furthermore, the dataset D processed through step 1...* Each frame contains state information s n and motor information u n The specific formula is as follows:

[0019] D * [n]={s n v n}

[0020] State information refers to known physical information that reflects the current motion state of a target, including but not limited to position, velocity, and orientation angle;

[0021] Maneuver information refers to physical information that reflects the current motion trend of a target and can indirectly or directly affect future state information, including but not limited to acceleration and angular velocity.

[0022] Furthermore, the specific steps of step 2 are as follows:

[0023] Step 2.1: Specify the grid coordinate range and resolution (grid size) for each state variable, including position, velocity, and orientation angle;

[0024] Step 2.2: For each state variable, within its value range, create an arithmetic sequence with the resolution as the step size, which is the state grid coordinate axis. The grid coordinate axes of all state variables determine the state grid coordinate system C.

[0025] Furthermore, T in step 3 is a set of state grid points S - For the index range, subtable T′ is a second-level nested table of values, satisfying the following mapping relationship:

[0026] T:S - →T′

[0027] Where T' is a table with the set of maneuver information U as the index range and the set of weights W as the value range, satisfying the following mapping relationship:

[0028] T′:U→W

[0029] Furthermore, the specific steps of step 3 are as follows:

[0030] Step 3.1: Input dataset D * The nth frame {s n u n}, find status information s n The position in the state grid coordinate system, with that position as the center, and the set of state grid points in its neighborhood.

[0031] Step 3.2: Calculate the probability contribution weight of the current frame to each state grid point in its neighborhood, and add the weight and maneuver information to each state grid point. The specific steps are as follows:

[0032] Step 3.2.1, for the set The m-th state grid point s is calculated using the specified kernel function K1 n Its probability contribution weight w n，m The specific formula is as follows:

[0033]

[0034] Step 3.2.2: If table T does not contain the state grid point, add the grid point to the table, and then add the maneuver information u. n and weight w n，m Add it to the sub-table T′ corresponding to that grid point;

[0035] Step 3.2.3: Repeat steps 3.2.1 and 3.2.2 until the set has been traversed.

[0036] Step 3.4: Repeat steps 3.1 and 3.2 until the dataset D has been traversed. * .

[0037] Furthermore, T in step 4 * It is a set of state grid points S - For the index range, the optimal set of maneuver information U * A table representing the range of values, satisfying the following mapping relationship:

[0038] T * :S - →U *

[0039] Furthermore, the specific steps of step 4 are as follows:

[0040] For the embodiments of the present invention, step 4.1 provides the following three implementation routes:

[0041] Route 1:

[0042] For the m-th state grid point in table T The corresponding sub-table T′ contains k maneuver information items and their corresponding weights, which are set U and set W, respectively. The probability density function of the maneuver is calculated using the kernel density estimation method with a specified kernel function K2. The maneuver with the highest probability density is taken as the optimal maneuver corresponding to the current state grid point. The specific formula is as follows:

[0043]

[0044]

[0045] Where p represents the probability density function of the maneuver, x is the independent variable of the probability density function, h is the specified bandwidth parameter, and u i For the i-th movement information in table T′, w i For u i The corresponding weights;

[0046] Route 2:

[0047] For the m-th state grid point in table T The corresponding sub-table T′ contains several maneuver information items and their corresponding weights. Histogram statistics are performed on T′ at a specified resolution h, and the midpoint of the maneuver interval with the maximum cumulative weight is taken as the optimal maneuver in the grid.

[0048] Route 3:

[0049] For the m-th state grid point in table T The corresponding sub-table T′ contains several maneuver information items and their corresponding weights. Treating T′ as a Gaussian mixture model, the expectation-maximization algorithm is used to estimate multiple means of the mixture distribution of maneuver information. The mean with the highest probability is then taken as the grid-optimal maneuver.

[0050] Step 4.2: Repeat step 4.1 until table T is traversed and obtained. * .

[0051] Furthermore, the specific steps of step 5 are as follows:

[0052] Step 5.1: Locate the position of the state s of the frame to be predicted in the state grid coordinate system, and take the state grid points in its neighborhood as the center.

[0053] Step 5.2, with The element is indexed in table T * By sequentially querying the data, the optimal set of maneuvers for each grid is obtained.

