Intelligent guide control method for anesthesia bomb

Abstract

The invention provides an intelligent guide control method for an anesthesia bomb, and relates to the technical field of information, and the method comprises the steps: obtaining a real-time motion parameter of a target; predicting a motion track of the target based on the obtained target motion parameters; calculating the optimal launching parameters of the anesthesia bomb based on the predicted motion trail; launching an anesthesia bomb; and tracking the motion states of the anesthesia bomb and the target. The flight attitude of the anesthetic bomb is dynamically adjusted according to the tracked motion states of the anesthetic bomb and the target, accurate intersection of the anesthetic bomb and the target is achieved, a proportional integral differential controller is established, target tracking data are processed in real time, the precision and reliability of the tracking data are improved, the anti-interference capability is enhanced, and the tracking stability is ensured. The precise striking on the uncertain moving target is realized; target classification and recognition are achieved by obtaining feature information such as the size, the shape and the texture of a target, combining with a feature template of a target type and adopting a support vector machine classification model.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to an intelligent guidance control method for anesthetic bullets. Background Art

[0002] In the intelligent guidance and control system of tranquilizer bullets, how to achieve real-time tracking and prediction of the target is a key technical issue. Since the target may be in a state of constant motion and changing posture, the traditional fixed-trajectory launch method cannot ensure the hit accuracy of the tranquilizer bullet. In order to improve the efficiency of anesthesia, the system needs to be able to dynamically adjust the launch angle and trajectory of the tranquilizer bullet according to the real-time motion information of the target. This requires the system to have high-precision target tracking and motion prediction capabilities, be able to obtain parameters such as the target's position, velocity, acceleration in real time, and calculate the optimal launch angle and trajectory curve based on these parameters. At the same time, since the target's motion trajectory may be random and uncertain, the system also needs to have a certain degree of adaptive capability and be able to modify the launch strategy in real time according to changes in the target's motion state. In addition, in complex environments, the target may be obscured by obstacles or interfered with. How to maintain the continuity and stability of tracking in this situation is also a technical problem that needs to be solved urgently. Summary of the Invention

[0003] The present invention provides an intelligent guidance control method for anesthetic bullets, which mainly includes: The target's real-time position, velocity, and acceleration parameters are acquired through comprehensive multi-sensor information fusion. The target's random and uncertain motion state is estimated and predicted using the Kalman filter algorithm to obtain the target's motion trajectory over a period of time. Based on the predicted target trajectory, a multi-objective optimization algorithm is used to calculate the optimal launch angle and trajectory parameters of the tranquilizer bullet, ensuring precise intersect between the bullet and the target in space and time. Based on the optimal launch angle and initial velocity parameters obtained through genetic algorithm optimization, the launch device is precisely adjusted and calibrated to ensure that the actual launch parameters of the tranquilizer bullet are consistent with the calculated optimal parameters. The flight trajectory of the tranquilizer bullet is continuously tracked after launch. At the same time, the PID control algorithm is used to dynamically adjust the flight attitude of the tranquilizer bullet based on the predicted target motion trajectory, ultimately achieving precise spatial and temporal rendezvous between the tranquilizer bullet and the target. During the launch of the tranquilizer bullet, the actual motion state of the target is continuously tracked. If the actual motion trajectory of the target deviates from the predicted trajectory, the trajectory of the tranquilizer bullet is adjusted in time. The trajectory is corrected online through servo control to ensure continuous intersection between the tranquilizer bullet and the target. In view of the randomness and uncertainty of the target's motion, the guidance of the tranquilizer bullet is optimized in real time using adaptive control. By dynamically adjusting the parameters of the guidance rate, the guidance can adapt to the changes in the target's motion state. A motion model is established based on the initial state of the anesthetic bullet launch and the initial motion state of the target. A visual sensor is used to continuously track and obtain the target's real-time position, velocity, and acceleration information. The extended Kalman filter algorithm is used to predict the target's motion and obtain its future motion trend. A fuzzy adaptive PID control algorithm is used to dynamically adjust the guidance rate parameters to improve the guidance accuracy and robustness of the anesthetic bullet. When the target is temporarily obstructed by an obstacle, the system uses historical data and prior knowledge of the target's motion to predict the target's motion state through motion model extrapolation, maintaining tracking continuity. At the same time, it obtains the position and shape information of the obstacle and calculates the optimal flight path for the tranquilizer bullet to bypass the obstacle and reach the target through path planning, thus avoiding the obstacle and achieving rendezvous with the target. Based on the 3D point cloud data of obstacles acquired by LiDAR scanning, the DBSCAN clustering and segmentation algorithm is used to extract the obstacle's position and shape information. When the target reappears after being obscured, the target is recaptured using the target detection algorithm, and the flight path of the tranquilizer bullet is corrected based on the actual position information to ensure the completion of the tracking and interception mission. Establish a proportional-integral-differential controller to process target tracking data in real time, improve the accuracy and reliability of tracking data, enhance anti-interference capabilities, ensure tracking stability, and achieve precise strikes on uncertain moving targets; By acquiring characteristic information such as target size, shape, and texture, combined with the characteristic template of the target type, and adopting the support vector machine classification model, target classification and recognition are achieved. The recognition results are integrated with the Kalman filter and neural network prediction results to obtain the target comprehensive state estimation, thereby improving the targeting and accuracy of tracking.

[0004] As a preferred embodiment, the intelligent guidance and control method for an anesthetic bullet comprises: obtaining real-time motion parameters of a target; predicting the motion trajectory of the target based on the obtained target motion parameters; calculating optimal launch parameters of the anesthetic bullet based on the predicted motion trajectory; launching the anesthetic bullet; tracking the motion states of the anesthetic bullet and the target; and dynamically adjusting the flight attitude of the anesthetic bullet based on the tracked motion states of the anesthetic bullet and the target to achieve precise intersection between the anesthetic bullet and the target, including: Based on the real-time position, velocity, and acceleration information of the target acquired by multiple sensors, the target motion state is filtered using the Kalman filter algorithm to obtain a filtered target motion state estimate. If the randomness and uncertainty of the target motion state are large, the process noise covariance matrix of the Kalman filter algorithm is increased to improve the adaptive ability of the filter algorithm and obtain a more accurate target motion state estimate. The target motion state estimate obtained by the Kalman filter is used as input, and the motion trajectory of the target in the future is predicted using a uniform acceleration motion model to obtain predicted target position, velocity, acceleration and other motion trajectory parameters. Based on the predicted target motion trajectory parameters, the flight trajectory of the tranquilizer bullet at different launch angles and initial velocities is calculated using a modified Euler ballistic model to obtain multiple sets of trajectory parameters such as the flight time and landing point of the tranquilizer bullet. A genetic algorithm is used to optimize the launch angle and initial velocity parameters of multiple groups of anesthetic bullets. By setting the minimum deviation between the anesthetic bullet landing point and the target position as the objective function, and considering constraints such as the anesthetic bullet range and flight time, the optimal anesthetic bullet launch angle and initial velocity parameters are obtained, so that they are best matched with the predicted target motion trajectory in space and time. According to the optimal anesthetic bullet launch angle and initial velocity parameters optimized by the genetic algorithm, the anesthetic bullet launch device is controlled to perform precise adjustment and calibration to ensure that the actual launch parameters of the anesthetic bullet are consistent with the calculated optimal parameters. After the anesthetic bullet is launched, its flight trajectory is continuously tracked and compared with the target motion trajectory predicted in the second step in real time. According to the deviation information between the two, the flight attitude and trajectory of the anesthetic bullet are dynamically adjusted through the PID control algorithm, ultimately achieving precise intersection with the target.

[0005] As a preferred solution, the intelligent guidance and control method for anesthetic bullets, wherein: the acquisition of the real-time motion parameters of the target includes: using multi-sensor fusion technology to obtain the real-time position, velocity and acceleration information of the target, and estimating and predicting the target motion state through the Kalman filter algorithm to obtain the motion trajectory of the target in the future period of time, including: According to the optimal launch angle and initial velocity parameters obtained by genetic algorithm optimization, the launch angle and initial velocity are precisely set by controlling the angle adjustment mechanism and velocity control unit of the launch device; a high-precision angle sensor and velocity sensor are used to collect the actual launch angle and initial velocity of the launch device in real time, and the collected actual parameters are compared with the optimal parameters. If the deviation exceeds the preset threshold, the PID control algorithm is triggered, and the launch angle and initial velocity are dynamically adjusted by adjusting the motor speed of the angle adjustment mechanism and the throttle opening of the velocity control unit until the deviation between the actual parameters and the optimal parameters is within the allowable range; after the anesthetic bullet is launched, its flight trajectory is continuously tracked by a high-speed camera, and the multi-view image information of the camera is used to calculate the real-time three-dimensional coordinates of the anesthetic bullet in space through the principle of triangulation, and then the velocity vector is calculated by the rate of change of the coordinates, and the acceleration vector is calculated by the rate of change of the velocity, thereby obtaining the state parameters such as the real-time spatial position, velocity and acceleration of the anesthetic bullet; at the same time, the historical position data of the target is collected by the GPS positioning device worn by the target, and it is input as an observation value into the Kalman filter model. The future motion trajectory of the target is predicted through the recursive calculation of the state equation and the observation equation. Prediction; the real-time state parameters of the anesthetic bomb and the predicted target motion trajectory are input into the pre-trained three-layer feedforward neural network model offline, and the optimal flight attitude adjustment of the anesthetic bomb is obtained through forward propagation calculation, including the adjustment values ​​of the pitch angle, yaw angle and roll angle; according to the optimal flight attitude adjustment output by the neural network, a PID control algorithm is adopted, and the deviation between the attitude adjustment value and the current attitude is used as input. The control signal of the servo motor is output through PID calculation to drive the control surface of the anesthetic bomb to produce corresponding deflection, thereby dynamically adjusting its flight attitude so that its flight trajectory is accurately consistent with the predicted target motion trajectory; during the flight of the anesthetic bomb During the operation, the relative position and speed between the anesthetic bomb and the target are continuously tracked, and the relative position and speed errors are input into the fuzzy controller based on Mamdani reasoning. By designing a reasonable membership function and fuzzy rule base, the proportional, integral and differential coefficients of the PID control algorithm are dynamically adjusted using fuzzy reasoning, making the control process more intelligent and adaptive. When the relative position error between the anesthetic bomb and the target is less than 10% of the target size and the relative speed error is less than 1m / s, it is determined that the anesthetic bomb has achieved precise intersection with the target, and a trigger command is sent to detonate the anesthetic bomb and release the anesthetic drug, completing precise anesthesia control of the target.

