Target dynamic and static attribute judgment method based on multi-frame data

Abstract

The invention provides a multi-frame data-based target dynamic and static attribute judgment method, which comprises the following steps of: firstly, generating all possible tracks by combining kinematics constraints, simultaneously accumulating value functions of associated scattering points, then carrying out track pruning according to the accumulated value functions, then carrying out threshold comparison, and finally obtaining a high-quality moving target track. The moving point target can be effectively tracked and recognized in the time dimension, and the judgment performance of the small moving target is improved. And the accumulation range of the trajectory is limited by utilizing kinematics constraint in the trajectory generation stage, and redundant trajectory information is quickly eliminated in the trajectory trimming stage. The calculation amount is effectively reduced, and it is ensured that the judgment efficiency of the moving target is not affected by calculation delay of the system.

Description

Technical Field

[0001] The present invention relates to the technical field of automobile radars, and in particular to a method for determining target dynamic and static attributes based on multi-frame data. Background Art

[0002] With the rapid development of Advanced Driving Assistance Systems (ADAS), the role of on-board sensors has become increasingly important. ADAS uses sensors to perceive the surrounding environment. Currently, millimeter-wave radar and cameras are the two primary sensors for ADAS. Millimeter-wave radar has high anti-interference capabilities and excellent penetration capabilities, enabling it to maintain stable performance in adverse weather conditions. Furthermore, compared to vision-based detection, radar typically measures distance and speed more accurately. Therefore, millimeter-wave radar, with its all-day and all-weather advantages, has become the primary sensor for ADAS applications such as adaptive cruise control, automatic emergency braking, and forward collision avoidance.

[0003] In practical applications, an on-board millimeter-wave radar emits electromagnetic waves and receives reflected signals. After analog processing and analog-to-digital conversion, the signal processing module performs target detection and parameter measurement, obtaining information such as the distance, angle, and relative velocity between the ego vehicle and surrounding objects. The trajectory processing module then processes the data and further tracks the measured information to construct a smooth and accurate target trajectory. The driving decision-making system then receives this information and makes rapid decisions based on the current situation to ensure driving safety. When implementing ADAS, we place particular emphasis on the detection and tracking of moving objects. The position and velocity of moving objects, such as vehicles, pedestrians, and obstacles, are crucial to driving safety. Accurately identifying and tracking these moving objects allows the driving system to react promptly and avoid potential collisions, thereby ensuring safe driving and accurate driving decisions.

[0004] Traditional methods for determining whether a target is moving or static rely on an accurate estimate of its velocity. Since the ego vehicle is in motion, this requires first compensating the target's velocity to determine its absolute velocity. Then, based on a single frame of data, the target is determined and track initiation is completed. This method can effectively determine and track moving targets in some simple scenarios, but it has difficulties determining moving point targets in complex dynamic environments.

[0005] First, traditional methods for compensating ego vehicle speed for targets introduce errors. High-precision compensation is difficult to achieve, leading to inaccurate determination of moving and static targets, affecting track initiation and even causing false tracks. Furthermore, due to measurement errors, the motion state of point targets cannot be accurately estimated based on single-frame data. Consequently, traditional methods perform poorly for determining the motion and static state of moving targets such as pedestrians and cyclists. Summary of the Invention

[0006] In response to the problems existing in the prior art, the present invention provides a method for determining the motion and still attributes of a target based on multi-frame data to solve the technical problem that the traditional methods in the prior art have poor performance in determining the motion and still attributes of moving targets such as pedestrians and bicycles.