[0054] Step 5.3, for The nth state grid point Calculate the optimal mesh maneuver using the specified kernel function K3. probability contribution weight The specific formula is as follows:

[0055]

[0056] Step 5.4, with For weight calculation The weighted average yields the optimal maneuver decision for the current state. The specific formula is as follows:

[0057]

[0058] Compared with the prior art, the beneficial effects of this invention are as follows:

[0059] I. The prediction model used in this invention can extract the maneuver decisions of maneuvering targets under different states from a large number of historical trajectories, overcoming the problem that existing technologies are difficult to accurately predict when the state of motion changes.

[0060] Second, this invention compresses data using probabilistic statistical methods and employs sparse storage, which can greatly reduce the space occupied by the prediction model.

[0061] Third, this invention completes most of the calculations in the prediction model establishment stage, compressing the information attached to a large amount of historical data into a small amount of information in the state grid points, so that it has a very fast computing speed in the prediction stage. Attached Figure Description

[0062] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0063] Figure 1 This is a flowchart of Embodiment 1 of the present invention;

[0064] Figure 2 The image shows the effect of single-step prediction of a certain trajectory according to an embodiment of the present invention, and a magnified view of a part thereof.

[0065] Figure 3 This is a diagram illustrating the effect of long-term prediction of a certain trajectory according to an embodiment of the present invention. Detailed Implementation

[0066] The present invention will now be further described with reference to the accompanying drawings and specific embodiments:

[0067] Example 1

[0068] A method for predicting the trajectory of a maneuvering target based on probabilistic statistical compression includes the following steps:

[0069] Step 1: Input the historical trajectory dataset D, preprocess it, and extract acceleration and angular velocity to obtain the processed dataset D.* ;

[0070] Step 2: Create a state grid coordinate system using longitude, latitude, velocity, and direction angle as state variables;

[0071] Step 3: Use the kernel density estimation method to process dataset D. * The state s of each frame n Discretize the data in the grid coordinate system C, and store the corresponding acceleration, angular velocity, and weights in a finite number of state grid points to obtain table T. a and Table T ω ;

[0072] Step 4: Use the Gaussian mixture model estimation method to estimate from Table T a and Table T ω The optimal acceleration and angular velocity strategy is extracted from the acceleration, angular velocity, and weight information to obtain the table. and table

[0073] Step 5: Input the state information of the frame to be predicted in the table. The optimal maneuver decision is queried in the grid, and the optimal maneuver information of the frame state to be predicted is calculated based on this decision.

[0074] Step 6: Apply the state information of the frame to be predicted and the optimal maneuver information obtained in Step 5 to the linear approximation turning rate and acceleration (TRA) model to predict the state at the next moment.

[0075] Step 7: Repeat steps 5 and 6 until the specified number of prediction steps is reached to achieve multi-step prediction.

[0076] Furthermore, the specific steps of step 1 are as follows:

[0077] Step 1.1: Input a trajectory from the historical trajectory dataset and reconstruct the time based on latitude, longitude, and speed data. The specific steps are as follows:

[0078] Step 1.1.1: Calculate the displacement d between two adjacent frames (the nth frame and the (n+1th frame) using the semi-versus formula. n The specific formula is as follows:

[0079]

[0080] Where Lon and Lat represent longitude and latitude respectively, and R is the Earth's radius, approximately 6,378,137 meters.

[0081] Step 1.1.2, Displacement d n Except for the speed v of the previous frame nThe interval between two frames can be obtained. By summing the intervals of all previous frames, the reconstruction time t of the current frame can be obtained. n The specific formula is as follows:

[0082]

[0083] Step 1.1.3: Repeat steps 1.1 and 1.2 until the entire trajectory has been traversed.

[0084] Step 1.2: Resample the trajectory processed in Step 1.1 at fixed time intervals. The specific steps are as follows:

[0085] Step 1.2.1: Create an arithmetic sequence as interpolation points using the time interval Δt = 10s and the time range of the current trajectory;

[0086] Step 1.2.2: For one state variable in the trajectory data, use the cubic spline interpolation method to obtain the system of equations and coefficients of cubic spline interpolation based on the reconstruction time fitting.

[0087] Step 1.2.3: Substitute the interpolation points obtained in Step 2.2 into the system of equations obtained in Step 2.1, and use the corresponding coefficients to calculate the function value, which is used as the resampling result of the variable;

[0088] Step 1.2.4, repeat steps 2.2 and 2.3 until all types of variables have been traversed.

