Unmanned platform autonomous navigation method, system and equipment based on small optical module

Abstract

The invention relates to an unmanned platform autonomous navigation method, system and equipment based on a small optical module. The method is applied to an unmanned platform carrying a small optical module, and the small optical module comprises a sensor module and a three-axis servo linkage control module. A target detection image of the environment where the small target is located is obtained through the sensor module. And extracting a plurality of small target calibration object images from the target detection image through an image detection model. And extracting position information characteristics of the small target calibration object in the small target calibration object image by adopting a long-time tracking algorithm, and obtaining a predicted position of the small target calibration object according to the position information characteristics and a preset regression model. And the three-axis servo linkage control module starts a multi-closed-loop nested loop servo control system according to the predicted position, so that when the unmanned platform controls the attitude of the unmanned platform to be stable, a navigation path between the position of the unmanned platform and the predicted position is generated. By adopting the method, the stability and precision of autonomous navigation of the unmanned platform can be improved.

Description

Technical Field

[0001] The present invention relates to the field of unmanned platform navigation technology, and in particular to an unmanned platform autonomous navigation method, system and equipment based on a small optical module. Background Art

[0002] At the scene of natural disasters (such as earthquakes and floods), ground communication facilities may be damaged and satellite navigation signals may be unstable, so unmanned platforms need to perform personnel search and rescue missions in harsh environments. Alternatively, when performing tasks such as forest fire monitoring and marine pollution monitoring, unmanned platforms need to maintain long-term autonomous operation in a vast and changing environment, and may face signal interference and obstruction problems.

[0003] In order to solve the navigation problem of unmanned platforms under the condition of interference with satellite navigation or communication links, the following solutions are usually adopted: inertial navigation system (INS), visual navigation system (VNS), multi-sensor fusion and ground station assisted navigation. The inertial navigation system measures the angular velocity and acceleration of the platform through sensors such as gyroscopes and accelerometers, and calculates the position and velocity by integration. Although the system can provide relatively accurate position and velocity information in a short time, due to the accumulation of sensor errors, long-term use will cause a significant decrease in navigation accuracy. The visual navigation system captures environmental images through a camera, uses computer vision technology to identify and track feature points in the environment, and performs positioning and navigation. However, in complex and dynamic environments, the visual navigation system is easily affected by factors such as light changes and occlusion, resulting in unstable positioning and navigation accuracy. The multi-sensor fusion solution combines multiple sensors such as inertial navigation systems, visual navigation systems, ultrasonic / lidar, etc., and improves navigation accuracy and robustness through data fusion algorithms. However, the system is highly complex, has a large amount of calculations, has high requirements on processor performance, and still has the problem of navigation error accumulation in strong interference environments. Ground station-assisted navigation uses ground stations to provide auxiliary navigation data and exchanges data with unmanned platforms through wireless communication. However, when the communication link is interfered or interrupted, the effectiveness of ground station-assisted navigation is greatly reduced, and it relies on the deployment and maintenance of ground station infrastructure. Summary of the invention

[0004] Based on this, it is necessary to provide an unmanned platform autonomous navigation method, system and equipment based on a small optical module that can improve the stability and accuracy of autonomous navigation in order to address the above technical problems.

[0005] An unmanned platform autonomous navigation method based on a small optical module is applied to an unmanned platform equipped with a small optical module, wherein the small optical module includes: a sensor module and a three-axis servo linkage control module. The method includes:

[0006] The target detection image of the environment where the small target is located is obtained through the sensor module.

[0007] The target detection image extracts multiple small target calibration images through the image detection model.

[0008] A long-term tracking algorithm is used to extract the position information features of the small target calibration object in the small target calibration object image, and the predicted position of the small target calibration object is obtained based on the position information features and the preset regression model.

[0009] The three-axis servo linkage control module starts the multi-closed-loop nested loop servo control system according to the predicted position, so that the unmanned platform can generate a navigation path between its own position and the predicted position when controlling its own posture to be stable.