[0006] As a preferred solution, the intelligent guidance and control method for anesthesia bullets, wherein: the calculation of the optimal launch parameters of the anesthesia bullet based on the predicted motion trajectory includes: using a multi-objective optimization algorithm, taking the minimum deviation between the landing point of the anesthesia bullet and the target position as the objective function, and taking into account the constraints such as the anesthesia bullet range and flight time, to calculate the optimal launch angle and initial velocity parameters, including: According to the initial state of the anesthetic bullet and the initial motion state of the target, a motion model of the anesthetic bullet and the target is established to predict the intersection trajectory of the anesthetic bullet and the target; at the same time, a visual sensor is used to continuously track the actual motion state of the target to obtain the real-time position, velocity and acceleration information of the target; the actual motion state of the target is compared with the predicted intersection trajectory, and the deviation between the actual trajectory and the predicted trajectory in position and velocity is calculated; the position deviation threshold δp and the velocity deviation threshold δv are set. If the actual deviation exceeds the preset threshold, that is, |Δp|>δp or |Δv|>δv, it is judged that the target motion has deviated and the trajectory correction mechanism is triggered; according to the size and direction of the position deviation Δp and the velocity deviation Δv, the PID control algorithm is used to calculate the adjustment amount of the servo, drive the servo motor to change the servo angle, adjust the flight attitude and trajectory of the anesthetic bullet, and realize online correction of the trajectory; the input of ... The input is position deviation and velocity deviation, and the output is the adjustment amount of the servo. Fast and smooth trajectory correction is achieved by adjusting the PID parameters. In view of the randomness and uncertainty of target motion, the extended Kalman filter (EKF) algorithm is used to predict the target motion. The target position, velocity and acceleration are used as state variables to establish the nonlinear state equation and measurement equation of the target motion. The EKF algorithm is used to achieve state estimation and prediction, and the target motion trend in the future period is obtained. According to the target motion trend predicted by EKF, the fuzzy adaptive PID control algorithm is used to dynamically adjust the parameters of the guidance law. The size of the position deviation and velocity deviation is used as the input of the fuzzy controller, and the PID parameters are dynamically adjusted through fuzzy reasoning, so that the guidance system can adapt to the changes in the target motion state and improve the guidance accuracy and robustness of the anesthetic bullet.

[0007] As a preferred embodiment, the intelligent guidance control method for anesthetic bullets, wherein: the launching of the anesthetic bullet includes: controlling the launch device to perform precise adjustment and calibration based on the calculated optimal launch angle and initial speed parameters to ensure that the actual launch parameters of the anesthetic bullet are consistent with the calculated optimal parameters, including: According to the initial state parameters of the anesthetic bullet and the target, the motion model of the anesthetic bullet and the target is established, and the state equation and observation equation are determined; the state equation describes the motion law of the anesthetic bullet and the target, and the observation equation describes the relationship between the sensor measurement value and the state variable; the real-time position, velocity and acceleration information of the target are continuously obtained through the visual sensor as the observation quantity of the extended Kalman filter algorithm; the extended Kalman filter recursively estimates the optimal value of the state variable and its covariance matrix by linearizing the state equation and observation equation, and predicts the motion trend of the target in the future; according to the target motion trend obtained by the extended Kalman filter, the deviation between the target position, velocity and the expected value is calculated, and the fuzzy control algorithm is used to evaluate the deviation; the fuzzy control obtains the adjustment amount of the PID control parameter based on the size and change rate of the deviation through the preset fuzzy rules; the fuzzy control The PID parameter adjustment obtained is used as the input of the adaptive PID control algorithm to dynamically adjust the proportional, integral and differential parameters in the guidance law; the adaptive PID control calculates the guidance instructions based on the adjusted parameters to control the flight attitude of the anesthetic bomb; the pitch angle, yaw angle and roll angle of the anesthetic bomb are controlled according to the optimized guidance instructions to enable it to continuously track the target; at the same time, the relative distance and speed between the anesthetic bomb and the target are calculated in real time, and when the trigger conditions are met, the anesthetic device is triggered to achieve precise guidance; the target motion state is continuously tracked, and the real-time position, velocity and acceleration information are obtained through the visual sensor as the new observation quantity of the extended Kalman filter, and the state estimation value and prediction value are corrected to form a closed-loop control; at the same time, the parameters of the fuzzy control and adaptive PID control are updated according to the new state estimation to improve the robustness and adaptability of the control system.

[0008] As a preferred solution, the intelligent guidance control method for anesthetic bullets, wherein: the tracking of the motion state of the anesthetic bullet and the target includes: continuously tracking the flight trajectory of the anesthetic bullet using a visual sensor, and obtaining the real-time position information of the target through a GPS positioning device, including: Based on the acquired historical motion data of the target, the Kalman filter algorithm is used to establish the target motion model; during the flight of the anesthetic bomb, the target is continuously tracked by the target detection algorithm; when the target cannot be detected in multiple consecutive frames, it is determined that the target is blocked by an obstacle; the established motion model is used to extrapolate and predict the current motion state parameters of the target, including position, speed, etc.; the three-dimensional point cloud data of the obstacle is obtained through laser radar scanning, and the position and shape information of the obstacle is extracted from the point cloud using the DBSCAN clustering segmentation algorithm; the obstacle information and the predicted target motion state are input into the path planning module, and the A search algorithm is used to improve the cost function by introducing the obstacle avoidance factor to search for an optimal flight path in three-dimensional space that avoids obstacles and can intersect with the target; according to the planned flight The PID control algorithm is used to adjust the flight attitude and thrust of the tranquilizer bullet in real time, so that it flies along the optimal path and minimizes the deviation from the predetermined path until it intersects with the target; during the flight of the tranquilizer bullet, the target is continuously tracked; when the target reappears after being blocked, the target is recaptured using the target detection algorithm, and the flight path is corrected according to the actual position; if the error between the target motion state estimated by the Kalman filter and the actual observation value exceeds the preset threshold during the flight, the path replanning is triggered; the flight path is replanned based on the latest target and obstacle information obtained; the corrected flight path is converted into attitude and thrust control instructions for the tranquilizer bullet, and the tranquilizer bullet is guided to continue flying along the corrected path by executing the control instructions until it successfully hits the target, completing the tracking and interception mission.

[0009] As a preferred solution, the intelligent guidance control method for anesthesia bullet, wherein: the flight attitude of the anesthesia bullet is dynamically adjusted according to the motion state of the tracked anesthesia bullet and the target, including: using a PID control algorithm, according to the relative position and speed deviation between the anesthesia bullet and the target, dynamically adjusting the flight attitude of the anesthesia bullet so that its flight trajectory accurately matches the predicted target motion trajectory, including: Obtain the 3D point cloud data of the obstacle through laser radar scanning, and transmit the point cloud data to the onboard computer for processing; use the DBSCAN clustering algorithm to cluster and segment the 3D point cloud data, and extract the position and shape information of the obstacle; based on the position and shape information of the obstacle, combined with the flight path of the tranquilizer bullet, use the artificial potential field method to determine whether there is a potential collision risk; if there is a collision risk, use the A search algorithm to plan an obstacle avoidance flight path based on the position and shape information of the obstacle, and control the tranquilizer bullet to fly along the path to bypass the obstacle; when the target reappears after being blocked by the obstacle , recapture the target through the YOLOv3 target detection algorithm and obtain the target's real-time position information; based on the target's real-time position information, combined with the current position and posture of the tranquilizer bullet, the PID control algorithm is used to correct the flight path of the tranquilizer bullet to ensure that the target can be continuously tracked; during the tracking process, obstacle detection and obstacle avoidance are continuously performed; through multi-threaded parallel processing, one thread is responsible for target tracking, and the other thread is responsible for obstacle detection and obstacle avoidance path planning; data exchange and synchronization are achieved between the two threads through shared memory; the flight path is corrected in real time according to changes in the target position until the tracking and interception mission is completed.