[0089] Step 1.3: Extract acceleration and angular velocity from the trajectory obtained in Step 1.2. The specific steps are as follows:

[0090] Step 1.3.1: Use forward differential processing to obtain the velocity change Δv n Divide by the resampling time interval Δt to obtain the acceleration a n ;

[0091] Step 1.3.2: Use forward differential processing to obtain the change in velocity direction Δθ. n Then, the angular velocity ω is calculated using the following formula. n :

[0092]

[0093] Step 1.3.3: Repeat steps 3.1 to 3.4 until the entire trajectory has been traversed.

[0094] Step 1.4: Repeat steps 1.1 to 1.3 until the historical trajectory dataset has been traversed.

[0095] Furthermore, the dataset D processed through step 1... *Each frame contains state information: longitude. n latitude n Speed ​​v n Direction angle θ n ; and motor information: acceleration a n and angular velocity ω n The specific formula is as follows:

[0096] D * [n] = {lon n lat n v n θ n a n ω n}

[0097] Furthermore, the specific steps of step 2 are as follows:

[0098] Step 2.1: Specify the grid coordinate range and resolution for longitude as [115.5°, 121.5°] and 0.01°, respectively; the grid coordinate range and resolution for latitude as [30.5°, 40.5°] and 0.01°, respectively; the grid coordinate range and resolution for velocity as [0m / s, 350m / s] and 1m / s, respectively; and the grid coordinate range and resolution for orientation angle as [0°, 360°] and 1°, respectively.

[0099] Step 2.2: For each state variable, within its range of values, create an arithmetic sequence, i.e., the grid coordinate axes, with the resolution as the step size. The grid coordinate axes of state variables highly correlated with acceleration (longitude, latitude, and velocity) determine the grid coordinate system C1; the grid coordinate axes of state variables highly correlated with angular velocity (longitude, latitude, and orientation angle) determine the grid coordinate system C2.

[0100] Furthermore, the specific steps of step 3 are as follows:

[0101] Step 3.1: Input the nth frame {lon} from the historical trajectory dataset. n lat n v n θ n a n ω n}, find its acceleration-related state s n ={lon n lat n v n The position in the grid coordinate system C1 is taken as the center, and the state grid points in a window with a width of 4 are taken as the set.

[0102] Step 3.2: Calculate the probability weight contributed by each state grid point in the neighborhood of the current frame, and compress the weight and acceleration into each state grid point. The specific steps are as follows:

[0103] Step 3.2.1: For the m-th state grid point in the neighborhood... The probability contribution weight w of the current frame is calculated using a Gaussian kernel function. n，m The specific formula is as follows:

[0104]

[0105] Where ∑ is the covariance matrix of the states, and its specific values ​​are as follows:

[0106]

[0107] Step 3.2.2, if table T a If the state grid point is not included in the table, add the grid point to the table, and then add the maneuver information u. n and weight w n，m Append to table T corresponding to this grid point a 'middle.

[0108] Step 3.2.3: Repeat steps 3.2.1 and 3.2.2 until the set has been traversed.

[0109] Step 3.4: Repeat steps 3.1 and 3.2 until the historical trajectory dataset has been traversed.

[0110] Step 3.5: Compress the angular velocities in the dataset into a table T indexed by longitude, latitude, and orientation angle. ω The specific process is the same as the process described above.

[0111] Furthermore, the specific steps of step 4 are as follows:

[0112] Step 4.1: For a given state grid point in the grid acceleration weight table, the acceleration set A and weight set W are treated as a Gaussian mixture model. The expectation-maximization algorithm is used to estimate multiple means of the acceleration mixture distribution, and the mean with the highest probability is taken as the optimal acceleration a of the grid. * ;

[0113] Use this optimal maneuver data to overwrite the acceleration weight list data.

[0114] Step 4.2: Repeat step 4.1 until the grid acceleration weight table is traversed to obtain the optimal grid acceleration table.

[0115] Step 4.3: Process the mesh angular velocity weight table using the same method as in Steps 4.1 and 4.2 to obtain the optimal mesh angular velocity table.