[0010] An unmanned platform autonomous navigation system based on a small optical module, the system comprising:

[0011] The target detection image acquisition module is used to acquire the target detection image of the environment where the small target is located through the sensor module.

[0012] The small target image extraction module is used to extract multiple small target calibration images from the target detection image through the image detection model.

[0013] The predicted position acquisition module is used to extract the position information features of the small target calibration object in the small target calibration object image by using a long-term tracking algorithm, and obtain the predicted position of the small target calibration object according to the position information features and a preset regression model.

[0014] The navigation path generation module is used for the three-axis servo linkage control module to start the multi-closed-loop nested loop servo control system according to the predicted position, so that the unmanned platform can control its own posture to be stable and generate a navigation path between its own position and the predicted position.

[0015] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0016] The target detection image of the environment where the small target is located is obtained through the sensor module.

[0017] The target detection image extracts multiple small target calibration images through the image detection model.

[0018] A long-term tracking algorithm is used to extract the position information features of the small target calibration object in the small target calibration object image, and the predicted position of the small target calibration object is obtained based on the position information features and the preset regression model.

[0019] The three-axis servo linkage control module starts the multi-closed-loop nested loop servo control system according to the predicted position, so that the unmanned platform can generate a navigation path between its own position and the predicted position when controlling its own posture to be stable.

[0020] The above-mentioned unmanned platform autonomous navigation method, system and equipment based on small optical modules obtain target detection images through sensor modules, and use image detection models to accurately extract small target calibration object images to ensure the efficiency and accuracy of target detection. These images extract the position information features of small target calibration objects through long-term tracking algorithms, and obtain the predicted position of small target calibration objects based on these features and preset regression models to achieve accurate tracking of small targets. Accurately identify and locate small targets in complex backgrounds to improve the accuracy and reliability of target detection and tracking. The three-axis servo linkage control module starts a multi-closed-loop nested loop servo control system to ensure that the unmanned platform generates a stable navigation path while adjusting its own posture, thereby improving the stability and accuracy of autonomous navigation. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 A schematic diagram of a process of an unmanned platform autonomous navigation method based on a small optical module in one embodiment;

[0022] Figure 2 is a schematic diagram of an LCT tracking algorithm in one embodiment;

[0023] Figure 3 A schematic diagram of a servo control system framework based on unmanned platform target tracking in one embodiment;

[0024] Figure 4 A schematic diagram of a framework of an unmanned platform autonomous navigation system based on a small optical module in one embodiment;

[0025] Figure 5 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0026] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0027] In one embodiment, Figure 1As shown, an autonomous navigation method for an unmanned platform based on a small optical module is provided, which is applied to an unmanned platform equipped with a small optical module, wherein the small optical module includes: a sensor module, a three-axis servo linkage control module and an image processing unit. The sensor is composed of a front downward infrared camera, a front downward visible light camera, a downward infrared camera, and a downward visible light camera. The three-axis servo linkage control module includes: a gyro IMU unit, a position encoding unit, a three-axis servo motor and a servo control unit. The image processing unit includes: an image tracking unit and an image navigation unit.

[0028] The specific steps include:

[0029] Step 102: Acquire a target detection image of the environment where the small target is located through a sensor module.

[0030] Step 104: extract multiple small target calibration object images from the target detection image through the image detection model.

[0031] Step 106: extract the position information features of the small target calibration object in the small target calibration object image using a long-term tracking algorithm, and obtain the predicted position of the small target calibration object based on the position information features and a preset regression model.

[0032] Step 108, the three-axis servo linkage control module starts the multi-closed-loop nested loop servo control system according to the predicted position, so as to generate a navigation path between the unmanned platform's own position and the predicted position when the unmanned platform controls its own posture to be stable.