[0010] As a preferred solution, the intelligent guidance and control method for tranquilizer bullets, wherein: when a target is blocked by an obstacle, the target's motion state is predicted by using historical data and prior knowledge of the target's motion, through motion model extrapolation, and the optimal flight path of the tranquilizer bullet to bypass the obstacle is calculated through a path planning algorithm, including: Acquire target tracking data, preprocess the data, use median filtering to remove outliers, use low-pass filtering to remove high-frequency noise, and improve data quality and reliability; analyze the target motion characteristics based on the preprocessed data, and estimate the target speed and acceleration; combine environmental condition information such as wind speed and temperature to adaptively adjust the proportional, integral, and differential coefficients of the PID controller to optimize controller performance and improve control accuracy and robustness; input the preprocessed tracking data into the Kalman filter, establish the target motion state equation and observation equation, recursively estimate the target's position, speed and other state variables, and predict the state at the next moment to reduce the impact of measurement error and process noise; the results of the Kalman filter are used as the input of the neural network, and the target motion model is established through training to learn the target's motion law and predict the target's position and speed in the future to provide a reference for control decisions; use the target feature information in the target tracking data, such as the target's size, shape, texture, etc., combined with prior knowledge, such as the feature template of the target type, to construct a support vector The quantitative machine classification model is used to realize the classification and identification of the target; the identification results are integrated with the results of Kalman filtering and neural network prediction to generate a comprehensive state estimation of the target, thereby improving the pertinence and accuracy of tracking; based on the comprehensive state estimation of the target, the model predictive control algorithm is used to solve the optimization problem online, dynamically adjust the output of the PID controller, generate the optimal control instructions, and drive the gimbal, launch mechanism and other actuators to track and strike the target; at the same time, the control instructions are compared with the prediction results, the tracking error is calculated, and the continuity and stability of tracking are guaranteed through feedback control; during the tracking process, target information and tracking effect evaluation indicators such as tracking error and off-target time are collected in real time; when the evaluation indicators exceed the preset threshold, the adaptive correction mechanism of the control strategy and algorithm parameters is triggered; through the optimization algorithm, the noise covariance matrix of the Kalman filter, the weight coefficient of the neural network, the kernel function parameters of the support vector machine, etc. are adjusted online to realize intelligent and adaptive target tracking control and improve the environmental adaptability and robustness of the tracking system.

[0011] The technical solution provided by the embodiment of the present invention may have the following beneficial effects: The present invention discloses a method for accurately striking uncertain moving targets. In order to solve the problem that the target's motion trajectory is highly random and easily obstructed by obstacles, the real-time motion parameters of the target are obtained through multi-sensor information fusion, the target state is predicted using the Kalman filter algorithm, and the optimal tranquilizer bullet launch parameters are calculated in combination with the multi-target optimization algorithm. During the strike process, the guidance rate parameters are optimized in real time through adaptive control. When the target deviates from the predicted trajectory, servo control is used to correct the trajectory; when the target is obstructed by an obstacle, the target state is predicted based on historical data and the optimal detour path is planned. At the same time, a proportional-integral-differential controller is established to process tracking data to improve accuracy and anti-interference ability. This method can effectively solve the problem of accurately striking uncertain moving targets, improve the success rate of strikes, and has strong practicality and reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions in the present invention or the prior art, a brief introduction is given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0013] Figure 1 The figure is a flow chart of an intelligent guidance control method for an anesthetic bullet of the present invention.

[0014] Figure 2 The figure is a schematic diagram of an intelligent guidance control method for an anesthetic bullet of the present invention.

[0015] Figure 3 This is another schematic diagram of an intelligent guidance control method for an anesthetic bullet according to the present invention; DETAILED DESCRIPTION In order to make the objectives, technical solutions and advantages of the present invention more clear, the present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0016] Example 1: like Figure 1-3 In this embodiment, a method for intelligent guidance and control of an anesthetic bullet may specifically include: Step 1: Utilizing multi-sensor information fusion to obtain the target's real-time position, velocity, and acceleration parameters, the Kalman filter algorithm is used to estimate and predict the target's random and uncertain motion state, thereby determining the target's trajectory for a period of time. Based on the predicted trajectory, a multi-objective optimization algorithm is used to calculate the optimal launch angle and trajectory parameters for the tranquilizer projectile, ensuring precise intersect between the projectile and the target in both space and time.

[0017] Based on the real-time position, velocity, and acceleration information of the target acquired by multiple sensors, the Kalman filter algorithm filters the target's motion state to obtain a filtered estimate of the target's motion state. If the target's motion state is highly random and uncertain, the process noise covariance matrix of the Kalman filter algorithm is increased to improve the algorithm's adaptability and obtain a more accurate estimate of the target's motion state. The Kalman filter-derived estimate of the target's motion state is used as input to predict the target's trajectory over a period of time using a uniform acceleration model. The predicted trajectory parameters, such as the target's position, velocity, and acceleration, are obtained. Based on the predicted trajectory parameters, the trajectory of the tranquilizer bullet at different launch angles and initial velocities is calculated using a modified Euler trajectory model. The trajectory parameters, such as flight time and impact point, are obtained for multiple sets of tranquilizer bullet launch angles and initial velocity parameters. A genetic algorithm is used to optimize the launch angle and initial velocity parameters for these multiple sets of tranquilizer bullets. By setting the objective function to minimize the deviation between the impact point and the target position, while also considering constraints such as the bullet's range and flight time, the optimal launch angle and initial velocity parameters are obtained, ensuring optimal spatial and temporal alignment with the predicted target trajectory. Based on the optimal launch angle and initial velocity parameters obtained through genetic algorithm optimization, the anesthetic bullet launcher is precisely adjusted and calibrated to ensure that the actual launch parameters of the anesthetic bullet are consistent with the calculated optimal parameters. After the anesthetic bullet is launched, its flight trajectory is continuously tracked and compared in real time with the target trajectory predicted in the second step. Based on the deviation between the two, the PID control algorithm dynamically adjusts the flight attitude and trajectory of the anesthetic bullet, ultimately achieving a precise rendezvous with the target.

[0018] For example, the Kalman filter algorithm can be used to estimate the motion state of an object. Imagine using multiple sensors to track a running hare. The sensors provide real-time information on the hare's position, velocity, and acceleration. However, sensor data inevitably contains noise and errors, such as occlusion from leaves or sensor accuracy limitations. The Kalman filter acts like an intelligent filter, fusing the observations from multiple sensors and incorporating the hare's motion patterns (for example, a hare cannot accelerate from a standstill to 100 km / h instantly) to estimate the hare's motion state more accurately. If the hare's motion is highly erratic, such as sudden turns or jumps, the process noise covariance matrix in the Kalman filter increases, meaning that the observed data is more reliable than predictions based on previous motion patterns. This allows the filter to better adapt to the hare's random motion and produce a more accurate state estimate. Once the hare's motion state is accurately determined, its future trajectory needs to be predicted. Assuming a uniform acceleration model is used, this assumes that the hare maintains a relatively constant acceleration over short periods of time. For example, if the Kalman filter estimates a hare's current speed at 5 m / s and its acceleration at 1 m / s², then its speed can be predicted to be 6 m / s in one second, with a corresponding change in position. Of course, the uniform acceleration model is a simplification; in reality, the hare's motion may be more complex. Next, the flight trajectory of the tranquilizer bullet needs to be calculated. The modified Euler trajectory model, which accounts for factors such as air resistance, can more accurately calculate the bullet's flight path. Different launch angles and initial velocities can be set, such as 30, 45, and 60 degrees, and initial velocities of 20, 25, and 30 m / s. The flight trajectory of the tranquilizer bullet under these parameters can be calculated, yielding multiple sets of data, including flight time and impact location. With the tranquilizer bullet's flight trajectory and the predicted hare's trajectory, a genetic algorithm can be used to optimize the bullet's launch parameters. Like an optimizer simulating natural selection, the genetic algorithm "breeds" and "mutates" different launch angle and initial velocity combinations, selecting those that result in a landing point closer to the hare's predicted position. The objective function is to minimize the deviation between the tranquilizer bullet's impact point and the hare's predicted position. Constraints must also be considered, such as the bullet's range must not exceed the device's maximum range, and its flight time must not be so long that the hare escapes range. A genetic algorithm can be used to find the optimal launch angle and initial velocity, ensuring the bullet has the best chance of hitting the hare. Once the optimal launch parameters are found, the tranquilizer bullet launcher must be precisely adjusted and calibrated. For example, based on the calculated optimal launch angle of 40 degrees and initial velocity of 23 meters per second, the launcher's elevation angle and charge are adjusted to ensure that the actual launch parameters are consistent with the calculated results. After the tranquilizer bullet is launched, its flight trajectory must be continuously tracked. For example, a high-speed camera can be used to capture the bullet's position and compare it in real time with the hare's predicted trajectory. If any deviation is detected, a PID control algorithm is used for dynamic adjustment.The PID control algorithm acts like an automatic corrector, adjusting the tranquilizer bullet's flight attitude and trajectory based on the magnitude and trend of the deviation. For example, if the tranquilizer bullet lags behind the predicted hare's position, the PID controller will increase the bullet's thrust or adjust its flight direction to bring it closer to the target. The ultimate goal is to ensure the tranquilizer bullet accurately hits the moving hare.

