Direction-adjustable unmanned aerial vehicle accessory

Abstract

The invention, which relates to the technical field of the unmanned aerial vehicle, discloses a direction-adjustable unmanned aerial vehicle accessory comprising an environment sensing module, a target tracking module, a control module, a direction adjustment execution wireless communication module and a data storage module. The environment sensing module is composed of a light sensor, a wind direction sensor and a temperature sensor, the image recognition module sub-module and the signal tracking module sub-module are combined, accurate tracking of a target is achieved, the image recognition sub-module can capture and analyze image features of a target object in real time, and the image recognition sub-module can recognize the image features of the target object in real time. Even if target features are changed due to environmental changes, the tracking accuracy can be ensured by dynamically adjusting the weight coefficient of a tracking algorithm and re-determining a feature region and a feature value, meanwhile, the signal tracking sub-module receives a signal sent by a target object, fusion verification is carried out on the signal and an image recognition result, and the tracking accuracy is improved. And the reliability and the stability of target tracking are further improved.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicles, and particularly to an attachment for an unmanned aerial vehicle with adjustable direction. Background Art

[0002] With the rapid development of technology, unmanned aerial vehicle technology has been widely applied in various fields, such as aerial photography, agricultural plant protection, and environmental monitoring. In these applications, the target tracking ability of unmanned aerial vehicles has become one of the important indicators to measure their performance. In order to achieve efficient and accurate target tracking, researchers have been constantly exploring and innovating, combining advanced image processing technology, signal processing technology, and environmental perception technology to improve the autonomous operation ability and intelligent level of unmanned aerial vehicles.

[0003] Although traditional unmanned aerial vehicles have achieved certain results in target tracking, there are still some deficiencies. On the one hand, traditional image recognition algorithms often have difficulty accurately capturing and identifying target features in the face of complex and changing environments, resulting in poor tracking effects. On the other hand, traditional signal tracking technology is limited by the signal transmission distance and interference factors, and it is easy to experience signal loss or unstable tracking. In addition, traditional unmanned aerial vehicles also have shortcomings in environmental adaptability and are difficult to maintain a stable operation state in extreme or complex environments, thus limiting their application scope.

[0004] Therefore, it is necessary to develop an attachment for an unmanned aerial vehicle with adjustable direction to improve the autonomous operation ability and intelligent level of unmanned aerial vehicles and provide strong support for the further development of unmanned aerial vehicle technology. Summary of the Invention

[0005] The purpose of the present invention is to make up for the deficiencies of the existing technology and provide an attachment for an unmanned aerial vehicle with adjustable direction. The system integrates three core modules: image recognition, signal tracking, and environmental perception, to solve the deficiencies of traditional unmanned aerial vehicles in target tracking and environmental adaptability. Through the image recognition module, the system can capture and analyze the image features of the target object in real time. The signal tracking module receives the signals emitted by the target object and fuses and verifies them with the image recognition results to further improve the reliability of tracking. At the same time, the environmental perception module collects the environmental information around the unmanned aerial vehicle in real time and dynamically adjusts the operation parameters of the unmanned aerial vehicle according to this information to keep it in the best working state in complex and changing environments.

[0006] To solve the above technical problems, the present invention provides the following technical solution: an attachment for an unmanned aerial vehicle with adjustable direction, the system includes: an environmental perception module, a target tracking module, a control module, a steering execution wireless communication module, and a data storage module;

[0007] The environmental perception module: It consists of a light sensor, a wind direction sensor, and a temperature sensor, which collect the environmental information around the UAV in real time through the sensors and transmit the collected data to the control module in real time;

[0008] The target tracking module: It includes an image recognition module sub-module and a signal tracking module sub-module, and uses a combination of image recognition technology and signal tracking technology to track the target;

