A photoelectric tracking feedforward control method for compensating for carrier translational disturbance

Abstract

The application provides an optical-electric tracking feedforward control method for compensating for carrier translation disturbance, which is used for accurately calculating the angular velocity control amount required by optical-electric tracking; the carrier is used for carrying an optical-electric turntable, an IMU and an image sensor; the optical-electric turntable, the IMU and the carrier are fixed with each other; the optical-electric tracking refers to adjusting the heading angular velocity and the pitching angular velocity of the image sensor by controlling the optical-electric turntable, so that the image sensor can point to the target in real time; the tracking control of the heading angular velocity and the pitching angular velocity is independent of each other; the application has the following advantages: the target distance and the carrier translation velocity measured by the IMU are used to estimate the carrier translation disturbance compensation amount; compared with the traditional method which only considers the carrier rotation disturbance, a more accurate error model is established; the translation disturbance compensation amount is introduced into the Kalman filter as a system control vector, the coupling influence of the target motion state estimation is eliminated, and more accurate target motion state estimation is realized.

Description

Technical Field

[0001] This invention relates to the field of photoelectric tracking and servo control technology, specifically to a photoelectric tracking feedforward control method suitable for compensating for translational disturbances of a carrier in close-range, high-dynamic scenarios of unmanned platforms. Background Technology

[0002] Photoelectric tracking refers to the process of continuously and precisely capturing and pointing a target in real time by controlling a photoelectric turntable with its onboard image sensor. The key to this process lies in whether the control method of the photoelectric turntable can accurately calculate the required control variables. For photoelectric tracking systems, the control methods mainly include line-of-sight stabilization control and tracking control. With the rapid development of intelligent unmanned platforms, photoelectric tracking systems based on unmanned platforms are widely used in complex environments such as urban streets and jungle off-road terrain. Compared with traditional long-distance tracking scenarios, these close-range, highly dynamic environments are characterized by high target mobility and complex and variable platform motion states, placing higher demands on the instantaneous response capability of photoelectric tracking systems. Given a fixed photoelectric tracking platform hardware, advanced tracking control methods are an effective means to improve the system's tracking accuracy.

[0003] In existing moving-base electro-optical tracking systems, line-of-sight stabilization technology mostly employs inertial measurement unit (IMU)-based inertial stabilization. This method primarily uses gyroscope feedback to suppress rotational disturbances, often neglecting the coupling interference (translational disturbance) of the vehicle's translational motion on the line of sight. While translational disturbances have a relatively small impact in long-range tracking, in close-range scenarios, vehicle translation generates significant line-of-sight angular velocity components, directly affecting tracking accuracy.

[0004] In target tracking control, tracking accuracy is primarily limited by the sampling and processing delays of image sensors. To address this, existing techniques commonly employ Kalman predictive filtering to compensate for delays by predicting the target state, thereby improving tracking accuracy. This method relies on an accurate perturbation error model, but existing perturbation error models typically ignore the translational perturbation error of the carrier, resulting in poor tracking accuracy in high-dynamic scenarios.

[0005] In close-range, high-dynamic scenarios, the impact of vehicle translational disturbances on tracking accuracy has become a significant factor that cannot be ignored. Errors introduced by translational disturbances are coupled into the estimation of the target state and are further amplified by observation delays, leading to a decrease in tracking accuracy. Under these conditions, traditional tracking control methods are no longer sufficient to meet the tracking accuracy requirements of such scenarios. Summary of the Invention

[0006] To address the above problems, this invention proposes a photoelectric tracking feedforward control method that compensates for carrier translational disturbances. This method is based on Kalman predictive filtering and combines IMU estimation of the carrier's translational velocity to compensate for the impact of carrier translational motion on target state observation. This solves the problem of poor tracking accuracy and stability caused by traditional predictive filtering methods neglecting carrier translational disturbances.