[0019] In the second step, based on the optimal launch angle and initial velocity parameters obtained by genetic algorithm optimization, the launch device is precisely adjusted and calibrated to ensure that the actual launch parameters of the anesthetic bullet are consistent with the calculated optimal parameters. The flight trajectory of the anesthetic bullet is continuously tracked after launch. At the same time, combined with the predicted target motion trajectory, the PID control algorithm is used to dynamically adjust the flight attitude of the anesthetic bullet, ultimately achieving precise spatial and temporal intersection between the anesthetic bullet and the target.

[0020] Based on the optimal launch angle and initial velocity parameters obtained through genetic algorithm optimization, the launcher's angle adjustment mechanism and velocity control unit are used to precisely set these parameters. High-precision angle and velocity sensors are used to collect the launcher's actual launch angle and initial velocity in real time. These parameters are compared with the optimal parameters. If the deviation exceeds a preset threshold, a PID control algorithm is triggered to dynamically adjust the launch angle and initial velocity by adjusting the motor speed of the angle adjustment mechanism and the throttle position of the velocity control unit until the deviation between the actual and optimal parameters is within an acceptable range. After the anesthetic bullet is launched, its flight trajectory is continuously tracked by a high-speed camera. Using the camera's multi-view image information, the real-time three-dimensional coordinates of the bullet in space are calculated using triangulation principles. The velocity vector is then calculated from the rate of change of the coordinates, and the acceleration vector is calculated from the rate of change of the velocity. This allows the real-time spatial position, velocity, and acceleration of the bullet to be determined. Simultaneously, historical position data collected from the target's GPS positioning device is used to obtain the target's historical motion trajectory. This historical motion trajectory is then fed into a Kalman filter model as observations. The target's future motion trajectory is then predicted through recursive calculations of the state and observation equations. The real-time state parameters of the tranquilizer bullet and the predicted target motion trajectory are input into a pre-trained, offline, three-layer feedforward neural network model. Forward propagation is used to calculate the optimal flight attitude adjustment for the tranquilizer bullet, including pitch, yaw, and roll angles. Based on the optimal flight attitude adjustment output by the neural network, a PID control algorithm is employed. The deviation between the attitude adjustment value and the current attitude is used as input. The PID calculation outputs a control signal for the servo motor, which drives the control surfaces of the tranquilizer bullet to produce corresponding deflections, thereby dynamically adjusting its flight attitude so that its flight trajectory precisely matches the predicted target motion trajectory. During the flight of the tranquilizer bullet, the relative position and velocity between the bullet and the target are continuously tracked, and the relative position and velocity errors are input into a fuzzy controller based on Mamdani inference. By designing a suitable membership function and fuzzy rule base, fuzzy inference is used to dynamically adjust the proportional, integral, and differential coefficients of the PID control algorithm, making the control process more intelligent and adaptive. When the relative position error between the anesthetic bullet and the target is less than 10% of the target size and the relative speed error is less than 1m / s, it is determined that the anesthetic bullet has achieved precise intersection with the target, and a trigger command is sent to detonate the anesthetic bullet, release the anesthetic drug, and complete precise anesthesia control of the target.

[0021] For example, the Kalman filter acts like an intelligent filter, fusing data from multiple sensors and estimating the target's motion state more accurately based on its motion patterns. For example, to track a running hare, a GPS tracker can provide the hare's location, but GPS signals are easily affected by factors such as tree obstruction, resulting in low positioning accuracy. In this case, if an accelerometer is attached to the hare, the Kalman filter can combine the GPS positioning data with the accelerometer data to more accurately estimate the hare's position and velocity. If the hare suddenly accelerates or turns, the Kalman filter will adjust its estimate of the hare's motion state based on the change in acceleration, thereby maintaining tracking accuracy. A genetic algorithm is an optimization algorithm that simulates the process of natural selection and can be used to find optimal launch parameters. For example, to determine the launch angle and initial velocity of a tranquilizer bullet, a series of candidate parameter combinations can be set. A genetic algorithm can then simulate the process of natural selection to select the optimal combination. Specifically, the genetic algorithm first evaluates the fitness of each parameter combination—that is, the probability of the tranquilizer bullet hitting the target under that parameter combination. Then, based on the fitness, it selects some parameter combinations for "crossover" and "mutation" to generate new parameter combinations. This process iterates continuously until the optimal combination is found. The PID control algorithm is a commonly used automatic control algorithm that dynamically adjusts the control variable based on the deviation between the actual and target parameters, bringing the actual parameters closer to the target parameters. For example, to control the launch angle of a tranquilizer bullet, an angle sensor can measure the actual launch angle and compare it with the optimal launch angle calculated by a genetic algorithm. If there is a deviation between the two, the PID controller adjusts the motor speed of the angle adjustment mechanism based on the magnitude and trend of the deviation, bringing the actual launch angle closer to the optimal angle. The proportional, integral, and differential coefficients in the PID controller can be adjusted according to the specific situation to achieve optimal control. High-speed cameras can be used to track the flight trajectory of a tranquilizer bullet. By using multiple high-speed cameras to capture the bullet from different angles, the three-dimensional coordinates of the bullet in space can be calculated using the principle of triangulation. Then, by calculating the rate of change of the coordinates, the velocity and acceleration of the bullet can be determined. This information can be used to assess the bullet's flight state and provide a basis for subsequent control. A feedforward neural network is an artificial intelligence model that can be used to predict the optimal flight attitude adjustment for a tranquilizer bullet. For example, the real-time state parameters of a tranquilizer bullet and the predicted trajectory of a target can be fed into a pre-trained three-layer feedforward neural network. Based on these inputs, the neural network outputs adjustments to the bullet's pitch, yaw, and roll angles. These adjustments can be used to control the bullet's control surfaces so that its flight trajectory precisely matches the target's trajectory. A fuzzy controller is a control algorithm based on fuzzy logic that can handle uncertainty and nonlinear problems.For example, in the flight control of a tranquilizer bullet, the target's trajectory may be uncertain, and the bullet's flight dynamics are relatively complex. A fuzzy controller can dynamically adjust the parameters of the PID controller based on the relative position and velocity errors between the bullet and the target, making the control process more intelligent and adaptive. For example, if the distance between the bullet and the target is far, the fuzzy controller will increase the PID controller's proportional coefficient to accelerate the bullet's approach. If the distance between the bullet and the target is close, the fuzzy controller will decrease the PID controller's proportional coefficient to prevent the bullet from overtaking the target.

[0022] Step three: During the launch of the tranquilizer bullet, the target's actual motion is continuously tracked. If the target's actual trajectory deviates from the predicted trajectory, the bullet's trajectory is promptly adjusted. Servo control is used to achieve online trajectory correction to ensure continuous rendezvous between the bullet and the target. To address the potential randomness and uncertainty of the target's motion, adaptive control is used to optimize the guidance of the tranquilizer bullet in real time. By dynamically adjusting the guidance rate parameters, the guidance system can adapt to changes in the target's motion.

[0023] Based on the initial state of the anesthetic bullet launch and the initial motion state of the target, a motion model of the anesthetic bullet and the target is established, and the intersection trajectory of the two is predicted. Simultaneously, a visual sensor continuously tracks the target's actual motion state, acquiring real-time position, velocity, and acceleration information. The target's actual motion state is compared with the predicted intersection trajectory, and the position and velocity deviations between the two are calculated. Position and velocity deviation thresholds δp and δv are set. If the actual deviation exceeds the preset thresholds (i.e., |Δp| ​​> δp or |Δv| > δv), the target is considered to have deviated, triggering a trajectory correction mechanism. Based on the magnitude and direction of the position and velocity deviations Δp and Δv, a PID control algorithm is used to calculate the servo adjustment amount, driving the servo motor to change the servo angle, adjusting the flight attitude and trajectory of the anesthetic bullet, and achieving online trajectory correction. The PID control algorithm takes position and velocity deviations as inputs, and outputs the servo adjustment amount. Adjusting the PID parameters enables fast and smooth trajectory correction. To address the randomness and uncertainty of target motion, an extended Kalman filter (EKF) algorithm is used to predict target motion. Using the target's position, velocity, and acceleration as state variables, nonlinear state equations and measurement equations are established for target motion. The EKF algorithm is used to estimate and predict the target's motion trend over a period of time. Based on the EKF-predicted target motion trend, a fuzzy adaptive PID control algorithm is used to dynamically adjust the guidance law parameters. Using the position and velocity deviations as inputs to the fuzzy controller, the PID parameters are dynamically adjusted through fuzzy reasoning, enabling the guidance system to adapt to changes in the target's motion state and improving the guidance accuracy and robustness of the anesthetic projectile.