[0007] The present invention provides a method for determining target dynamic and static attributes based on multi-frame data, comprising:

[0008] S1. Collect radar measurement data, initialize all measurement points in the first frame of radar measurement data, and calculate their value functions;

[0009] S2. Match the measurement points of each frame in the radar measurement data with the accumulation range of the measurement points of the next frame according to the kinematic constraints, accumulate the value functions of the measurement points that meet the constraints, output the accumulated value functions, and obtain all estimated trajectories and their corresponding accumulated value functions;

[0010] S3. Prune all estimated trajectories, retain the trajectory with the largest cumulative value function among the same measurement points, and delete the measurement points associated with the pruned retained trajectory to obtain an updated measurement point set. Perform steps S1-S2 again on the measurement point set until all measurement points of the last frame are deleted, and obtain the pruned estimated trajectory and its corresponding cumulative value function.

[0011] S4. Performing a threshold comparison on the accumulated value function of the estimated trajectory. If the threshold is exceeded, defining the measurement point associated with the second estimated trajectory as a moving point; otherwise, classifying it as a stationary point.

[0012] S5. Estimating the speed of the estimated trajectory of the moving target, generating a speed grid, estimating the target speed in the speed grid, outputting the horizontal and vertical speeds of the target, and determining whether the target is moving or still.

[0013] Optionally, the collecting radar measurement data, initializing all measurement points in the first frame of the radar measurement data, and calculating their value functions include:

[0014] S101. Use a millimeter-wave radar to obtain L-frame radar measurement data of the surrounding environment, which is defined as:

[0015]

[0016] Represents the point index set of the l-th frame measurement data at the j-th iteration, and records the measurement data of the l-th frame as:

[0017]

[0018] The measurement data of the first frame is obtained by D l The kth measurement point is recorded as:

[0019]

[0020] in, Indicates the time when the measurement point is obtained, Respectively represent the measured relative distance, angle and Doppler frequency;

[0021] S102, when l=1 frame of measurement data, initialize all measurement points, and for measurement point s k,1 , based on IMU to obtain the vehicle speed v s =[v sx ,v sy ] T , calculate the Doppler frequency of the measured angle and vehicle speed as Where λ is the wavelength, and the function to calculate its value is:

[0022]

[0023] Optionally, matching the measurement points of each frame in the radar measurement data with the accumulation range of the measurement points of the next frame according to the kinematic constraints, accumulating the value functions of the measurement points that satisfy the constraints, and outputting the accumulated value functions to obtain all estimated trajectories and their corresponding accumulated value functions includes:

[0024] S201. When 2≤l≤L, calculate s k,l-1 and D l Measurement points n,l The radial velocity v r,l and tangential velocity v t,l , and calculate their velocity errors and Expressed as:

[0025]

[0026] Combined k,l-1 The measurement point s associated with the previous frame m,l-2 , calculate s k,l-1 and D l Measurement points n,l Radial acceleration a r,l and tangential acceleration a t,l, and calculate their acceleration errors and Expressed as:

[0027]

[0028]

[0029] S202, set the error weight β, combined with the speed error and Acceleration error and Calculate measurement points k,l-1 and s n,l and determine whether the speed and acceleration are less than the maximum speed v max and the maximum acceleration a max , expressed as:

[0030]

[0031] If less than s k,l-1 and s n,l The match is successful, otherwise it fails, where x + =max{x,0};

[0032] S203, record the measurement point index n matched in the first frame into the trajectory set M k,l If all measurement points fail to match, the index is recorded as 0; and in 2≤l≤L, for the measurement point s k,l-1 While matching, k,l Each trajectory of ξ k,l Accumulate and get the accumulated value function F k,l , if the lth frame does not have s k,l-1 Match the successful measurement point, take η as the accumulation parameter of l frame, and the accumulation value function F k,l Expressed as:

[0033] F k,l =F k,l-1 +z n,l ,n∈ξ k,l

[0034] Among them, the number of frames p that are not matched to points in a row does not exceed P.