[0009] The image recognition sub-module: It includes a camera and an image processing chip. The camera is used to capture the images around the UAV and transmit them to the image processing chip. During the tracking process, when the initial set features of the target object change due to environmental changes, the chip performs spectral analysis on the captured images. By comparing the spectral feature changes of the pixels in the same area of the images at different times, each group of image data is divided into k regions, and the spectral feature change amount ΔS of each region of the target object at different times is calculated j,t where j represents the jth region, t represents different times, and the formula is: ΔS j ,t = ∑ i = 1 n ( S i,t - S i,t-1) 2 , where S i ,t represents the spectral value of the ith pixel at time t, n is the total number of pixels in the selected region j. When the spectral feature change amount exceeds the preset threshold, the feature region and feature value for identifying the target are re-determined. At the same time, according to the real-time motion state and feature change situation of the target object, the weight coefficient of the tracking algorithm is dynamically adjusted. Let the calculation formula for the weight coefficient adjustment amount ΔW in the target tracking algorithm be: where α and β are weight distribution coefficients used to balance the influence of speed and feature change on weight adjustment. Let the initial weight coefficient of the tracking algorithm be W 0 = ( w 0x ,w 0y ) , then the adjusted weight coefficient W = ( w x ,w y ) is: w x = w 0x + ΔW × cos ( γ ) , w y = w 0y + ΔW × sin ( γ ) , where γ is the angle between the motion direction of the target object and the positive direction of the x-axis of the image coordinate system, and the calculation formula is:

[0010] The signal tracking sub-module: Equipped with multi-band signal receiving antennas and a signal processing unit, it receives the signals emitted by the target object, mutually fuses and verifies with the image recognition results, and at the same time, transmits the position and motion state information of the target to the control module;

[0011] The control module: Receives the environmental information from the environmental perception module and the target information from the target tracking module. For the environmental information, it calculates the environmental adaptability adjustment parameters according to the preset environmental adaptation strategy; for the target information, it combines the adaptive adjustment parameters of the target tracking module and uses the target tracking filtering algorithm to calculate the accurate target tracking adjustment parameters;

[0012] The orientation execution mechanism: According to the instructions of the control module, it adjusts the direction of the UAV accessory by using the motor drive and transmission mechanism;

[0013] The wireless communication module transmits the data to the ground control station. The ground operator monitors the working state of the accessory and the monitoring data situation in real time for the received data. When an abnormality is found or the monitoring task strategy needs to be adjusted, an instruction is remotely sent to the control module, and the control module adjusts the working mode or parameters of the accessory according to the instruction;

[0014] The data storage module: Stores the environmental perception data, target tracking data, and accessory operation state data through the internal memory. The system includes: an environmental perception module, a target tracking module, a control module, an orientation execution wireless communication module, and a data storage module.

[0015] Further, a preset threshold T is set in the target tracking module s The calculation formula is: where ω j is the weight coefficient of the j-th region.

[0016] Furthermore, for the determination of the new feature region in the target tracking module, assuming the image region is R and the pixel point (x, y) is within the region R, the region change degree D R is defined to judge whether to re-analyze the feature region, and the calculation formula is: where I(x, y) is the gray value of the pixel (x, y) at the current moment, I prev (x, y) is the gray value of this pixel at the previous moment, N R is the total number of pixels within the region R. When D R >T d , T d is the preset region change threshold, mark this region as a region to re-analyze the feature region. For the determination of the contour of the new feature region, assuming the gradient amplitude of the pixel point (x, y) is G(x, y), the direction is θ(x, y), and the contour saliency is S(x, y), the calculation formula is: where N(x, y) is the set of neighboring pixels of pixel (x, y), and pixels with S(x, y) greater than the threshold T are selected s to form a preliminary contour, and then the contour feature region is obtained through morphological operations for optimization.

[0017] Furthermore, in the target tracking module, a texture feature analysis method based on the gray-level co-occurrence matrix is used to extract eigenvalues. Let the gray-level co-occurrence matrix be P d，θ (i, j) represents the occurrence probability that in a pair of pixels with a distance of d in the direction θ, one pixel has a gray level of i and the other has a gray level of j. The calculation formula for contrast C is: C = ∑ i ∑ j (i - j) 2 P d,Δ (i, j), and the calculation formula for entropy E is: E = -∑ i ∑ j P d，θ (i, j) log(P d，θ (i, j)), and the obtained eigenvalues are combined into a new feature vector

[0018] Furthermore, for the calculation of the real-time motion state in the target tracking module, let the coordinates of the target object on the image plane of the image processing chip be (x i , y i ), and the coordinate sequence after n image acquisitions is {(x 1 , y 1 ), (x 2 , y 2 ), …, (x n , y n )}, and the motion speed vector of the target object is The calculation formula is: where k vx and k vy are speed correction coefficients, which are related to the light intensity I, wind direction angle θ w and temperature T in the environmental perception module, and their calculation formula is: a 1 , a 2 , a 3 , b 1 , b 2 , b 3 are environmental impact weight coefficients.