[0024] For example, the Kalman filter algorithm is a powerful tool for estimating the state of dynamic systems, such as tranquilizer rounds and target motion. In this scenario, it can be used to filter sensor noise and estimate the target's true position, velocity, and acceleration. For example, suppose a visual sensor detects a target's position as (10, 5, 2) and velocity as (2, 1, 0). However, these measurements may contain errors due to environmental interference or sensor limitations. The Kalman filter combines the target's motion model (e.g., uniform motion or uniform acceleration) with previous state estimates and performs a weighted average of these measurements to produce a more accurate state estimate, such as (11, 9, 9) and velocity as (9, 1, 1). This filtered state estimate is closer to the target's true motion. If the target suddenly turns or accelerates, the Kalman filter adjusts the state estimate based on the changes in the measurements, adapting to the target's motion changes. The extended Kalman filter (EKF) is designed to handle nonlinear systems. In a tranquilizer round guidance system, both the target's motion and the trajectory of the tranquilizer round may be nonlinear. The EKF approximates nonlinear systems by linearizing them. For example, if the target moves in a curve, its equation of motion will contain nonlinear terms. The EKF linearizes the equations of motion around the current state estimate and then applies the Kalman filter update step. This allows for better tracking of the target's nonlinear trajectory, thereby improving the accuracy of the tranquilizer bullet's impact. Establishing motion models for the tranquilizer bullet and the target is crucial. This can be based on a modified Euler trajectory model, taking into account factors such as air resistance and gravity. The target's motion model can be selected based on the target's type and motion characteristics, such as uniform linear motion, uniformly accelerated linear motion, or a more complex curved motion model. Once these two models are established, the trajectory of the tranquilizer bullet and the target can be predicted over a period of time and their intersection point calculated. For example, assume that the tranquilizer bullet is launched with a certain initial velocity and angle, and the target is moving in a straight line at a constant velocity. By calculating the trajectory of the tranquilizer bullet and the trajectory of the target, it can be predicted that the two will meet at the position (20, 15, 3) in 5 seconds. The trajectory correction mechanism is key to ensuring the accuracy of the tranquilizer bullet's impact. First, the position deviation threshold and velocity deviation threshold must be set. These two thresholds determine when trajectory correction is initiated. For example, the position deviation threshold can be set to 5 meters and the velocity deviation threshold to 2 meters per second. During flight, the target's actual motion is continuously tracked and compared with the predicted rendezvous trajectory. If the actual position deviation exceeds 5 meters or the actual velocity deviation exceeds 2 meters per second, the trajectory correction mechanism is triggered. The PID control algorithm is a classic control algorithm used to adjust the system output to achieve the desired value. In this example, the PID controller's inputs are position and velocity deviations, and its output is the servo adjustment. Based on the magnitude and trend of the deviations, the PID controller calculates the required servo adjustment angle, thereby correcting the tranquilizer bullet's flight attitude and trajectory.For example, if a tranquilizer bullet deviates by 6 meters from its target and the deviation is increasing, the PID controller will calculate a larger servo adjustment, such as 5 degrees, to quickly steer the bullet toward the target. If the deviation is small and stable, the PID controller will calculate a smaller adjustment, such as 5 degrees, to avoid overcorrection. The fuzzy adaptive PID control algorithm can further improve the robustness and accuracy of the guidance system. It dynamically adjusts the PID controller parameters using fuzzy logic. For example, if the target is moving rapidly, with large and fluctuating position and velocity deviations, the fuzzy controller will adjust the PID parameters more aggressively, such as increasing the proportional coefficient, making the control system more sensitive to deviation changes and enabling faster trajectory corrections. If the target is moving smoothly, with small and slowly changing deviations, the fuzzy controller will adjust the PID parameters more conservatively, such as decreasing the proportional coefficient, to avoid overcorrection and energy waste. Vision sensors continuously track the target's actual motion, acquiring real-time position, velocity, and acceleration information. This information forms the basis for trajectory corrections. For example, a high-speed camera can capture images of the target and calculate its position, velocity, and acceleration using image processing algorithms. These data are fed into the Kalman filter for state estimation and prediction.

[0025] In step 4, a motion model is established based on the initial state of the anesthetic bullet launch and the initial motion state of the target. The visual sensor is used to continuously track and obtain the real-time position, velocity and acceleration information of the target. The target motion is predicted by the extended Kalman filter algorithm to obtain the future motion trend. The fuzzy adaptive PID control algorithm is used to dynamically adjust the guidance parameters to improve the guidance accuracy and robustness of the anesthetic bullet.

[0026] Based on the initial state parameters of the tranquilizer projectile and target, a motion model of the projectile and target is established, and the state equation and observation equation are determined. The state equation describes the motion of the projectile and target, while the observation equation describes the relationship between sensor measurements and state variables. A visual sensor continuously acquires the target's real-time position, velocity, and acceleration information as observations for the extended Kalman filter algorithm. The extended Kalman filter linearizes the state and observation equations to recursively estimate the optimal values ​​of the state variables and their covariance matrix, and predicts the target's motion trend over a period of time. Based on the target motion trend obtained by the extended Kalman filter, the deviation between the target position and velocity and the expected values ​​is calculated, and the deviation is evaluated using a fuzzy control algorithm. Based on the magnitude and rate of change of the deviation, the fuzzy control algorithm uses pre-set fuzzy rules to infer the adjustment of the PID control parameters. The PID parameter adjustments obtained by the fuzzy control are used as input to the adaptive PID control algorithm, which dynamically adjusts the proportional, integral, and differential parameters in the guidance law. Based on the adjusted parameters, the adaptive PID control algorithm calculates guidance commands to control the flight attitude of the tranquilizer projectile. The optimized guidance commands control the pitch, yaw, and roll angles of the tranquilizer projectile to continuously track the target. Simultaneously, the system calculates the relative distance and velocity between the anesthetic bullet and the target in real time. When the trigger conditions are met, the anesthesia device is triggered, achieving precise guidance. The target's motion is continuously tracked, and real-time position, velocity, and acceleration information are acquired through visual sensors. These information is used as new observations for the extended Kalman filter, which then modifies the state estimate and prediction to form a closed-loop control system. Simultaneously, the parameters of the fuzzy control and adaptive PID control are updated based on the new state estimate, improving the robustness and adaptability of the control system.

[0027] For example, precise guidance of a tranquilizer bullet after launch is a complex control problem requiring the integrated application of multiple technical approaches. Here, we will analyze and illustrate a tranquilizer bullet guidance system based on an extended Kalman filter, fuzzy control, and adaptive PID control. First, we need to establish motion models for the tranquilizer bullet and the target. The motion model for the tranquilizer bullet can be simplified as parabolic motion under the influence of gravity and air resistance. Assume the initial velocity of the tranquilizer bullet is 20 m / s, the launch angle is 45 degrees, and the air resistance coefficient is 0.1. The target's motion model can be selected based on the actual situation, such as uniform linear motion, uniformly accelerated linear motion, or more complex curved motion. Assume the target is moving in a uniform linear motion at a speed of 5 m / s. These two models can be described using state equations, such as the position and velocity of the tranquilizer bullet and the position and velocity of the target. The observation equation describes the relationship between sensor measurements and state variables. For example, a visual sensor measures the pixel coordinates of the target in an image, which need to be converted into actual three-dimensional coordinates. The visual sensor continuously tracks the target, acquiring its position, velocity, and acceleration information. For example, at a given moment, a visual sensor measures the target's position as (10, 5, 2) and its velocity as (2, 1, 0). These measurements may contain errors due to sensor noise and environmental interference. The extended Kalman filter combines the motion model and sensor measurements to estimate and predict the target's state. Suppose the Kalman filter estimates the target's true position as (11, 6, 2) and its velocity as (2.2, 1.1, 0). It predicts the target's position at the next moment as (13.2, 7.7, 2). The Kalman filter effectively filters noise and predicts the target's motion trend. The extended Kalman filter is used because the motion models of the target and the tranquilizer bullet can be nonlinear, and the extended Kalman filter can handle such nonlinearities. Next, fuzzy control is used to adjust the parameters of the PID controller based on the deviation between the predicted and desired target positions. Assuming the desired position is (13, 7, 2), the position deviation is (0.2, 0.7, 0). Fuzzy control converts these deviations into fuzzy quantities based on pre-set fuzzy rules, such as "small deviation," "moderate deviation," or "large deviation." Then, based on fuzzy reasoning, the adjustment amount for the PID parameters is obtained. For example, if the deviation is large and increasing, the proportional coefficient of the PID controller needs to be increased to speed up the response. The advantage of this is that the control parameters can be dynamically adjusted according to changes in the target motion state, improving the robustness of the control system. The adaptive PID controller dynamically adjusts the proportional, integral, and differential parameters based on the adjustment amount of the fuzzy control output. For example, the original PID parameters are Kp=1, Ki=0.1, and Kd=0.01. After fuzzy control adjustment, the new PID parameters may become Kp=1.2, Ki=0.12, and Kd=0.012.Based on the adjusted parameters, the PID controller calculates guidance commands, such as the servo deflection angle. Based on these commands, the tranquilizer bullet's flight attitude is adjusted to keep it approaching the target. For example, if the tranquilizer bullet needs to deflect to the right, the PID controller outputs a positive servo deflection angle, controlling the servo to rotate rightward, thereby changing the bullet's flight direction. When the relative distance and speed between the tranquilizer bullet and the target meet the preset trigger conditions—for example, the distance is less than 0.5 meters and the relative speed is less than 1 meter per second—the tranquilizer device is triggered, releasing the anesthetic agent. The entire guidance process is a closed-loop control system. The visual sensor continuously tracks the target and inputs new measurements into the extended Kalman filter to update the state estimate and prediction. Simultaneously, the parameters of the fuzzy control and adaptive PID control are also updated, enabling the control system to adapt to changes in the target's motion state. For example, if the target suddenly turns, the visual sensor detects this change, the extended Kalman filter updates the target's state estimate, and the parameters of the fuzzy control and adaptive PID control are adjusted accordingly, allowing the tranquilizer bullet to adjust its trajectory in a timely manner and continue tracking the target.