[0007] The technical solution adopted in this invention is as follows:

[0008] A feedforward control method for photoelectric tracking to compensate for translational disturbances of the carrier is disclosed, used to accurately calculate the angular velocity control quantities required for photoelectric tracking. The carrier supports a photoelectric turntable, an IMU (Integrated Device Unit), and an image sensor, with the turntable, IMU, and carrier rigidly connected to each other. Photoelectric tracking refers to adjusting the yaw and pitch angular velocities of the image sensor's line of sight (center axis of the field of view) by controlling the turntable, ensuring it points towards the target in real time. The tracking control of the yaw and pitch angular velocities is independent of each other. The translational disturbance of the carrier refers to the interference of the carrier's velocity components along the X, Y, and Z axes in the carrier coordinate system on the estimated target yaw (or pitch) angular velocity. The feedforward control method, taking the yaw angular velocity tracking control to compensate for translational disturbances along the carrier's x-axis as an example, comprises the following steps:

[0009] S1. Establish and align the carrier coordinate system and the IMU coordinate system;

[0010] S2. Obtain sensor output information for use in the calculation of various state variables in subsequent steps;

[0011] S3. The total angular velocity control quantity required to achieve photoelectric tracking is divided into three parts: the target azimuth angular velocity, the angular velocity to compensate for the translational disturbance of the carrier, and the angular velocity to suppress the line-of-sight error. The specific steps are as follows:

[0012] S3.1 Establish a Kalman filter to compensate for the translational disturbance of the carrier and complete the initialization;

[0013] S3.2. Implement Kalman prediction filtering to predict the target azimuth angular velocity at time k;

[0014] S3.3 Calculate the angular velocity of the compensating carrier translational disturbance at time k using the predicted value of the target motion state at time k;

[0015] S3.4 Calculate the angular velocity at time k that suppresses the target's line-of-sight error;

[0016] S3.5 Obtain the total angular velocity control value at time k; combine it with the target azimuth angular velocity obtained in step S3.2.3. The angular velocity of the translational disturbance of the compensated carrier obtained in step S3.3 The angular velocity obtained in step S3.4.2 to suppress target line-of-sight error Calculate the total angular velocity control quantity required to ultimately achieve photoelectric tracking. :

[0017] ;

[0018] S4. The total angular velocity control value obtained in step S3.5 The image is input to the encoder motor control unit, and after the new image is sampled, steps S2 to S4 are repeated to achieve photoelectric tracking.

[0019] Compared with the prior art, the present invention has the following advantages:

[0020] 1. This invention estimates the translational disturbance compensation amount of the carrier by using the target distance and the carrier translational velocity measured by the IMU. Compared with the traditional method that only considers the carrier rotational disturbance, it establishes a more accurate error model.

[0021] 2. This invention introduces the translational disturbance compensation amount as a system control vector into the Kalman filter, eliminating its coupling effect on the target motion state estimation and achieving more accurate target motion state estimation;

[0022] 3. Based on the established error model, this invention calculates the required angular velocity control quantity into three parts: the target azimuth angular velocity, the angular velocity to compensate for the translational disturbance of the carrier, and the angular velocity to suppress the line-of-sight error, and finally achieves real-time tracking. Attached Figure Description

[0023] To better understand and describe the technical solution of the present invention, further explanation is provided below in conjunction with the accompanying drawings:

[0024] Figure 1 This is a schematic diagram of translational disturbance compensation;

[0025] Figure 2 Flowchart of the feedforward control method for compensating for translational disturbances;

[0026] Figure 3 This describes the specific setup of the carrier and target vehicle for the load photoelectric tracking system in this embodiment;

[0027] Figure 4 The target line-of-sight error time series obtained by the pure PID method and the method of the present invention are compared during the three cycles of the reciprocating motion of the carrier trolley.

[0028] Figure 5 The target StMTE curves obtained by the pure PID method and the method of this invention are compared during the three cycles of the reciprocating motion of the carrier vehicle. Detailed Implementation

[0029] To make the technical solution and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and a specific embodiment. It should be understood that the specific embodiment described herein is only for explaining the present invention and is not intended to limit the present invention.

[0030] See Figure 1 The flow chart of the feedforward tracking control method in this embodiment is specifically divided into the following steps:

[0031] S1. Establish and align the carrier coordinate system and the IMU coordinate system;

[0032] S2. Obtain sensor output information for calculation of various state variables in subsequent steps. Different sensors have different output frequencies. In order to align sensor information obtained at different times, the current time is defined as time k, the previous time is defined as time k-1, the time when the target is first detected by the image sensor is time 1, and the current time is reset to time 0 whenever the target is lost.