[0019] Furthermore, for the calculation of the degree of change in image features in the target tracking module, let the area of the image feature region of the target object in the initial state be A 0 , and the area of the corresponding feature region after environmental change be A 1, the change ratio of the area of the feature region For the eigenvalue, let the initial eigenvalue be V 0 , and the changed eigenvalue be V 1 , the eigenvalue change amount △V = |V 1 - V 0 |, the degree of change of the image feature △C is calculated by the weighted sum of the change ratio of the area of the feature region and the eigenvalue change amount, and the calculation formula is: △C = ω 1 ×△A + ω 2 ×△V, where ω 1 and ω 2 are weight coefficients.

[0020] Furthermore, in the target tracking module, for the processing of the signals emitted by the received object, after receiving the signals from the target object, a digital filter is used to remove the noise and interference components in the signals, perform signal demodulation, and according to different signal types, restore the information carried by the original signals. For GPS signals, by calculating the satellite ephemeris data and the signal propagation time difference, the longitude and latitude coordinate information of the target object is calculated. For RF signals, according to its specific coding format and protocol, the custom position and status data of the target object are parsed.

[0021] Furthermore, for the fusion of the signal and the image recognition result in the target tracking module, let the target position obtained by image recognition be (x img , y img ), and its confidence level be C img , the target position obtained by signal tracking be (x sig , y sig ), and its confidence level be (C sig ), the calculation formula for the fused target position (x f , y f ) is: Among them, the image recognition confidence level C img The calculation formula is: N f is the number of successfully matched target features, N t is the total number of preset target features, S c is the clarity score of the target area in the image, E m is the error value of the image feature matching, E max is the maximum acceptable error value set according to experience, λ 1 , λ 2 , λ 3 are weight coefficients, and the signal tracking confidence level C sig The calculation formula is: S s is the received signal strength, Ss，max is the maximum intensity of this signal type under ideal conditions, Pc is the correct proportion of the data after signal demodulation, N err is the number of error codes during signal transmission, N total is the total number of transmitted code elements, μ 1 、μ 2 、μ 3 are weighting coefficients.

[0022] Furthermore, in the control module, the environmental adaptability adjustment parameter is calculated according to a preset environmental adaptation strategy. Let the light direction vector be the wind direction vector be the current direction vector of the accessory be then the environmental adaptability adjustment angle θ a The calculation formula is: where c 1 is the proportionality coefficient of the influence of the wind direction on the adjustment angle, k a is the comprehensive environment correction coefficient, and the calculation formula related to temperature is: where c2 is the temperature influence coefficient.

[0023] Furthermore, in the control module, the precise target tracking adjustment parameter is calculated by using the target tracking filtering algorithm. Let the predicted position of the target at the current moment be (x p , y p ), the actual position at the previous moment be (x l , y l ), the target motion speed vector be the target tracking algorithm weighting coefficient be W = (w x , w y ), then the tracking adjustment parameter a x in the x direction and the tracking adjustment parameter a y in the y direction are calculated as follows: where △t is the time interval.

[0024] Compared with the prior art, this adjustable-direction drone accessory has the following beneficial effects:

[0025] First, by combining the image recognition module sub-module and the signal tracking module sub-module, the present invention realizes precise tracking of the target. The image recognition sub-module can capture and analyze the image features of the target object in real time. Even if the target features change due to environmental changes, the accuracy of tracking can be ensured by dynamically adjusting the weighting coefficients of the tracking algorithm and re-determining the feature region and feature values. At the same time, the signal tracking sub-module receives the signals emitted by the target object and performs fusion verification with the image recognition results, further improving the reliability and stability of target tracking.

[0026] 2. The present invention collects the environmental information around the drone in real time through the environmental perception module, including light, wind direction, and temperature, and dynamically adjusts the operation parameters of the drone according to this information. The control module calculates the environmental adaptability adjustment parameters based on the preset environmental adaptation strategy, and uses an algorithm to calculate the precise target tracking adjustment parameters to ensure that the drone can maintain the best working state under various environmental conditions. This improvement in environmental adaptability enables the drone to operate stably in complex and changeable environments, improving the practicality and application scope of the drone.

[0027] Other advantages, objectives, and features of the present invention will be described to some extent in the subsequent specification, and to some extent, will be obvious to those skilled in the art based on the study of the following text, or can be taught from the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings based on these drawings without creative efforts.