[0035] Optionally, the method of pruning all estimated trajectories, retaining the trajectory with the largest cumulative value function in the same measurement point, and deleting the measurement points associated with the pruned retained trajectory to obtain an updated measurement point set, and performing steps S1-S2 on the measurement point set again until all measurement points of the last frame are deleted, obtaining the pruned estimated trajectory and its corresponding cumulative value function, includes:

[0036] S301, measuring point s k,1 The set of all trajectories M k,L Perform pruning, if M k,L Contains more than one trajectory, retaining the maximum cumulative value function F k,L The trajectory τ k,L Defined as s k,1 The best trajectory of

[0037] S302, delete the best trajectory τ k,L Associated measurement points to obtain the updated measurement point set

[0038] S303, and its corresponding L frame measurement point executes steps S1-S2 again until The iteration is stopped when , and the estimated trajectory after pruning and its corresponding accumulation value function are obtained.

[0039] Optionally, performing a threshold comparison on the accumulated value function of the estimated trajectory, and defining the measurement point associated with the second estimated trajectory as a moving point if the threshold is exceeded, and otherwise classifying the measurement point as a stationary point, includes:

[0040] S401, the optimal trajectory τ k,L The corresponding cumulative value function F k,L Compared with the threshold γ, if F k,L is greater than γ, then determine τ k,L The associated measurement points are from moving targets, and vice versa are classified as stationary points, which can be expressed as:

[0041]

[0042] Optionally, performing velocity estimation on the estimated trajectory of the moving target, generating a velocity grid, estimating the target velocity in the velocity grid, outputting the lateral and longitudinal velocities of the target, and determining whether the target is moving or still, includes:

[0043] S501, divide the lateral velocity grid into The longitudinal velocity grid is divided into Generate I x ×I y Group velocity grid;

[0044] S502, the optimal trajectory τ k,L Associated measurement point of the first frame Calculate the distance between the measurement point and each set of velocity grids, and then use the value function z k,l Compare and get I x ×I y Ratio function of group velocity grid Expressed as:

[0045]

[0046] S503, for the optimal trajectory τ k,L The L frame measurement points are summed for each group of velocity grids, and the velocity grid corresponding to the minimum value is output. Combine it with the vehicle speed v s =[v sx ,v sy ] T By taking the difference, we can get the horizontal and vertical velocities of the trajectory, which can be expressed as:

[0047]

[0048] Compared with the prior art, the present invention:

[0049] (1) Improve the performance of moving point target judgment in dynamic environments by accumulating value functions over multiple frames

[0050] The present invention eliminates the need to compensate scattering points for vehicle speed when determining moving targets. Instead, it combines measurement point information and vehicle speed to design a value function for detecting moving targets. Specifically, all possible trajectories are generated using kinematic constraints, while the value functions of associated scattering points are accumulated. Trajectories are then pruned based on the accumulated value functions, followed by a threshold comparison to obtain high-quality moving target trajectories. This method effectively tracks and identifies moving point targets in the temporal dimension, improving the detection performance of small moving targets.

[0051] (2) Effectively reduce the impact of measurement errors and improve the performance of point target state estimation

[0052] The present invention can accurately obtain the absolute lateral and longitudinal velocities of a moving target. Specifically, single-frame data can only determine the relative radial velocity between the target and the radar, but cannot accurately determine the target's absolute velocity. The present invention utilizes multi-frame trajectory measurement data, combined with the Doppler effect, to discretize the lateral and longitudinal velocities into a velocity grid. The target's lateral and longitudinal velocities are then found within this velocity grid, thereby determining the target's motion state.

[0053] (3) Reduce the amount of computation in the system and improve decision efficiency

[0054] When processing multi-frame data, the present invention uses kinematic constraints to limit the range of trajectory accumulation during the trajectory generation phase and quickly removes redundant trajectory information during the trajectory pruning phase. This effectively reduces the amount of computation and ensures that the system's efficiency in detecting moving targets is not affected by computational delays. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0056] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0057] Figure 1 Schematic diagram of the process of the present invention;

[0058] Figure 2 It is a schematic diagram of the system block diagram of the present invention;

[0059] Figure 3 Schematic diagram of calculating relative radial velocity and tangential velocity in the present invention;

[0060] Figure 4 This is a schematic diagram of the simulation of the median function of the present invention;

[0061] Figure 5 This is a simulation diagram of the median function of the present invention. DETAILED DESCRIPTION

[0062] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other implementation cases obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application. The functional units with the same labels in the examples of the present invention have the same and similar structures and functions.