[0033] S3. The total angular velocity control quantity required to achieve photoelectric tracking is divided into three parts: the target azimuth angular velocity, the angular velocity to compensate for the translational disturbance of the carrier, and the angular velocity to suppress the line-of-sight error. The specific steps are as follows:

[0034] S3.1 Establish a Kalman filter to compensate for the translational disturbance of the carrier and complete the initialization;

[0035] S3.2. Implement Kalman prediction filtering to predict the target azimuth angular velocity at time k;

[0036] S3.3 Calculate the angular velocity of the compensating carrier translational disturbance at time k using the predicted value of the target motion state at time k;

[0037] S3.4 Calculate the angular velocity at time k that suppresses the target's line-of-sight error;

[0038] S3.5 Obtain the total angular velocity control value at time k; combine it with the target azimuth angular velocity obtained in step S3.2.3. The angular velocity of the translational disturbance of the compensated carrier obtained in step S3.3 The angular velocity obtained in step S3.4.2 to suppress target line-of-sight error Calculate the total angular velocity control quantity required to ultimately achieve photoelectric tracking. :

[0039] ;

[0040] S4. The total angular velocity control value obtained in step S3.5 The image is input to the encoder motor control unit, and after the new image is sampled, steps S2 to S4 are repeated to achieve photoelectric tracking.

[0041] In S1, the steps for establishing and aligning the carrier coordinate system and the IMU coordinate system are as follows: For the plane where the heading angle of the line of sight controlled by the photoelectric turntable is located, take the front-right direction of the turntable as the xy axis to establish a two-dimensional Cartesian coordinate system b, that is, the carrier coordinate system. At the same time, install and calibrate the IMU to align the XY axis of the IMU coordinate system with the b system.

[0042] In S2, the specific steps for obtaining sensor information are as follows:

[0043] S2.1. Utilize IMU output information to perform inertial navigation pose calculation, and obtain the velocity components of the carrier's x-axis at time k. , where the subscript P indicates that the value is a state variable of the carrier;

[0044] S2.2. The angle of the line of sight relative to the x-axis of the carrier at time k-1 is obtained through the output of the photoelectric turntable heading angle encoder, i.e., the line of sight heading angle. The subscript LOS indicates that the value is a state variable of the line of sight.

[0045] S2.3 Detect the angular difference between the target and the line of sight using an image sensor, i.e., the target line of sight error. Since image sensor sampling and processing are time-consuming, the measurement result obtained at the current time k is used as the target line-of-sight error at time k-1. ;

[0046] S2.4 Calculate the target azimuth at time k-1 It is the sum of the line-of-sight error and the line-of-sight heading angle of the target at time k-1:

[0047] ;

[0048] S2.5. While detecting the target using an image sensor, the target distance at time k-1 is obtained using a lidar. ;

[0049] In S3.1, a Kalman filter for compensating for translational disturbances of the carrier is established and initialized, specifically including the following steps:

[0050] S3.1.1 To facilitate control implementation, polar coordinates are used in the b-frame to describe the target's motion state, including the target's azimuth angle relative to the origin of the b-frame. azimuth angular velocity azimuth acceleration ,distance radial velocity Radial acceleration ;

[0051] S3.1.2. Establish a Kalman filter based on the traditional tracking control method. Use the CA motion model, assuming the target azimuth angular velocity... and radial velocity The azimuth and radial acceleration derivatives are treated as zero-mean Gaussian white noise, exhibiting uniform variation. , Determine the noise variance; determine the system state vector of the Kalman filter. for:

[0052] ,

[0053] The system state transition matrix is:

[0054] ,

[0055] T is the sampling period of the image sensor;

[0056] The process noise covariance of the system is:

[0057] ,in ;

[0058] The system's observation vector is:

[0059] ,

[0060] The system observation transfer equation is:

[0061] ,

[0062] The measurement noise covariance matrix is:

[0063] ,in and These are the noise variances for the target azimuth and range, respectively.