[0029] Figure 1 It is a schematic structural diagram of an attachment for a drone with adjustable direction;

[0030] Figure 2 It is a schematic flow diagram of an attachment for a drone with adjustable direction. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0031] The following will clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0032] Embodiment 1:

[0033] Drone Traffic Monitoring in Urban Environments

[0034] In the urban traffic monitoring scenario, the drone is equipped with this attachment for a drone with adjustable direction to perform tasks.

[0035] The light sensor of the environmental perception module perceives light intensity, direction, and color information in real time. For example, in a street lined with high-rise buildings, the light sensor can detect reflected light from different directions and light changes in shadow areas, providing a reference for light conditions for subsequent image processing. The wind direction sensor captures wind direction and speed information. When the drone flies between high-rise buildings, the wind direction sensor can timely monitor the changing air flow direction and speed caused by building blockages, ensuring the stable flight of the drone. The temperature sensor monitors the environmental temperature, providing temperature data support for the operating state of the drone equipment in hot summers or cold winters.

[0036] The target tracking module includes an image recognition module sub-module and a signal tracking module sub-module, and uses a combination of image recognition technology and signal tracking technology to track the target.

[0037] The image recognition sub-module captures images of vehicles on the road through a camera and transmits them to the image processing chip. During the tracking process, if the light changes due to the vehicle entering a tunnel and the initially set vehicle color characteristics change, the image processing chip performs spectral analysis on the images. For each set of image data, the target object (vehicle) is divided into k regions, and the spectral feature change amount △S of each region of the target object at different times is calculated. j，t Among them, j represents the jth region, t represents different times, and the formula is △S j , t = ∑ i = 1 n (S i，t - S i，t-1 ) 2 , where S i , t represents the spectral value of the ith pixel at time t, n is the total number of pixels in the selected region j. If the spectral feature change amount exceeds the preset threshold T s , T s The calculation formula is: Among them, ω j is the weight coefficient of the jth region, and the characteristic region and characteristic value of the vehicle are re-determined. For the determination of the new characteristic region, let the image region be R, and the pixel point (x, y) is within the region R. The region change degree D R is defined to judge whether it is a characteristic region that needs to be re-analyzed. The calculation formula is: Among them, I(x, y) is the gray value of the pixel (x, y) at the current moment, I prev (x, y) is the gray value of this pixel at the previous moment, N R is the total number of pixels in the region R. When D R > T d , T dWhen it is the preset regional change threshold, mark this area as the feature area to be re-analyzed. For determining the contour of the new feature area, let the gradient magnitude of the pixel point (x, y) be G(x, y), the direction be θ(x, y), and the contour saliency be S(x, y). The calculation formula is: where N(x, y) is the set of neighboring pixels of the pixel (x, y). Select the pixel points where S(x, y) is greater than a certain threshold T s to form a preliminary contour, and then optimize it through morphological operations to obtain a stable contour feature area. For extracting new feature values, let the gray-level co-occurrence matrix P d，θ (i, j) represent the occurrence probability that in a pair of pixels with a distance of d in the direction θ, one pixel has a gray level of i and the other has a gray level of j. The calculation formula for the contrast C is: C = ∑ i ∑ j (i - j) 2 P d，θ (i, j), and the calculation formula for the entropy E is: E = -∑ i ∑ j P d,θ (i, j) log(P d,θ (i, j)), and combine the obtained feature values into a new feature vector for target recognition. At the same time, according to the real-time motion state and feature change situation of the vehicle, dynamically adjust the weight coefficient of the tracking algorithm. Let the calculation formula for the adjustment amount △W of the weight coefficient in the target tracking algorithm be: where α and β are weight distribution coefficients used to balance the influence of speed and feature change on weight adjustment. Let the initial weight coefficient of the tracking algorithm be W 0 =(w 0x , w 0y ), then the adjusted weight coefficient W = (w x , w y ) is: w x = w 0x + △W × cos(γ), w y = w 0y + △W × sin(γ), where γ is the angle between the moving direction of the target object and the positive direction of the x-axis of the image coordinate system. Then the calculation formula is: Here, the motion speed vector v of the vehicle = (vx, vy), and its calculation formula is: where, k vx and k vy are speed correction coefficients, which are related to the light intensity I, wind direction angle θ w and temperature T in the environmental perception module. Its calculation formula is: a 1 , a 2 , a3 , b 1 , b 2 , b 3 is the environmental impact weight coefficient.