[0116] Furthermore, the specific steps of step 5 are as follows:

[0117] Step 5.1: Find the position of the state s = {lon, lat, v, θ} of the frame to be predicted in coordinate systems C1 and C2 respectively. Take the state grid points in its neighborhood as the center of the position.

[0118] Step 5.2, use S - The elements are respectively in the table and table The optimal maneuver decision is obtained by querying the query results, resulting in the optimal maneuver set U for the neighborhood grid points. *

[0119] Step 5.3: For the nth state grid point in the neighborhood... Calculating the optimal grid maneuver using the Gaussian kernel function probability contribution weight The specific formula is as follows:

[0120]

[0121] Step 5.4, with For weight calculation The weighted average yields the optimal maneuver decision for the current state. The specific formula is as follows:

[0122]

[0123] Furthermore, step 6 is specifically expressed by the following formula:

[0124]

[0125] Where s′ represents the predicted state after time Δt.

[0126] In summary, the probabilistic statistical compression-based trajectory prediction method for maneuvering targets proposed in this invention has the following technical effects and advantages:

[0127] 1. Historical trajectory data preprocessing and maneuver information extraction: By reconstructing time, resampling, and extracting maneuver information, historical trajectory data is effectively processed to provide an accurate data foundation for subsequent predictions.

[0128] 2. Probability and statistical compression and state grid establishment: The maneuver information is compressed twice into a table indexed by state grid points, reducing storage space requirements.

[0129] 3. Fast query and prediction: Most of the calculations are completed during the data compression stage. During prediction, the machine can quickly query the motion information of neighboring state grid points and make predictions in combination with the physical model, which improves the prediction speed.

[0130] 4. Solving the problem of changes in motion state: This invention can extract the maneuver decisions of maneuvering targets in different states from a large number of historical trajectories, effectively solving the problem that existing technologies are difficult to accurately predict when the motion state changes.

[0131] 5. Multi-step prediction and high accuracy: Multi-step prediction is achieved by repeating query and prediction steps, while maintaining high prediction accuracy.

[0132] In summary, this invention provides an effective method for predicting the trajectory of maneuvering targets, especially in handling changes in the motion state of maneuvering targets, while also featuring fast prediction speed and small storage space requirements.

Claims

1. A method for predicting the trajectory of a maneuvering target based on probabilistic statistical compression, characterized in that, Includes the following steps: Step 1: Input the historical trajectory dataset D, preprocess it, and extract the maneuver information to obtain the processed dataset D. * ; Step 2: Create the state grid coordinate system C; Step 3: Transfer dataset D * The state information s of each frame n Discretize in the grid coordinate system C, and then assign the corresponding maneuver information u. n The weights are stored in a finite number of state grid points to obtain table T; Step 4: Extract the optimal maneuver decision from the maneuver information and weights in table T, thus obtaining table T containing the optimal maneuver decision. * ; Step 5: Input the state of the frame to be predicted in table T. * The system queries the maneuver decisions of neighboring grid points and calculates the optimal maneuver decision corresponding to the state to be predicted based on these decisions. Step 6: Apply the state of the frame to be predicted and the optimal maneuver decision obtained in step 5 to the physical model G to predict the state at the next moment. Step 7: Repeat steps 5 and 6 until the specified number of prediction steps is reached to achieve multi-step prediction.

2. The method for predicting the trajectory of a maneuvering target based on probabilistic statistical compression according to claim 1, characterized in that, Dataset D processed by step 1 * Each frame contains state information s n and motor information u n The specific formula is as follows: D * [n]={s n ,u n } State information refers to known physical information that reflects the current motion state of a target, including position, velocity, and orientation angle. Maneuver information refers to physical information that reflects the current motion trend of a target and can indirectly or directly affect future state information, including acceleration and angular velocity.

3. The method for predicting the trajectory of a maneuvering target based on probabilistic statistical compression according to claim 1, characterized in that, The specific steps of step 2 are as follows: Step 2.1: Specify the grid coordinate range and resolution for each state variable, including position, velocity, and orientation angle; Step 2.2: For each state variable, within its value range, create an arithmetic sequence with the resolution as the step size, which is the state grid coordinate axis. The grid coordinate axes of all state variables determine the state grid coordinate system C.