[0064] S3.1.3 Calculate the carrier translational disturbance compensation amount and introduce it as the system control vector into the Kalman filter; the carrier translational disturbance compensation amount specifically includes compensating for the angular velocity and radial velocity of the carrier translational disturbance, which are the azimuth angular velocity and radial velocity of the carrier relative to the target, respectively, which will affect the observed target azimuth angle and distance; calculate the angular velocity of the compensated carrier translational disturbance. radial velocity for:

[0065] ,

[0066] ,

[0067] in The compensation amount is used as the system control vector for the X-axis velocity component of the carrier. Introducing a Kalman filter:

[0068] ,

[0069] The system's control input matrix is:

[0070] ,

[0071] The filtering equations of the established discrete Kalman filter are described as follows:

[0072] ,

[0073] The superscripts - and + in the upper right corner represent the prior and posterior estimates, respectively, and the ^ symbol in the state variables indicates that it is an estimate. Let be the mean square error matrix of the system state vector. Let be the system noise covariance matrix. To observe the noise covariance matrix, Here is the Kalman filter gain matrix, and the subscript k indicates time k;

[0074] S3.1.4 Initialize the Kalman filter; Initialize the posterior estimate of the state vector of the Kalman filter system at time -1. System control vector Prior estimation of the state vector of the Kalman filter system at time 0. ; Initialize the mean square error matrix of the system state vector Set the system noise covariance matrix. Covariance matrix of observation noise ; and The value of is determined by the specific parameters of the system; after initialization, the initial state is substituted into the Kalman filter equation, and the Kalman prediction filter iteration is implemented according to the process described in step S3.2.

[0075] In S3.2, Kalman prediction filtering is implemented to predict the target azimuth angular velocity at time k; the filtering iteration process is as follows:

[0076] S3.2.1 Calculate the optimal posterior estimate of the target motion state at time k-1; using the prior estimate of the Kalman filter system state at time k-1, combined with step S2, obtain the observation vector at time k-1. :

[0077] ,

[0078] Prior estimation of the system state at time k-1 based on the measurement update equation of Kalman filtering After correction, the optimal posterior estimate of the target's motion state at time k-1 is obtained. :

[0079] ,

[0080] ,

[0081] S3.2.2 Calculate the best a priori estimate of the target motion state at time k; using the best a priori estimate of the target motion state at time k-1 obtained in step S3.2.1. As the initial value, combined with the system control vector at time k-1 : Using Kalman filtering, the state transition equation Perform a one-step prediction to obtain the highest priority estimate of the target's motion state at time k. :

[0082] ,

[0083] ;

[0084] S3.2.3 Obtain the predicted values ​​of the target azimuth angular velocity, target azimuth angle, and target distance at time k; use the highest priority prior art estimate of the target motion state at time k obtained in step S3.2.2. As the predicted value of the target's motion state at that moment, the predicted result of the target's azimuth angular velocity at time k is read from it. Simultaneously, the target azimuth angle at time k is obtained from the predicted target motion state. Distance to target These two predicted values ​​will be used in subsequent step S3.3 to calculate the angular velocity and radial velocity of the compensated carrier translational disturbance at time k;

[0085] S3.2.4 Update the Kalman filter error covariance matrix Kalman gain This includes the following steps:

[0086] ,

[0087] ,

[0088] .

[0089] In S3.3, the predicted value of the target's motion state at time k is used to calculate the angular velocity of the compensated carrier translational disturbance at time k, as follows:

[0090] Combined with the carrier velocity at time k obtained in step S2 The target azimuth angle at time k obtained in step S3.2.3 Target distance Calculate the angular velocity of the compensated carrier translational disturbance at time k. and radial velocity :

[0091] ,

[0092] ,

[0093] The calculated result is also the system control vector of the Kalman filter at time k. :

[0094] .

[0095] In S3.4, the angular velocity that suppresses the target line-of-sight error at time k is calculated, which specifically includes the following steps:

[0096] S3.4.1 Predict the target line-of-sight error at time k; Step S2 yielded the target line-of-sight error at time k-1. However, real-time tracking requires suppressing the target line-of-sight error at time k, therefore it is necessary to predict the target line-of-sight error at time k; using the target line-of-sight error at time k-1 obtained in step S2... The target azimuth angular velocity at time k-1 obtained by filtering in step S3.2 and the target azimuth angular velocity at time k And the angular velocity control quantity output by the PID controller at time k-1. Predict the target line-of-sight error at time k for:

[0097] ;

[0098] S3.4.2 Using the predicted target line-of-sight error at time k The PID controller calculates the angular velocity at time k that suppresses the target line-of-sight error. :

[0099] ,

[0100] in , , These are the proportional coefficient, integral coefficient, and derivative coefficient of the PID controller, respectively, and their specific values ​​are obtained through debugging.