[0038] In the image recognition sub-module, the degree of change in image features △C is calculated by the weighted sum of the change ratio of the feature region area and the change amount of the feature value. The calculation formula is: △C = ω 1 ×△A + ω 2 ×△V, where ω 1 and ω 2 are the weight coefficients. Let the area of the image feature region of the target object (vehicle) in the initial state be A 0 , and the area of the corresponding feature region after the environmental change be A 1 . The change ratio of the feature region area For the feature value, let the initial feature value be V 0 , and the changed feature value be V 1 . The change amount of the feature value △V = |V 1 - V 0 |.

[0039] The signal tracking sub-module receives the GPS signal of the vehicle and fuses and verifies it with the image recognition result. Let the target position obtained by image recognition be (x img , y img ), and its credibility is where N f is the number of successfully matched target features, N i is the total number of preset target features, S c is the clarity score of the target area in the image, E m is the error value of image feature matching, E max is the maximum acceptable error value set according to experience, λ 1 , λ 2 , λ 3 are the weight coefficients), and the target position obtained by signal tracking is (x sig , y sig ), and its credibility is where, S s is the received signal strength, S s，max is the maximum strength of this signal type under ideal conditions, P c is the correct verification ratio of the data after signal demodulation, N err is the number of error codes during signal transmission, N total is the total number of transmitted code elements, μ 1 , μ 2 , μ 3(where is the weight coefficient), the calculation formula for the fused target position (x f , y f ) is:

[0040] The control module calculates the adjustment parameters according to the environmental information and target information. When the environmental light is dim and affects the image recognition effect, the control module calculates the environmental adaptability adjustment angle θ x , l y ) and the wind direction vector w = (w x , w y ) information, and the calculation formula is; a Among them, c is the proportionality coefficient of the influence of the wind direction on the adjustment angle, k 1 is the comprehensive environmental correction coefficient, which is related to the temperature, and the calculation formula is: α Among them, c is the temperature influence coefficient, adjusts the direction of the drone attachment to make the camera better align with the target vehicle. For the target tracking adjustment parameters, assume that the predicted position of the target at the current moment is (x 2 , y p ), the actual position at the previous moment is (x p , y l ), the target motion speed vector is v = (v l , v x ), and the weight coefficient of the target tracking algorithm is W = (w y , w x ), then the tracking adjustment parameter a y in the x direction of the target and the tracking adjustment parameter a x in the y direction of the target y The calculation formula is: Among them, △t is the time interval, and the drone attitude is precisely adjusted to ensure that the target vehicle is always tracked.

[0041] The direction adjustment actuator quickly adjusts the direction of the drone attachment according to the control module instruction, using the motor drive and transmission mechanism, so that the camera always aligns with the target vehicle and obtains clear image data.

[0042] The wireless communication module transmits the collected data to the ground control station. The ground operator can monitor the traffic conditions in real time. If it is found that there is traffic congestion or an accident in a certain section, an instruction can be remotely sent to the control module, and the control module adjusts the flight path and monitoring focus of the drone to improve the traffic monitoring efficiency.

[0043] The data storage module stores the environmental perception data, target tracking data, and attachment operation status data, providing data support for subsequent traffic flow analysis and accident cause investigation.

[0044] In summary, in the urban traffic monitoring scenario, the direction-adjustable drone accessory in this direction obtains light, wind direction, and temperature information in real time through the environmental perception module, providing basic data for subsequent processing. The image recognition sub-module of the target tracking module uses a camera and an image processing chip to track vehicles, can re-determine the feature region and feature values according to the change of light, and dynamically adjust the weight coefficient of the tracking algorithm. The signal tracking sub-module receives the GPS signal of the vehicle and fuses it with the image recognition result to improve the positioning accuracy. The control module calculates the adjustment parameters according to the environmental and target information, and the direction adjustment actuator adjusts the direction of the drone accessory accordingly, so that the camera always aims at the target vehicle. The wireless communication module transmits the data to the ground control station, facilitating the operator to monitor and adjust. The data storage module stores relevant data for subsequent analysis.

[0045] Embodiment 2:

[0046] Forest fire monitoring

[0047] In the forest fire monitoring scenario, the drone is equipped with the accessory of the present invention to carry out operations.

[0048] The light sensor of the environmental perception module continuously monitors the light conditions in the forest, the wind direction sensor captures the wind direction information in real time, and the temperature sensor closely monitors the environmental temperature changes.