[0028] Step 5: When the target is briefly obstructed by an obstacle, the system uses historical data and prior knowledge of the target's motion to predict its motion state through motion model extrapolation, maintaining tracking continuity. The system also obtains information about the obstacle's position and shape, and uses path planning to calculate the optimal flight path for the tranquilizer bullet to circumvent the obstacle and reach the target, thereby avoiding the obstacle and achieving rendezvous with the target.

[0029] A Kalman filter algorithm is used to build a target motion model based on historical target motion data. During the flight of the tranquilizer bullet, a target detection algorithm continuously tracks the target. If the target cannot be detected for multiple consecutive frames, it is determined to be obstructed by an obstacle. The established motion model is used to extrapolate and predict the target's current motion parameters, including position and velocity. Three-dimensional point cloud data of the obstacle is acquired through lidar scanning, and the DBSCAN clustering and segmentation algorithm is used to extract the obstacle's position and shape from the point cloud. This obstacle information and the predicted target motion state are input into the path planning module. The A-search algorithm, which improves the cost function by introducing an obstacle avoidance factor, searches for an optimal flight path in three-dimensional space that avoids obstacles and achieves intersection with the target. Based on the planned flight path, a PID control algorithm is used to adjust the tranquilizer bullet's flight attitude and thrust in real time to ensure that it follows the optimal path, minimizing deviation from the planned path until it intersects with the target. The target is continuously tracked throughout the flight of the tranquilizer bullet. When the target reappears after being obstructed, the target detection algorithm is used to recapture the target and adjust the flight path based on its actual position. If the error between the target's motion state estimated by the Kalman filter and the actual observed state exceeds a preset threshold during flight, path replanning is triggered. The flight path is replanned based on the latest acquired target and obstacle information. This revised flight path is converted into attitude and thrust control commands for the tranquilizer projectile. These commands are then executed to guide the tranquilizer projectile along the revised path until it successfully impacts the target, completing the tracking and interception mission.

[0030] For example, the Kalman filter algorithm is a commonly used algorithm for estimating system states, effectively handling noise and uncertainty. In this example, a Kalman filter can be used to build a target motion model. For example, assume that a visual sensor acquires historical target motion data, including position, velocity, and acceleration. This data is inevitably subject to measurement errors. Based on the target's motion patterns (e.g., uniform motion, uniform acceleration, etc.) and the measurement data, the Kalman filter can estimate the target's true motion state and predict its future trajectory. For example, assuming the target is moving in a straight line at a uniform speed, the Kalman filter can predict the target's position in the next frame based on the target's position and velocity information from the past few frames, even if the target in the current frame is temporarily obscured by an obstacle. During the flight of a tranquilizer bullet, continuous target tracking is required. Target detection algorithms can identify and locate targets from sensor data (e.g., images or videos). For example, deep learning-based target detection algorithms, such as YOLO or Faster R-CNN, can be used to detect the target's position in real time. If the target cannot be detected for multiple consecutive frames, it can be determined to be obscured by an obstacle. When a target is obstructed, the previously established motion model can be used to predict its current position and velocity. For example, if the target's velocity before occlusion was 10 meters per second and heading due east, it can be predicted that the target's position has moved 10 meters eastward one second after occlusion. LiDAR (Lidar) emits a laser beam and measures the time it takes for the beam to reflect back, thereby acquiring 3D point cloud data of the surrounding environment. The DBSCAN clustering and segmentation algorithm can be used to extract obstacle information from this point cloud data. For example, suppose a LiDAR scans a dense set of point clouds that are spatially close together but distant from other point clouds. The DBSCAN algorithm can cluster these point clouds into a cluster and identify them as obstacles. The position and shape of the obstacle can then be calculated based on the coordinates of these point clouds. The path planning module searches for an optimal flight path in 3D space that avoids obstacles and intersects the target. The A* search algorithm is a commonly used path planning algorithm. To avoid obstacles, an obstacle avoidance factor can be introduced to improve the cost function. For example, the cost of paths close to obstacles can be set higher, while the cost of paths away from obstacles can be set lower. This way, the A* search algorithm will tend to choose paths away from obstacles. Suppose the target is at (10, 10, 10), the tranquilizer bullet is at (0, 0, 0), and there is an obstacle at (5, 5, 5). The A* search algorithm will search for a path from the tranquilizer bullet to the target, trying to avoid obstacles. For example, it might choose a path that passes through (2, 2, 2), (7, 7, 7), and then to (10, 10, 10). The PID control algorithm is a classic control algorithm that adjusts the system output based on the deviation between the target value and the actual value.In this example, a PID control algorithm can be used to adjust the tranquilizer bullet's flight attitude and thrust. For example, if the tranquilizer bullet deviates from the planned path, the PID controller calculates adjustments to the servo and engine based on the magnitude and trend of the deviation to return the bullet to the planned path. When the target reappears, the target needs to be reacquired and the flight path corrected based on its actual position. For example, suppose the predicted target position is (10, 10, 10), while the observed target position is (12, 12, 12). The flight path needs to be corrected based on this deviation to ensure the tranquilizer bullet can more accurately hit the target. If the error between the target's motion state estimated by the Kalman filter and the actual observed value exceeds a preset threshold, the flight path needs to be replanned. For example, suppose the target suddenly changes direction, and the Kalman filter's predicted target position deviates from the actual position by more than 1 meter. In this case, the flight path needs to be replanned to accommodate the target's new motion state. The corrected flight path needs to be converted into attitude and thrust control commands for the tranquilizer bullet. For example, if the flight path needs to deflect upward by 10 degrees, this command needs to be sent to the servo to adjust the tranquilizer bullet's attitude upward by 10 degrees.

[0031] Step 6: Based on the three-dimensional point cloud data of the obstacle obtained by the lidar scanning, the DBSCAN clustering segmentation algorithm is used to extract the obstacle position and shape information. When the target reappears after being blocked, the target is recaptured through the target detection algorithm, and the flight path of the tranquilizer bullet is corrected according to the actual position information to ensure the completion of the tracking and interception mission.

[0032] LiDAR scanning acquires three-dimensional point cloud data of obstacles and transmits this data to the onboard computer for processing. The DBSCAN clustering algorithm is used to cluster and segment the 3D point cloud data, extracting the obstacle's position and shape information. Based on this obstacle's position and shape information and the tranquilizer bullet's flight path, an artificial potential field method is used to determine whether there is a potential collision risk. If a collision risk exists, an A-search algorithm is used to plan an obstacle-avoiding flight path based on the obstacle's position and shape information. The tranquilizer bullet is then directed along this path to circumvent the obstacle. When the target reappears after being obscured by an obstacle, the YOLOv3 object detection algorithm is used to recapture the target and obtain its real-time position. Based on this real-time position information and the current position and attitude of the tranquilizer bullet, a PID control algorithm is used to correct the flight path of the tranquilizer bullet to ensure continuous tracking of the target. During the tracking process, obstacle detection and avoidance are continuously performed. Multithreaded parallel processing is employed: one thread is responsible for target tracking, while another thread is responsible for obstacle detection and avoidance path planning. Data exchange and synchronization between the two threads is achieved through shared memory. The flight path is corrected in real time according to changes in the target position until the tracking and interception mission is completed.