[0101] According to step S1 of the present invention, see Figure 2The carrier coordinate system and the IMU coordinate system are established and aligned. The photoelectric turntable is fixed to the carrier. A two-dimensional Cartesian coordinate system b, i.e., the carrier coordinate system, is established with the rotation center of the photoelectric turntable as the origin and the plane containing the heading angle controlled by the turntable. Simultaneously, the IMU is installed and calibrated, aligning the X and Y axes of the IMU coordinate system with the b system. The geometric centers of the carrier and the target are respectively... , The carrier linear velocity is The direction coincides with the polar axis of the b-system, and the line of sight points to... ;Target Compared to a turntable The Euclidean distance is The angle relative to the line of sight is the line of sight error. The azimuth angle in the b system is To facilitate turntable control, polar coordinates are used in the b-frame to describe the target's motion state, including the target's azimuth angle relative to the origin of the b-frame. azimuth angular velocity azimuth acceleration ,distance radial velocity Radial acceleration .

[0102] See Figure 3 This describes the specific setup of the carrier and target vehicle carrying the photoelectric tracking system in this embodiment. The carrier vehicle and target vehicle are placed diagonally on two parallel, 3m apart, 3m long straight tracks. Before the experiment begins, the line of sight of the photoelectric tracking system carried by the carrier vehicle is aligned with the target vehicle. After the experiment begins, they move towards each other and back and forth at a speed of 1m / s. In this embodiment, the sampling period of the image sensor is 1 / 60s, and the initial system noise covariance matrix of the Kalman filter is... Observation noise covariance matrix The proportional gain of the PID controller Integral coefficient and differential coefficients .

[0103] According to step S2 of the present invention, let the start time of the experiment be time 1, and assume that the experiment has progressed to time k. The specific implementation steps of this embodiment are described below using time k as an example:

[0104] According to step S2 of the present invention, the x-axis velocity component of the carrier at time k is obtained. The heading angle of the line of sight at time k-1 and the target line-of-sight error at time k-1 Target azimuth Distance to target .

[0105] According to step S3 of the present invention, the total angular velocity control amount required to achieve photoelectric tracking at time k is calculated. .

[0106] According to step S4 of the present invention, the obtained total angular velocity control quantity is... The data is input to the encoder motor control unit, and after the new image is sampled, steps S2 to S4 are repeated to achieve photoelectric tracking.

[0107] This embodiment includes a control group controlled by a pure PID method. The main difference lies in: obtaining the target line-of-sight error at time k-1 according to step S2 of this invention. Then, directly substitute it into step S3.4.2, and use the same PID controller parameters to calculate the angular velocity control quantity at time k. :

[0108]

[0109] Then skip the other steps and obtain the angular velocity control quantity. The data is input to the encoder motor control unit, and the above process is repeated after a new image is sampled.

[0110] See Figure 4 The target line-of-sight error time series is obtained by the pure PID method and the method of this invention during the three cycles of the carrier vehicle's reciprocating motion. It can be seen that the method proposed in this invention significantly suppresses the target line-of-sight error during the tracking process and improves the tracking accuracy. The root mean square error (RMSE) of the entire target line-of-sight error series is used to measure the overall tracking accuracy of the system, and its calculation formula is as follows:

[0111]

[0112] in The length of the error sequence. The first in the error sequence The RMSE of the pure PID method is 1.6843°, while the RMSE of the method of this invention is 0.6050°, an improvement of 64.08%.

[0113] To more precisely quantify the instantaneous response characteristics and stability of the method in the time domain, a short-time mean tracking error (StMTE) was calculated as an evaluation index based on the target line-of-sight error time series. The calculation formula is as follows:

[0114]

[0115] in, The sampling frame rate of the image sensor. Target line of sight error sequence window The first Each value. In this embodiment... .