[0049] The target tracking module includes an image recognition module sub-module and a signal tracking module sub-module, and uses a combination of image recognition technology and signal tracking technology to track the target.

[0050] The image recognition sub-module takes pictures of the forest images. When a fire occurs, the image processing chip analyzes the images. For each group of image data, the flame and smoke areas are divided into k regions, and the spectral feature change amount △S of each region at different times is calculated j，t 、The formula is △S j , t = ∑ i = 1 n (S i，t - S i，t-1 ), 2 , where S i , t represents the spectral value of the i-th pixel at time t, n is the total number of pixels in the selected region j. When the spectral feature change amount exceeds the preset threshold T s , the calculation formula is: Re-determine the feature region and feature values of the flame and smoke, extract the new feature values, and set the gray-level co-occurrence matrix P d,θ (i, j) represents the occurrence probability that in a pixel pair with a distance of d in the direction θ, one pixel has a gray level of i and the other has a gray level of j. The calculation formula for the contrast C is; C = ∑ i ∑j (i - j) 2 P d，θ (i, j), the calculation formula of entropy E is: E = -∑ i ∑ j P d,θ (i, j)log(P d,θ (i, j)), combine the obtained eigenvalues into a new eigenvector Calculation of real - time motion state. Let the coordinates of the target object on the image plane of the image - processing chip be (x i , y i ). After n image acquisitions, the coordinate sequence is {(x 1 , y 1 ), (x 2 , y 2 ), …, (x n , y n )}. The motion speed vector of the target object is Its calculation formula is: where k vx and k vy are speed correction coefficients, related to the light intensity I, wind direction angle θ w and temperature T in the environmental perception module. Its calculation formula is: Here, a 1 , a 2 , a 3 , b 1 , b 2 , b 3 are environmental impact weight coefficients. According to the real - time motion state and characteristic changes of the flame and smoke, dynamically adjust the weight coefficients of the tracking algorithm. Let the calculation formula of the weight coefficient adjustment amount ΔW be: where α and β are weight distribution coefficients, used to balance the influence of speed and characteristic changes on weight adjustment. Let the initial weight coefficient of the tracking algorithm be W 0 =(w 0x , w 0y ). Then the adjusted weight coefficient W=(w x , w y ) is: w x =w 0x +ΔW×cos(γ), w y =w 0y +ΔW×sin(γ), where γ is the angle between the motion direction of the target object and the positive direction of the x - axis of the image coordinate system. Then the calculation formula is: Here, the motion speed vector of the flame and smoke v=(v x , vy ), and its calculation method is similar to vehicle tracking. The degree of change in image features ΔC is calculated by the weighted sum of the change ratio of the feature region area and the change amount of the feature value. The calculation formula is: △C = ω 1 ×△A + ω 2 ×△V.

[0051] The signal tracking sub-module receives the radio frequency signals sent by the sensors in the forest and fuses them with the image recognition results. Suppose the target position obtained by image recognition is (x img , y img ), and its credibility is C img , and the target position obtained by signal tracking is (x sig , y sig ), and its credibility is (C sig ). The calculation formula for the fused target position (x f , y f ) is: x f = And transmit the position and motion state information of the target to the control module.

[0052] The control module calculates and adjusts parameters according to the environmental and target information, and calculates the environmental adaptability adjustment angle θ according to the information of the wind direction vector a , and the calculation formula is: Adjust the direction of the drone attachment, and at the same time combine the target tracking to adjust the parameters. The calculation formula is: Command the drone to fly towards the core area of the fire. The direction adjustment actuator quickly adjusts the direction of the drone attachment according to the control module's instructions. The wireless communication module transmits the fire data to the ground control station in real time. The operator judges the fire situation based on this and organizes the rescue. The data storage module stores the relevant data to provide reference for subsequent work.

[0053] In summary, in the forest fire monitoring scenario, the environmental perception module of the attachment carried by the drone monitors the forest light, wind direction and temperature. The image recognition sub-module of the target tracking module extracts the features of the flame and smoke and adjusts the tracking weight coefficient. The signal tracking sub-module receives the radio frequency signal and fuses it with the image recognition result to determine the location and scope of the fire. The control module calculates and adjusts parameters according to the environmental and target information, controls the direction of the drone attachment to better monitor the fire, the direction adjustment actuator responds quickly, the wireless communication module transmits the fire data to the ground control station in real time, and the data storage module stores the data to provide reference for subsequent fire analysis and prevention.