[0033] For example, lidar is like a bat's echolocation system: it emits a laser beam and senses its surroundings by measuring the return time. Suppose the lidar on a tranquilizer bomb scans the surrounding environment, acquiring a large amount of point cloud data representing the three-dimensional coordinates of surrounding surfaces. This point cloud data is transmitted to the onboard computer for processing. After receiving the point cloud data, the onboard computer uses the DBSCAN clustering algorithm to process it. The DBSCAN algorithm can be imagined as a pile of sand. We want to group several densely packed piles together. The DBSCAN algorithm distinguishes which sand grains belong to the same pile based on the distance between the grains. Suppose the lidar scans a tree ahead. The leaves and trunk reflect numerous laser points, forming a dense point cloud cluster, while the point clouds on the ground and sky are relatively sparse. The DBSCAN algorithm can cluster this dense point cloud into a single cluster and identify it as an obstacle—a tree. By calculating the boundary points of this point cloud cluster, we can obtain the tree's location and approximate shape information, such as its height and width. After obtaining the position and shape of the obstacle, it is necessary to determine whether the tranquilizer bullet's flight path will collide with the obstacle. The artificial potential field method is like a magnetic field, with the obstacle considered a repulsive force field and the target an attractive force field. The tranquilizer bullet is subject to the combined effects of these two force fields during flight. If the obstacle is very close to the flight path, the repulsive force will be strong, indicating a potential collision risk. Suppose the tranquilizer bullet is flying towards its target, and the LiDAR scan indicates a wall in front of it. In this case, the repulsive force of the wall will be strong, indicating a collision risk. If the artificial potential field method determines a collision risk, a new obstacle-avoiding path must be planned. The A* search algorithm can be thought of as finding an exit in a maze. It searches all possible paths and selects the one with the lowest cost. In this scenario, the path cost is composed of distance and an obstacle avoidance factor. Paths closer to the obstacle have higher costs, while paths farther away have lower costs. Assuming the tranquilizer bullet needs to circumvent the wall mentioned earlier, the A* algorithm will search for several possible paths: over the wall, around it from the left, and around it from the right. The A* algorithm compares the costs of these three paths and selects the path with the lowest cost, such as bypassing the wall from the right side. When the target reappears after being briefly obscured by an obstacle, it is necessary to recapture the target. The YOLOv3 target detection algorithm is like a scout with sharp eyes and can quickly identify and locate the target. When the target reappears, the YOLOv3 algorithm can quickly identify and lock onto the target and obtain the target's real-time location information, such as the target's coordinates (x, y, z). After obtaining the target's real-time location information, the flight path of the tranquilizer bullet needs to be corrected to ensure that it can continue to track the target. The PID control algorithm is like a precise autopilot, which can automatically adjust the flight attitude and thrust of the tranquilizer bullet based on the deviation between the target position and the current position of the tranquilizer bullet.If the target moves to the left, the PID controller adjusts the tranquilizer bullet's servos based on this deviation, causing it to also veer left, and adjusts engine thrust to maintain distance from the target. Throughout the tracking process, obstacle detection, obstacle avoidance, and target tracking occur simultaneously. Imagine two robots: one responsible for detecting obstacles and planning a path, while the other tracks the target. These two robots communicate in real time via radio, sharing information. The tranquilizer bullet continuously adjusts its flight path based on changes in the target's position and obstacle information, ultimately flying toward the target with the precision of a guided missile, completing the tracking and interception mission.

[0034] Step seven: Establish a proportional-integral-differential controller to process the target tracking data in real time, improve the accuracy and reliability of the tracking data, enhance the anti-interference ability, ensure the stability of tracking, and achieve precise strikes on uncertain moving targets.

[0035] Target tracking data is acquired and preprocessed. Median filtering is used to remove outliers and low-pass filtering is used to remove high-frequency noise, improving data quality and reliability. Based on the preprocessed data, the target's motion characteristics are analyzed to estimate its velocity and acceleration. Environmental conditions, such as wind speed and temperature, are then adaptively adjusted to optimize controller performance, improving control accuracy and robustness. The preprocessed tracking data is fed into a Kalman filter to establish the target's motion state equation and observation equation. This recursive estimation of the target's position, velocity, and other state variables is used to predict the next state, minimizing the effects of measurement error and process noise. The Kalman filter results serve as input to a neural network, which is trained to build a target motion model, learn the target's motion patterns, and predict the target's position and velocity over time, providing a reference for control decisions. Target feature information from the tracking data, such as size, shape, and texture, is combined with prior knowledge, such as a characteristic template of the target type, to construct a support vector machine classification model for target classification and recognition. The recognition results are then integrated with the Kalman filter and neural network predictions to generate a comprehensive target state estimate, improving tracking targeting and accuracy. Based on the comprehensive target state estimate, a model predictive control algorithm is used to solve the optimization problem online, dynamically adjust the output of the PID controller, generate optimal control instructions, and drive the gimbal, launch mechanism, and other actuators to track and strike the target. Simultaneously, the control instructions are compared with the predicted results, and the tracking error is calculated. Feedback control is used to ensure the continuity and stability of tracking. During the tracking process, target information and tracking effect evaluation indicators such as tracking error and off-target time are collected in real time. When the evaluation indicators exceed the preset threshold, an adaptive correction mechanism for the control strategy and algorithm parameters is triggered. Through the optimization algorithm, the noise covariance matrix of the Kalman filter, the weight coefficients of the neural network, and the kernel function parameters of the support vector machine are adjusted online to achieve intelligent and adaptive target tracking control, improving the environmental adaptability and robustness of the tracking system.

[0036] For example, the first step in target tracking is to acquire and preprocess target tracking data. This is like capturing a moving object clearly with a camera. The captured video then needs to be processed, for example, to remove noise and blur. Specifically, a median filter can be used to remove outliers, such as the sudden appearance of birds or other distracting objects. A low-pass filter can also be used to remove high-frequency noise, such as camera shake or electronic interference, thereby improving data quality and reliability. This is similar to removing blur and smearing in a video to create a clearer image. Next, the preprocessed data needs to be analyzed to estimate the target's velocity and acceleration. For example, the change in the target's position over several consecutive frames can be used to calculate its velocity, while the change in its velocity can be used to calculate its acceleration. Furthermore, environmental conditions, such as wind speed and temperature, need to be incorporated to adaptively adjust controller parameters. For example, if the wind speed is high, the controller's proportional coefficient needs to be increased to respond more quickly to the target's movement. This is like flying a kite in strong winds; the kite's string angle needs to be adjusted more quickly to maintain stability. To more accurately estimate the target's state, a Kalman filter is used. The Kalman filter acts like an intelligent predictor. It estimates a target's current position, velocity, and other state variables based on its historical motion data and current observations, and predicts its state at the next moment. For example, even if a target is briefly obscured by trees, the Kalman filter can still predict its position and velocity during the obstruction based on its previous trajectory, much like being able to roughly estimate the position of the vehicle ahead even when your eyes are briefly closed. To learn the target's motion patterns, the Kalman filter output can be used as input to a neural network, which is trained to build a target motion model. A neural network is like an experienced hunter. It can learn the target's motion patterns based on its historical trajectory and predict its position and velocity over a period of time, providing a reference for control decisions, just as a hunter can predict its prey's next move based on its footprints and habits. To more accurately identify targets, target feature information from target tracking data, such as size, shape, and texture, can be combined with prior knowledge, such as a characteristic template of the target type, to construct a support vector machine classification model. A support vector machine is like an experienced appraiser, accurately classifying and identifying a target based on its features, just as an expert can determine the type of gemstone based on its color, luster, and hardness. By fusing the results of Kalman filtering, neural networks, and support vector machines, a comprehensive state estimate of the target can be generated. This is like combining the opinions of multiple experts to arrive at a more comprehensive and accurate assessment. Based on this comprehensive state estimate, model predictive control algorithms can be used to solve optimization problems online, dynamically adjust the output of the PID controller, and generate optimal control instructions to drive actuators such as the gimbal and launch mechanism to track and strike the target.This is like adjusting the muzzle direction and firing timing in real time based on the target's motion to ensure a hit. During the tracking process, target information and tracking performance evaluation metrics, such as tracking error and off-target time, must be collected in real time. When these metrics exceed preset thresholds, adaptive correction mechanisms for control strategies and algorithm parameters are triggered. This is like constantly adjusting a scope to ensure accuracy during a shooting operation. Optimization algorithms can be used to adjust the Kalman filter's noise covariance matrix, the neural network's weight coefficients, and the support vector machine's kernel function parameters online, enabling intelligent, adaptive target tracking control and improving the tracking system's environmental adaptability and robustness. This is like adjusting a firearm's configuration and shooting techniques based on different environments and targets to ensure accurate hits in all situations.

[0037] Step eight, by obtaining feature information such as target size, shape and texture, combined with the feature template of the target type, and using the support vector machine classification model, target classification and recognition are achieved. The recognition results are integrated with the Kalman filter and neural network prediction results to obtain the target comprehensive state estimation, thereby improving the targeting and accuracy of tracking.

[0038] Image data of the target to be tracked is acquired and segmented using the grabCut algorithm in OpenCV to extract the target's contour and texture features. Based on prior knowledge of the target type, a large amount of sample image data is used to construct a feature template library for different target types through feature extraction and clustering algorithms. This is then used to match target features. The target image feature data is divided into training and test sets. The training set data is used to train a support vector machine model, optimizing model parameters and improving classification accuracy. The extracted target features are then input into the trained support vector machine classification model, and the target's class label is obtained through model inference. The state vector and covariance matrix of the Kalman filter are initialized based on the target's initial position and velocity. At each time step, the filter's state estimate and covariance matrix are updated based on the current observations to obtain the target's position and velocity estimate for the next moment. The target is continuously tracked using a target detection algorithm to obtain its trajectory data. The trajectory data is preprocessed to extract a sequence of feature vectors. This sequence of feature vectors is used to train a long short-term memory neural network model, optimizing model parameters and improving prediction accuracy. The trained model is then used to predict the target's future motion trends. The support vector machine classification results, Kalman filter prediction results, and neural network prediction results are weighted and averaged. The weights can be set based on the respective confidence levels or prior knowledge to obtain a comprehensive state estimate of the target. Based on the comprehensive state estimate, it is determined whether the target motion has undergone a sudden change or anomalies such as occlusion have occurred. If an anomaly is detected, the Kalman filter process noise covariance matrix is ​​adjusted to increase prediction uncertainty. Simultaneously, the decision threshold of the support vector machine classification model is adjusted to improve recognition tolerance. By adaptively adjusting the tracking algorithm parameters, the robustness and accuracy of tracking are improved, achieving stable tracking of the target.