[0063] See also Figure 1 The present invention provides a method for determining target dynamic and static attributes based on multi-frame data, comprising:

[0064] S1. Collect radar measurement data, initialize all measurement points in the first frame of radar measurement data, and calculate their value functions;

[0065] S2. Match the measurement points of each frame in the radar measurement data with the accumulation range of the measurement points of the next frame according to the kinematic constraints, accumulate the value functions of the measurement points that meet the constraints, output the accumulated value functions, and obtain all estimated trajectories and their corresponding accumulated value functions;

[0066] S3. Prune all estimated trajectories, retain the trajectory with the largest cumulative value function among the same measurement points, and delete the measurement points associated with the pruned retained trajectory to obtain an updated measurement point set. Perform steps S1-S2 again on the measurement point set until all measurement points of the last frame are deleted, and obtain the pruned estimated trajectory and its corresponding cumulative value function.

[0067] S4. Performing a threshold comparison on the accumulated value function of the estimated trajectory. If the threshold is exceeded, defining the measurement point associated with the second estimated trajectory as a moving point; otherwise, classifying it as a stationary point.

[0068] S5. Estimating the speed of the estimated trajectory of the moving target, generating a speed grid, estimating the target speed in the speed grid, outputting the horizontal and vertical speeds of the target, and determining whether the target is moving or still.

[0069] See also Figure 1 and Figure 2 In this embodiment, all scattering points in the first frame are initialized and their value functions are calculated. Then, for each scattering point, the value function is accumulated from the second frame to the last frame. The kinematic constraints are used to limit the accumulation range of the trajectory, and only the value functions of the scattering points that meet the constraints are accumulated. At the same time, the scattering points are associated. Finally, all possible trajectories and their corresponding accumulated value functions are obtained, specifically:

[0070] S101, using millimeter wave radar as a sensor to obtain L-frame radar measurement data of the surrounding environment, defining Represents the point index set of the measurement data of the first frame at the jth iteration. The measurement data of the first frame is recorded as The frame is composed of D l The kth measurement point is recorded as:

[0071]

[0072] In the above formula, Indicates the time when the measurement point is obtained, Represent the measured relative distance, angle and Doppler frequency respectively.

[0073] S102, in the first frame l=1, initialize all measurement points S1. k,1 , based on IMU to obtain the vehicle speed v s =[v sx ,v sy ] T , calculate the Doppler frequency of the measured angle and vehicle speed as Where λ is the wavelength. The function to calculate its value is:

[0074]

[0075]

[0076] According to the kinematic constraints, the measurement points of each frame in the radar measurement data are matched with the accumulation range of the measurement points of the next frame, and the value functions of the measurement points that meet the constraints are accumulated. The accumulated value function is output to obtain all estimated trajectories and their corresponding accumulated value functions, specifically:

[0077] S201, when 2≤l≤L, for the measurement point s k,l-1 , according to the kinematic constraints, let it and the measurement point S of the next frame l To match, such as Figure 3 Calculate s k,l-1 and D l Measurement points n,l The radial velocity v r,l and tangential velocity v t,l , and calculate their velocity errors and Combined k,l-1 The measurement point s associated with the previous frame m,l-2 , calculate s k,l-1 and D l Measurement points n,l Radial acceleration a r,l and tangential acceleration a t,l , and calculate their acceleration errors and

[0078]

[0079]

[0080] S202, set the error weight β, calculate the measurement point s k,l-1 and s n,l If the speed and acceleration between them are less than the maximum speed v max and the maximum acceleration a max , then s k,l-1 and s n,l The match was successful.