4. The method for predicting the trajectory of a maneuvering target based on probabilistic statistical compression according to claim 1, characterized in that, In step 3, T is a set of state grid points S. - For the index range, subtable T′ is a second-level nested table of values, satisfying the following mapping relationship: T:S - →T′ Where T′ is a table with the set of maneuver information U as the index range and the set of weights W as the value range, satisfying the following mapping relationship: T′:U→W.

5. The method for predicting the trajectory of a maneuvering target based on probabilistic statistical compression according to claim 1, characterized in that, The specific steps of step 3 are as follows: Step 3.1: Input dataset D * The nth frame {s n u n }, find status information s n The position in the state grid coordinate system, with that position as the center, and the set of state grid points in its neighborhood. Step 3.2: Calculate the probability contribution weight of the current frame to each state grid point in its neighborhood, and add the weight and maneuver information to each state grid point. The specific steps are as follows: Step 3.2.1, for the set The m-th state grid point s is calculated using the specified kernel function K1 n Its probability contribution weight w n，m The specific formula is as follows: Step 3.2.2: If table T does not contain the state grid point, add the grid point to the table, and then add the maneuver information u. n and weight w n，m Add it to the sub-table T′ corresponding to that grid point; Step 3.2.3: Repeat steps 3.2.1 and 3.2.2 until the set has been traversed. Step 3.3: Repeat steps 3.1 and 3.2 until the dataset D has been traversed. * .

6. The method for predicting the trajectory of a maneuvering target based on probabilistic statistical compression according to claim 1, characterized in that, T in step 4 * It is a set of state grid points S - For the index range, the optimal set of maneuver information U * A table representing the range of values, satisfying the following mapping relationship: T * :S - →U * 。 7. The method for predicting the trajectory of a maneuvering target based on probabilistic statistical compression according to claim 1, characterized in that, The specific steps of step 4 are as follows: Step 4.1: For the m-th state grid point in table T The corresponding sub-table T′ contains k maneuver information items and their corresponding weights, which are set U and set W, respectively. The probability density function of the maneuver is calculated using the kernel density estimation method with a specified kernel function K2. The maneuver with the highest probability density is taken as the optimal maneuver corresponding to the current state grid point. The specific formula is as follows: Where p represents the probability density function of the maneuver, x is the independent variable of the probability density function, h is the specified bandwidth parameter, and u i For the i-th movement information in table T′, w i For u i The corresponding weights; Step 4.2: Repeat step 4.1 until all state grid points in table T have been traversed, thus obtaining table T. * .

8. The method for predicting the trajectory of a maneuvering target based on probabilistic statistical compression according to claim 7, characterized in that, The specific steps of step 4 are as follows: Step 4.1: For the m-th state grid point in table T The corresponding sub-table T′ contains several maneuver information items and their corresponding weights. Treating T′ as a Gaussian mixture model, the expectation-maximization algorithm is used to estimate multiple means of the mixture distribution of maneuver information. The mean with the highest probability is taken as the optimal maneuver corresponding to the current state grid point. Step 4.2: Repeat step 4.1 until all state grid points in table T have been traversed, thus obtaining table T. * .

9. The method for predicting the trajectory of a maneuvering target based on probabilistic statistical compression according to claim 1, characterized in that, The specific steps of step 4 are as follows: Step 4.1: For the m-th state grid point in table T The corresponding sub-table T′ contains several maneuver information and their corresponding weights. Histogram statistics are performed on T′ at a specified resolution h, and the midpoint of the maneuver interval with the maximum cumulative weight is taken as the optimal maneuver corresponding to the current state grid point. Step 4.2: Repeat step 4.1 until all state grid points in table T have been traversed, thus obtaining table T. * .

10. The method for predicting the trajectory of a maneuvering target based on probabilistic statistical compression according to claim 1, characterized in that, The specific steps of step 5 are as follows: Step 5.1: Locate the position of the state s of the frame to be predicted in the state grid coordinate system, and take the state grid points in its neighborhood as the center. Step 5.2, with The element is indexed in table T * By sequentially querying the data, the optimal set of maneuvers for each grid is obtained. Step 5.3, for The nth state grid point Calculate the optimal mesh maneuver using the specified kernel function K3. probability contribution weight The specific formula is as follows: Step 5.4, with The weighted average of the optimal maneuver set for each grid is calculated to obtain the optimal maneuver decision for the current state. The specific formula is as follows:

Images