[0116] See Figure 5 The target StMTE curves obtained by the pure PID method and the method of this invention were compared during the three cycles of the carrier's reciprocating motion. The StMTE curve of the pure PID control exhibited regular pulse-like peaks. After compensating for the translational disturbance of the carrier using the method of this invention, the error peaks were effectively suppressed. The StMTE peak value of the pure PID method was 2.4483°, while the StMTE peak value of the method of this invention was 0.7556°, an improvement of 69.14%.

Claims

1. A method of feedforward control of optoelectronic tracking to compensate for translational disturbances of a carrier, characterized in that, The method comprises the following steps: S1, establishing and aligning the carrier coordinate system and the IMU coordinate system; S2, obtaining sensor output information for calculating state variables in subsequent steps; S3, dividing the total angular velocity control quantity required for realizing photoelectric tracking into target azimuth angular velocity, angular velocity for compensating for carrier translational disturbance, and angular velocity for suppressing the boresight error, and calculating the three parts, the specific steps being as follows: S3.1, establishing a Kalman filter for compensating for carrier translational disturbance, and completing initialization; the specific steps including the following: S3.1.

1. For ease of implementation control, use polar coordinates to describe the target motion state in b-frame, including the azimuth angle of the target relative to the origin of the b-frame azimuth angle velocity azimuth angle acceleration distance radial velocity radial acceleration ; S3.1.2, Establish Kalman filter based on traditional tracking control method; use CA motion model, assume target azimuth angle velocity and radial velocity Uniformly change, take the azimuth angle acceleration derivative and the radial acceleration derivative as zero-mean Gaussian white noise , Noise variance; determine the system state vector of Kalman filter is: ， The system state transition matrix is: ， T is the image sensor sampling period; The system process noise covariance is: wherein ; The system observation vector is: ， The system observation transition equation is: ， The measurement noise covariance matrix is: where and are the noise variances of the target azimuth and range, respectively; S3.1.3, calculate the carrier translational disturbance compensation quantity and introduce it into the Kalman filter as the system control vector; the carrier translational disturbance compensation quantity specifically includes the angular velocity and the radial velocity compensating for the carrier translational disturbance, which are the azimuth angle velocity and the radial velocity of the carrier relative to the target motion, and will affect the observed target azimuth angle and distance; the angular velocity compensating for the carrier translational disturbance is calculated as follows: the radial velocity compensating for the carrier translational disturbance is calculated as follows: ​ ， ， wherein is the carrier X-axis velocity component, and the compensation amount is the system control vector Introduce the Kalman filter: ， The system control input matrix is: ， The filtering equation of the established discrete Kalman filter is described as follows: ， where the upper right corner superscripts - and + represent the prior estimate and the posterior estimate respectively, and the ^ symbol in the state quantity indicates that it is an estimated value, is the mean square error matrix of the system state vector, is the system noise covariance matrix, is the observation noise covariance matrix, is the Kalman filter gain matrix, and the subscript k represents the kth moment; S3.1.4, initializing Kalman filter; initializing the posterior estimation of the system state vector at time point -1 , system control vector , represents a 6-row 1-column 0 vector matrix, represents a 2-row 1-column 0 vector matrix; initializing the prior estimation of the system state vector at time point 0 ; initializing the mean square deviation matrix of the system state vector ; setting the system noise covariance matrix and the observation noise covariance matrix ; and are determined by the specific parameters of the system; after the initialization is completed, the initial state is substituted into the Kalman filter equation, and the Kalman prediction filter iteration is implemented according to the process described in step S3.2; S3.2, implementing Kalman prediction filtering to predict the target azimuth angular velocity at time k; the implementation process of filtering iteration is as follows: S3.2.1, calculate the optimal posterior estimation of the target motion state at time k-1; use the prior estimation of the system state of the Kalman filter at time k-1, in combination with step S2 to obtain the observation vector at time k-1 : ， The prior estimate of the system state at time k-1 is updated according to the measurement update equation of the Kalman filter The optimal posterior estimate of the target motion state at time k-1 is obtained by correction : ， ， S3.2.2, compute the optimal prior estimate of the target motion state at time k; using the optimal posterior estimate of the target motion state at time k-1 obtained in step S3.2.1 as an initial value, combine the system control vector at time k-1 , use the Kalman filter state transition equation to make a one-step prediction to obtain the optimal prior estimate of the target motion state at time k :​ ， ； S3.2.3, obtain the predicted values of the target azimuthal angular velocity, the target azimuthal angle and the target range at time k; and obtain the optimal prior estimation of the target motion state at time k from step S3.2.2 As the predicted values of the target motion state at this time, the predicted result of the target azimuthal angular velocity at time k is read therefrom ; at the same time, the target azimuthal angle at time k is obtained from the predicted target motion state and the target range These two predicted values will be used in the subsequent step S3.3 to calculate the angular velocity and the radial velocity of the compensation carrier for the disturbance in the tangential direction at time k; S3.2.4, updating the Kalman filter error covariance matrix , the Kalman gain comprising the steps of: ， ， ； S3.3, using the predicted value of the target motion state at time k to calculate the angular velocity for compensating for carrier translational disturbance at time k; the specific steps being as follows: the carrier movement velocity at time k obtained in step S2 and the target azimuth at time k obtained in step S3.2.3 , the target distance the angular velocity at time k for compensating the carrier's translational disturbance and the radial velocity : ， ， The calculation result is also the system control vector of Kalman filter at time k : ； S3.4, calculating the angular velocity for suppressing the target boresight error at time k; the specific steps including the following: S3.4.1, predict the target visual axis error at time k; the target visual axis error at time k-1 is obtained in step S2 But the target visual axis error at time k needs to be suppressed for real-time tracking, so the target visual axis error at time k needs to be predicted; the target visual axis error at time k-1 obtained in step S2 is used , the target azimuth angle velocity at time k-1 filtered in step S3.2 and the target azimuth angle velocity at time k , and the angular velocity control amount output by the PID controller itself at time k-1 , the target visual axis error at time k is predicted : T is the image sensor sampling period; S3.4.2, Using predicted k-time target gaze error and PID controller calculates angular velocity of k-time suppression target gaze error : ， wherein , , P, I and D are respectively the proportional coefficient, integral coefficient and differential coefficient of the PID controller, and the specific values thereof are obtained through debugging. S3.5, obtain the total angular velocity control quantity at time k; combine the target azimuth angular velocity obtained in step S3.2.3 , the angular velocity compensating for the translational disturbance of the carrier obtained in step S3.3 , and the angular velocity suppressing the target line-of-sight error obtained in step S3.4.2 , calculate the total angular velocity control quantity required for finally implementing photoelectric tracking : ； S4, the total angular velocity control quantity obtained in step S3.5 The input is to the encoder motor control unit, and after a new image sample is completed, steps S2 to S4 are repeatedly executed to achieve photoelectric tracking.