[0039] For example, image data of the target to be tracked can be acquired using devices such as high-definition cameras and infrared imagers. For example, in a scenario where a drone is tracking a ground vehicle, the drone's onboard high-definition camera can capture real-time image data of the vehicle. Using the grabCut algorithm in OpenCV to segment the image, the target can be separated from its background. For example, by inputting a captured vehicle image into the grabCut algorithm, the vehicle's outline can be extracted from the surrounding background, such as the road and trees, much like cutting the vehicle out of the image with scissors. Contour features of the target are extracted, such as its perimeter, area, and shape complexity. Texture features of the target are also extracted, such as its surface color and texture roughness. These features can be used to distinguish different target types. For example, the outline of a vehicle is typically rectangular, while the outline of a pedestrian is typically elongated. Building a library of feature templates for different target types can improve target recognition accuracy. For example, image data of a large number of different vehicle models can be collected, their features extracted, and a library of vehicle feature templates can be built. When a new target needs to be identified, its features can be matched against those in the template library to determine whether it is a vehicle. Dividing the target image feature data into a training set and a test set allows the generalization ability of the support vector machine model to be evaluated. For example, 80% of the vehicle image data is used as the training set to train the support vector machine model, while the remaining 20% ​​is used as the test set to evaluate the model's recognition accuracy. Using the support vector machine model for object classification allows objects to be classified into different categories. For example, by inputting the extracted target features into a trained support vector machine model, it can determine whether the target is a car, truck, or pedestrian. The Kalman filter's state vector can be initialized based on the target's initial position and velocity, such as the target's initial position coordinates and velocity. The covariance matrix represents the uncertainty of the state estimate, and the initial value can be set based on the actual situation. For example, if the target's initial velocity is highly uncertain, the covariance matrix element corresponding to velocity can be set larger. The Kalman filter can update the target's state estimate based on current observations, such as the target's position in the image. For example, if the target is observed to move to the right, the Kalman filter will update its position and velocity estimates accordingly. The Kalman filter can predict the target's position and velocity at the next moment. For example, based on the current target state estimate, a Kalman filter can predict the target's position and velocity at the next moment, which is used to guide the tracker's movement. Object detection algorithms can continuously track the target and obtain its trajectory data, such as its position coordinates in each image frame. Preprocessing this trajectory data, such as removing noise and outliers, can improve prediction accuracy. For example, if a target suddenly disappears from a frame, perhaps due to occlusion or other reasons, this frame of data can be treated as an outlier and removed.A sequence of feature vectors, such as the target's velocity, acceleration, and direction changes, is extracted from the trajectory data. These features can be used to describe the target's motion pattern. For example, if the target's trajectory is a straight line, its velocity and direction changes are relatively small. Using a long short-term memory neural network model, the target's motion patterns can be learned and its future motion trends can be predicted. For example, if the target has been moving at a constant speed in a straight line, the model can predict that it will continue to do so for a period of time in the future. A weighted average of the support vector machine classification results, the Kalman filter prediction results, and the neural network prediction results can be used to leverage the strengths of each algorithm and improve the accuracy of state estimation. For example, if the support vector machine model has a high confidence score, it can be given a higher weight. Based on the comprehensive state estimation results, it can be determined whether the target's motion has undergone a sudden change. For example, if the target's velocity suddenly changes dramatically, a sudden change may have occurred. Occlusion refers to the obstruction of observation of the target by other objects. For example, if the target is obscured by trees, the target is no longer visible in the image. If an anomaly is detected, such as a sudden change in target motion or occlusion, the Kalman filter parameters need to be adjusted. Increasing the process noise covariance matrix can increase prediction uncertainty, thereby better adapting to changes in target motion. For example, if the target undergoes a sudden change, increasing the process noise covariance matrix can enable the Kalman filter to converge to the new state more quickly. Adjusting the decision threshold of the support vector machine classification model can improve recognition tolerance. For example, if the target is occluded, the extracted features may be incomplete or inaccurate. Lowering the decision threshold can improve the model's recognition rate. By adaptively adjusting tracking algorithm parameters, the robustness and accuracy of tracking can be improved. For example, if the target undergoes a sudden change, the process noise covariance matrix of the Kalman filter can be increased, and the decision threshold of the support vector machine model can be lowered, allowing the tracking algorithm to better adapt to changes in target motion.

[0040] The description of the above embodiments is only used to help understand the technical solutions and core ideas of this application; ordinary technicians in this field should understand that they can still modify the technical solutions recorded in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A method for intelligent guidance and control of anesthetic bullets, characterized in that: include: Obtain the real-time motion parameters of the target; Predicting the target's motion trajectory based on the acquired target motion parameters; Calculate the optimal launch parameters of the tranquilizer bullet based on the predicted motion trajectory; Fire tranquilizer bullets; Tracking the movement of tranquilizer bullets and targets; According to the motion state of the tracked anesthetic bullet and the target, the flight posture of the anesthetic bullet is dynamically adjusted to achieve precise intersection with the target.

2. The intelligent guidance control method for anesthesia bullet according to claim 1, characterized in that: The acquiring of real-time motion parameters of the target includes: Multi-sensor fusion technology is used to obtain the real-time position, velocity and acceleration information of the target, and the Kalman filter algorithm is used to estimate and predict the target's motion state to obtain the target's motion trajectory in the future.

3. The intelligent guidance control method for anesthesia bullet according to claim 2, characterized in that: The method of calculating the optimal launch parameters of the anesthetic bullet based on the predicted motion trajectory includes: A multi-objective optimization algorithm is used to calculate the optimal launch angle and initial velocity parameters by taking the minimum deviation between the anesthetic bullet landing point and the target position as the objective function and considering the anesthetic bullet range and flight time constraints.

4. The intelligent guidance control method for anesthesia bullet according to claim 3, characterized in that: The launching of the tranquilizer bullet comprises: According to the calculated optimal launch angle and initial speed parameters, the launch device is controlled to perform precise adjustment and calibration to ensure that the actual launch parameters of the anesthetic bullet are consistent with the calculated optimal parameters.

5. The intelligent guidance control method for anesthesia bullet according to claim 4, characterized in that: The tracking of the motion state of the anesthetic bullet and the target includes: Visual sensors are used to continuously track the flight trajectory of the anesthetic bullet, and the real-time location information of the target is obtained through the GPS positioning device.

6. The intelligent guidance control method of anesthesia bullet according to claim 5, characterized in that: The method of dynamically adjusting the flight attitude of the tranquilizer bullet according to the motion state of the tracked tranquilizer bullet and the target includes: The PID control algorithm is used to dynamically adjust the flight attitude of the anesthetic bullet according to the relative position and speed deviation between the anesthetic bullet and the target, so that its flight trajectory accurately matches the predicted target motion trajectory.

7. The intelligent guidance control method of anesthetic bullet according to claim 6, characterized in that: Also includes: When the target is blocked by an obstacle, the target's motion state is predicted by using the historical data and prior knowledge of the target's motion and by extrapolating the motion model. The optimal flight path of the tranquilizer bullet to bypass the obstacle is calculated through a path planning algorithm.

8. The intelligent guidance control method for anesthesia bullet according to claim 7, characterized in that: The method of predicting the motion state of the target by extrapolating the motion model includes the following steps: The tracking data is input into the Kalman filter to establish the target motion state equation and observation equation; Recursively estimate the position and velocity state variables of the target and predict the state at the next moment; The result of Kalman filtering is used as the input of the neural network. Through training, the target motion model is established, the target motion law is learned, and the position and speed of the target in the future are predicted. Utilizing target feature information in target tracking data and combining it with prior knowledge, a support vector machine classification model is constructed to achieve target classification and recognition. The recognition results are fused with the results of Kalman filtering and neural network prediction to generate a comprehensive state estimate of the target.

9. The intelligent guidance control method for anesthesia bullet according to claim 8, characterized in that: The method of calculating the optimal flight path of the tranquilizer bullet to bypass obstacles by using a path planning algorithm comprises the following steps: Based on the target comprehensive state estimation, the optimization problem is solved online using the model predictive control algorithm; Dynamically adjust the output of the PID controller to generate the optimal flight path.

10. The intelligent guidance control method for anesthesia bullet according to claim 9, characterized in that: After generating the optimal flight path, the actuator is driven to track and strike the target. At the same time, the control instructions are compared with the predicted results, the tracking error is calculated, and the continuity and stability of tracking are guaranteed through feedback control.

Images