[0081]

[0082] Among them, x + =max{x,0}.

[0083] S203, record the measurement point index n matched in the first frame into the trajectory set M k,l If all measurement points fail to match, the index is recorded as 0. When 2≤l≤L, for the measurement point s k,l-1While matching, k,l Each trajectory of ξ k,l Accumulate and get the accumulated value function F k,l If the first frame does not have k,l-1 For the measurement point that is successfully matched, η is used as the accumulation parameter of this frame, and the number of frames p without matching points does not exceed P.

[0084] F k,l =F k,l-1 +z n,l , n∈ξ k,l (15)

[0085] All estimated trajectories are pruned. If multiple trajectories come from the same measurement point, only the trajectory with the largest cumulative value function is retained. Subsequently, the measurement points associated with the pruned trajectory are deleted to obtain an updated set of measurement points. Step 1 is performed again for the updated multi-frame measurement points to perform a new round of trajectory generation until all measurement points in the last frame are deleted. Finally, the pruned trajectory and its corresponding cumulative value function are obtained, specifically:

[0086] S301, measuring point s k,1 The trajectory set M k,L Perform pruning, if M k,L Contains multiple trajectories, and the maximum cumulative value function F k,L The trajectory τ k,L Defined as s k,1 The best trajectory.

[0087] S302, delete the best trajectory τ k,L Associated points, get the updated measurement point set

[0088] S303, and its corresponding L frame measurement points execute step 1 again to perform a new round of trajectory generation and pruning. Stop iteration when .

[0089] The cumulative value function of the generated trajectory is compared with the threshold. If it exceeds the threshold, the measurement point associated with the trajectory is declared as a moving point, otherwise it is classified as a stationary point. Specifically:

[0090] S401, the optimal trajectory τ k,L The corresponding cumulative value function F k,L Compared with the threshold γ, if F k,L is greater than γ, then determine τ k,L The associated points come from the moving target.

[0091]

[0092] Estimate the velocity of the moving target's trajectory, generate a velocity grid, estimate the target velocity in the velocity grid, and output the target's lateral and longitudinal velocities, specifically:

[0093] S501, divide the lateral velocity grid into The longitudinal velocity grid is divided into Can get I x ×I y Group velocity grid.

[0094] S502, for the trajectory τ k,L Associated measurement point of the first frame Calculate the distance between it and each set of velocity grids, and then use the value function z k,l Compare and get I x ×I y Ratio function of group velocity grid

[0095]

[0096] S503, for the trajectory τ k,L The L frame measurement points are summed for each group of velocity grids, and the velocity grid corresponding to the minimum value is found. Combine it with the vehicle speed v s =[v sx ,v sy ] T By taking the difference, we can get the velocity of the trajectory.

[0097]

[0098] In another embodiment, a simulation data experiment is conducted on the above technical solution. In this embodiment, five stationary targets and one moving target are set. Their initial position parameters are shown in the following table:

[0099] Table 1 Simulation target parameters

[0100]

[0101]

[0102] Other parameter settings:

[0103] Processing frame number L = 10

[0104] The time interval between two frames is T = 0.05

[0105] Variance of distance error

[0106] Variance of angular error

[0107] Variance of velocity error

[0108] First, verify the effect of the value function of formula 2 on the dynamic and static target judgment. Let the vehicle speed be v s =[40,0] T , set the fast and slow target speeds to v fast =[1.2,1.2] T and v slow =[0.6,0.6] T The distance diagram from the measurement point to the velocity profile is shown as follows: Figure 4 As shown in the figure, when the target moves, the scattering points deviate from the self-velocity distribution. At this time, the value function z is larger. It can be seen that the value function can distinguish between moving and static targets, and has a better effect on distinguishing moving targets with high speed.