2. The method of claim 1, wherein the method further comprises: determining a position of the object in the object coordinate system; and determining a position of the object in the camera coordinate system. In S1, the steps of establishing and aligning the carrier coordinate system and the IMU coordinate system are as follows: for the plane where the boresight heading angle controlled by the photoelectric turntable is located, a two-dimensional Cartesian coordinate system b system, i.e. the carrier coordinate system, is established by taking the front-right direction of the turntable as the x-y axis, and the IMU is installed and calibrated to align the IMU coordinate system X-Y axis with the b system.

3. The photoelectric tracking feedforward control method for compensating for translational disturbances of a carrier according to claim 1, characterized in that: In S2, the specific steps of obtaining sensor information are as follows: S2.1, implement the inertial navigation pose solution by using the IMU output information, and obtain the velocity component of the carrier x axis at time k wherein the subscript P indicates that the value is a state quantity of the carrier; S2.2, the angle of the line-of-sight axis relative to the carrier x-axis at time k-1 is obtained from the output of the photoelectric turntable heading angle encoder, i.e. the line-of-sight heading angle where the subscript LOS indicates that the value is a state quantity of the line-of-sight axis; S2.3, detecting the angle difference of the target relative to the visual axis by the image sensor, i.e. target visual axis error ; due to the high time consumption of image sensor sampling and processing, the measurement result obtained at the current k moment is taken as the target visual axis error at the k-1 moment ; S2.4, calculate the target azimuth angle at time k-1 which is the sum of the boresight error of the target at time k-1 and the boresight heading angle: ； S2.5, obtaining the target distance at time k-1 by laser radar while detecting the target by the image sensor .

Images