[0054] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above-described exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be embraced within the present invention. Any reference signs in the claims should not be construed as limiting the claims involved.

Claims

1. A direction-adjustable drone accessory, characterized in that: The system includes: an environment perception module, a target tracking module, a control module, a direction adjustment execution wireless communication module and a data storage module; The environmental perception module is composed of a light sensor, a wind direction sensor and a temperature sensor, which collects environmental information around the drone in real time through the sensors and transmits the collected data to the control module in real time; The target tracking module includes an image recognition module submodule and a signal tracking module submodule, and uses a combination of image recognition technology and signal tracking technology to track the target; The image recognition submodule includes a camera and an image processing chip. The camera is used to capture images around the drone and transmit them to the image processing chip. During the tracking process, when the target object changes due to environmental changes, the chip performs spectral analysis on the collected image. By comparing the changes in spectral characteristics of pixels in the same area of ​​the image at different times, each group of image data is divided into k areas, and the spectral characteristic changes of each area of ​​the target object at different times are calculated. j，t Where j represents the jth region, t represents different moments, and the formula is: ΔS j , t = ∑ i =1 n (S i，t -S i，t-1 ) 2 , where S i , t represents the spectral value of the i-th pixel at time t, n is the total number of pixels in the selected area j, when the spectral feature change exceeds the preset threshold, the feature area and feature value used to identify the target are re-determined, and at the same time, the weight coefficient of the tracking algorithm is dynamically adjusted according to the real-time motion state and feature change of the target object. The calculation formula of the weight coefficient adjustment amount ΔW in the target tracking algorithm is as follows: Among them, α and β are weight distribution coefficients, which are used to balance the impact of speed and feature changes on weight adjustment. The initial tracking algorithm weight is W0 = (w 0x , w 0y ), then the adjusted weight coefficient W = (w x , w y ) is: w x =w 0x +ΔW×cos(γ),w y =w 0y +ΔW×sin(γ), where γ is the angle between the target object's moving direction and the positive direction of the x-axis of the image coordinate system. The calculation formula is: The signal tracking submodule is equipped with a multi-band signal receiving antenna and a signal processing unit to receive the signal sent by the target object and integrate it with the image recognition result for verification. At the same time, the position and motion state information of the target is transmitted to the control module; The control module receives the environmental information from the environmental perception module and the target information from the target tracking module, calculates the environmental adaptability adjustment parameters according to the preset environmental adaptation strategy for the environmental information, and calculates the accurate target tracking adjustment parameters by combining the adaptive adjustment parameters of the target tracking module with the target tracking filtering algorithm for the target information; The direction adjustment actuator: according to the instruction of the control module, the direction of the drone attachment is adjusted by using the motor drive and transmission mechanism; The wireless communication module transmits data to the ground control station. The ground operator receives the data to monitor the working status of the accessory and the monitoring data in real time. If an abnormality is found or the monitoring task strategy needs to be adjusted, the operator sends a command to the control module remotely. The control module adjusts the working mode or parameters of the accessory according to the command. The data storage module stores environment perception data, target tracking data and accessory operation status data through an internal memory.

2. The direction-adjustable drone accessory according to claim 1, characterized in that: The target tracking module presets a threshold T s The calculation formula is: where ω j is the weight coefficient of the jth region.

3. The direction-adjustable drone accessory according to claim 1, characterized in that: Determination of the new feature area in the target tracking module, assuming that the image area is R, the pixel point (x, y) is within the area R, and the area change degree D is defined R To determine whether to re-analyze the feature area, the calculation formula is: Where I(x, y) is the grayscale value of the pixel (x, y) at the current moment, I prev (x, y) is the gray value of the pixel at the previous moment, N R is the total number of pixels in region R, when D R >T d , T d When it is the preset regional change threshold, the region is marked as a re-analyzed feature region. For the new feature region contour determination, the gradient amplitude of the pixel point (x, y) is G(x, y), the direction is θ(x, y), and the contour significance is S(x, y). The calculation formula is: Where N(x, y) is the neighborhood pixel set of pixel (x, y), and select S(x, y) greater than the threshold T s The pixels of the pixels form a preliminary outline, which is then optimized through morphological operations to obtain the contour feature area.