[0109] Next, we verify that multi-frame processing can improve the judgment performance of moving targets compared to single-frame processing. Set the two vehicle speeds, fast and slow, to v s1 =[40,0] T and v s2 =[20,0] T , the moving angle of the moving target is 45°, the speed of the moving target is changed, and the Monte Carlo method is used to obtain the moving target judgment probability.

[0110] First, in the H0 scenario, the target speed is v = [0,0] T Then, in the H1 scenario, the absolute speed of the moving target is changed, the number of times it exceeds the decision threshold is calculated, and the decision probability is obtained. Figure 5 As shown in the figure, it can be seen that multi-frame processing can significantly improve the probability of moving target judgment and can effectively detect targets with very low speed.

[0111] Finally, estimate the target motion state. We know 10 frames of target data, including the measured speed and measured angle. According to step 4, we can estimate the horizontal and vertical speed of the target. Let v x The value range is 0 to 25, v y The value range is -5 to 5, and the discrete interval of both is 0.01. Let the vehicle speed v s =[20,0] T , the target velocity is v = [1.2, 1.2] T , the estimated speed of the final target is [1.22,1.21] T , approaching the actual speed of the target.

[0112] The above experiments demonstrate that the present invention can improve the performance of moving point target detection in dynamic environments, particularly for small moving targets, by accumulating value functions over multiple frames. Furthermore, when processing multi-frame data, kinematic constraints are used to limit the range of trajectory accumulation during the trajectory generation phase, and redundant trajectory information is quickly removed during the trajectory pruning phase. This effectively reduces computational complexity and ensures that computational delays do not affect the efficiency of moving target detection.

[0113] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or device comprising the element.

[0114] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is intended to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A method for determining target dynamic and static attributes based on multi-frame data, characterized in that: include: S1. Collect radar measurement data, initialize all measurement points in the first frame of radar measurement data, and calculate their value functions; S2. Match the measurement points of each frame in the radar measurement data with the accumulation range of the measurement points of the next frame according to the kinematic constraints, accumulate the value functions of the measurement points that meet the constraints, output the accumulated value functions, and obtain all estimated trajectories and their corresponding accumulated value functions; S3. Prune all estimated trajectories, retain the trajectory with the largest cumulative value function among the same measurement points, and delete the measurement points associated with the pruned retained trajectory to obtain an updated measurement point set. Perform steps S1-S2 again on the measurement point set until all measurement points of the last frame are deleted, and obtain the pruned estimated trajectory and its corresponding cumulative value function. S4. Performing a threshold comparison on the accumulated value function of the estimated trajectory. If the threshold is exceeded, defining the measurement point associated with the second estimated trajectory as a moving point; otherwise, classifying it as a stationary point. S5. Estimating the speed of the estimated trajectory of the moving target, generating a speed grid, estimating the target speed in the speed grid, outputting the horizontal and vertical speeds of the target, and determining whether the target is moving or still.

2. The method for determining target dynamic and static attributes based on multi-frame data according to claim 1, wherein: The collecting of radar measurement data, initializing all measurement points in the first frame of the radar measurement data, and calculating the value function thereof include: S101. Use a millimeter-wave radar to obtain L-frame radar measurement data of the surrounding environment, which is defined as: Represents the point index set of the l-th frame measurement data at the j-th iteration, and records the measurement data of the l-th frame as: The measurement data of the first frame is obtained by D l The kth measurement point is recorded as: in, Indicates the time when the measurement point is obtained, Respectively represent the measured relative distance, angle and Doppler frequency; S102, when l=1 frame of measurement data, initialize all measurement points, and for measurement point s k,1 , based on IMU to obtain the vehicle speed v s =[v sx ,v sy ] T , calculate the Doppler frequency of the measured angle and vehicle speed as Where λ is the wavelength, and the function to calculate its value is:

3. The method for determining target dynamic and static attributes based on multi-frame data according to claim 2, wherein: The method of matching the measurement points of each frame in the radar measurement data with the accumulation range of the measurement points of the next frame according to the kinematic constraints, accumulating the value functions of the measurement points that meet the constraints, and outputting the accumulated value functions to obtain all estimated trajectories and their corresponding accumulated value functions includes: S201. When 2≤l≤L, calculate s k,l-1 and D l Measurement points n,l The radial velocity v r,l and tangential velocity v t,l , and calculate their velocity errors and Expressed as: Combined k,l-1 The measurement point s associated with the previous frame m,l-2 , calculate s k,l-1 and D l Measurement points n,l Radial acceleration a r,l and tangential acceleration a t,l , and calculate their acceleration errors and Expressed as: S202, set the error weight β, combined with the speed error and Acceleration error and Calculate measurement points k,l-1 and s n,l and determine whether the speed and acceleration are less than the maximum speed v max and the maximum acceleration a max , expressed as: If less than s k,l-1 and s n,l The match is successful, otherwise it fails, where x + =max{x,0}; S203, record the measurement point index n matched in the first frame into the trajectory set M k,l If all measurement points fail to match, the index is recorded as 0, and when 2≤l≤L, the measurement point s k,l-1 While matching, k,l Each trajectory of ξ k,l Accumulate and get the accumulated value function F k,l , if the first frame does not have s k,l-1 Match the successful measurement point, take η as the accumulation parameter of l frame, and the accumulation value function F k,l Expressed as: F k,l =F k,l-1 +z n,l ,n∈ξ k,l Among them, the number of frames p that are not matched to points in a row does not exceed P.

4. The method for determining target dynamic and static attributes based on multi-frame data according to claim 3, wherein: The method of trimming all estimated trajectories, retaining the trajectory with the largest cumulative value function among the same measurement points, and deleting the measurement points associated with the pruned retained trajectory to obtain an updated measurement point set, and performing steps S1-S2 on the measurement point set again until all measurement points of the last frame are deleted, obtaining the pruned estimated trajectory and its corresponding cumulative value function, includes: S301, measuring point s k,1 The set of all trajectories M k,L Perform pruning, if M k,L Contains more than one trajectory, retaining the maximum cumulative value function F k,L The trajectory τ k,L Defined as s k,1 The best trajectory of S302, delete the best trajectory τ k,L Associated measurement points to obtain the updated measurement point set S303, and its corresponding L frame measurement point executes steps S1-S2 again until The iteration is stopped when , and the estimated trajectory after pruning and its corresponding accumulation value function are obtained.

5. The method for determining target dynamic and static attributes based on multi-frame data according to claim 4, wherein: The step of comparing the accumulated value function of the estimated trajectory with a threshold value, and defining the measurement point associated with the second estimated trajectory as a moving point if the threshold value is exceeded, and classifying the measurement point as a stationary point if the threshold value is exceeded, includes: S401, the optimal trajectory τ k,L The corresponding cumulative value function F k,L Compared with the threshold γ, if F k,L is greater than γ, then determine τ k,L The associated measurement points are from moving targets, and vice versa are classified as stationary points, which can be expressed as:

6. The method for determining target dynamic and static attributes based on multi-frame data according to claim 5, wherein: The method of performing velocity estimation on the estimated trajectory of the moving target, generating a velocity grid, estimating the target velocity in the velocity grid, outputting the horizontal and vertical velocities of the target, and determining whether the target is moving or still includes: S501, divide the lateral velocity grid into The longitudinal velocity grid is divided into Generate I x ×I y Group velocity grid; S502, the optimal trajectory τ k,L Associated measurement point of the first frame Calculate the distance between the measurement point and each set of velocity grids, and then use the value function z k,l Compare and get I x ×I y Ratio function of group velocity grid Expressed as: S503, for the optimal trajectory τ k,L The L frame measurement points are summed for each group of velocity grids, and the velocity grid corresponding to the minimum value is output. Combine it with the vehicle speed v s =[v sx ,v sy ] T By taking the difference, we can get the horizontal and vertical velocities of the trajectory, which can be expressed as:

Images