4. The direction-adjustable drone accessory according to claim 1, characterized in that: The target tracking module uses a texture feature analysis method based on gray-level co-occurrence matrix to extract eigenvalues. Suppose the gray-level co-occurrence matrix P d，θ (i, j) represents the probability of one pixel with grayscale i and the other with grayscale j in a pair of pixels with a distance d in the direction θ. The calculation formula of contrast C is: C = ∑ i ∑ j (ij) 2 P d，θ (i, j), the entropy E calculation formula is: E = -∑ i ∑ j P d，θ (i, j)log(P d，θ (i, j)), and combine the obtained eigenvalues ​​into a new eigenvector 5. The direction-adjustable drone accessory according to claim 1, characterized in that: The calculation of the real-time motion state in the target tracking module assumes that the coordinates of the target object on the image plane of the image processing chip are (x i ,y i ), the coordinate sequence after n image acquisitions is {(x1, y1), (x2, y2), …, (x n ,y n )}, the target object’s velocity vector is The calculation formula is: where k vx and k vy is the speed correction coefficient, which is related to the light intensity I and wind direction angle θ in the environment perception module w It is related to the temperature T, and its calculation formula is: a1, a2, a3, b1, b2, b3 are environmental impact weight coefficients.

6. The direction-adjustable drone accessory according to claim 1, characterized in that: The calculation of the degree of change of image features in the target tracking module is as follows: the area of ​​the image feature region of the target object in the initial state is A0, the area of ​​the corresponding feature region after the environment changes is A1, and the area change ratio of the feature region is For the eigenvalue, let the initial eigenvalue be V0, the changed eigenvalue be V1, and the eigenvalue change ΔV = |V l -V0|, the degree of image feature change ΔC is calculated by the weighted sum of the feature region area change ratio and the feature value change, and the calculation formula is: ΔC = ω1×ΔA+ω2×ΔV, where ω1 and ω2 are weight coefficients.

7. The direction-adjustable drone accessory according to claim 1, characterized in that: The target tracking module processes the signal sent by the received object. After receiving the signal from the target object, a digital filter is used to remove the noise and interference components in the signal, and the signal is demodulated. According to different signal types, the information carried by the original signal is restored. For GPS signals, the longitude and latitude coordinate information of the target object is calculated by solving the satellite ephemeris data and the signal propagation time difference. For radio frequency signals, the customized position and status data of the target object are parsed according to its specific coding format and protocol.

8. The direction-adjustable drone accessory according to claim 1, characterized in that: The fusion of the signal and image recognition results in the target tracking module, assuming that the target position obtained by image recognition is (x img ,y img ), whose credibility is C img , the target position obtained by signal tracking is (x sig ,y sig ), its credibility is (C sig ), the fused target position (x f ,y f ) is calculated as: Among them, the image recognition credibility C img The calculation formula is: N f is the number of successfully matched target features, N t is the total number of features of the target preset, S c is the clarity score of the target area in the image, E m is the error value of image feature matching, E max is the maximum acceptable error value set based on experience, λ1, λ2, λ3 are weight coefficients, and signal tracking credibility C sig The calculation formula is: S s is the received signal strength, S s，max is the maximum strength of the signal type under ideal conditions, Pc is the correct verification ratio of the data after signal demodulation, N err is the number of error codes during signal transmission, N total is the total number of code elements transmitted, μ1, μ2, μ3 are weight coefficients.

9. The direction-adjustable drone accessory according to claim 1, characterized in that: The control module calculates the environmental adaptability adjustment parameter according to the preset environmental adaptability strategy, assuming that the light direction vector is The wind direction vector is The current direction vector of the attachment is Then the environmental adaptability adjustment angle θ a The calculation formula is: Among them, c1 is the proportional coefficient of the wind direction on the adjustment angle, k a is the comprehensive environmental correction factor, and the calculation formula related to temperature is: Where c2 is the temperature influence coefficient.

10. The direction-adjustable drone accessory according to claim 1, characterized in that: The control module uses the target tracking filter algorithm to calculate the accurate target tracking adjustment parameters. Suppose the predicted position of the target at the current moment is (x p ,y p ), the actual position at the previous moment is (x l ,y l ), the target motion velocity vector is The weight coefficient of the target tracking algorithm is W = (w x , w y ), then the tracking adjustment parameter a of the target in the x direction x and the tracking adjustment parameter a in the y direction y The calculation formula is: Where Δt is the time interval.

Images