[0033] In the above-mentioned unmanned platform autonomous navigation method based on a small optical module, the target detection image is obtained through the sensor module, and the image detection model is used to accurately extract the small target calibration object image to ensure the efficiency and accuracy of target detection. These images extract the position information characteristics of the small target calibration object through the long-term tracking algorithm. According to these characteristics and the preset regression model, the predicted position of the small target calibration object is obtained to achieve accurate tracking of the small target. Accurately identify and locate small targets in complex backgrounds to improve the accuracy and reliability of target detection and tracking. The three-axis servo linkage control module starts the multi-closed-loop nested loop servo control system to ensure that the unmanned platform generates a stable navigation path while adjusting its own posture, thereby improving the stability and accuracy of autonomous navigation.

[0034] In one embodiment, the sensor module includes: a front downward infrared camera, a downward infrared camera, a front downward visible light camera, and a downward visible light camera. The sensor module mounted on the unmanned platform obtains the target detection image in the environment where the small target is located:

[0035] f D (x,y)=f T (x,y)+f B(x,y)+f N (x,y)

[0036] Among them, f D (x,y) is the target detection image, f T (x,y) is a small target, f B (x, y) is the background image, f N (x,y) is the noise, (x,y) is the position of the pixel.

[0037] In one embodiment, a foreground target matrix of small target pixels in a target detection image is constructed based on a rank sparse matrix, and an image detection model is constructed based on the foreground target matrix and the low-rank background matrix. The image detection model uses robust principal component analysis to separate the foreground and background of the target detection image. If the binarization of the target detection image is greater than a preset threshold, the foreground target image is separated. If the binarization of the target detection image is less than a preset threshold, the background image is separated.

[0038] In one embodiment, the image detection model is pre-trained using a machine learning algorithm and standard template data, and the foreground target image is input as a training sample data set into the pre-trained image detection model to extract the target position and pixel size of the small target pixel in the foreground target image. The foreground target image is subjected to feature extraction in a deep learning network according to the target position, pixel size, and preset feature extraction rules to obtain a small target feature map, and a candidate region is generated in the small target feature map. The relative position and size of the object real frame and the candidate region in the small target feature map are marked to obtain a candidate label value, and the candidate label value and the candidate region are input into the deep learning network for feature prediction, and a small target calibration object image corresponding to the small target prediction value is output.

[0039] In one embodiment, a long-term tracking algorithm is used to intercept the search window to extract the position information features of the small target calibration object in the small target calibration object image, and the position information features are input into a preset regression model for position prediction. After the RC (temporal context regressor: solving deformation and occlusion problems) regression model in the regression model performs a fast Fourier transform on each frame of the small target calibration object image, it is convolved with the cosine weight matrix to obtain the image features of the small target calibration object to be predicted, and the image features of the small target calibration object to be predicted are subjected to an inverse Fourier transform to obtain a response value, and the pixel position corresponding to the maximum value of the response value is used as the predicted position of the small target calibration object to be confirmed. A target multi-scale pyramid is constructed near the predicted position to be confirmed by the RT (target regressor: used for scale estimation) regression model in the regression model, and the optimal scale corresponding to the maximum value of the response value is calculated according to the target multi-scale pyramid to extract the small target directional gradient histogram, and the predicted position of the small target calibration object in the small target directional gradient histogram is output.

[0040] It is worth noting that in the process of tracking the target, the target may move quickly or be blocked. At the same time, due to the limitation of volume and power consumption, the tracking algorithm is required to have good real-time performance and the ability to re-capture the target after losing it. Therefore, the long-term correlation tracking algorithm (LCT) is used, such as Figure 2 As shown in the figure, the current position of the target is effectively estimated by estimating the scale and translation through discriminative correlation filters, which is accurate, real-time and robust, and also has a target re-detection mechanism. The specific steps include:

[0041] 1) Intercept the search window and extract features. The initial window is given by the target recognition algorithm, and the subsequent position is given by the detected target position;

[0042] 2) Calculate the offset of the target through the regression model RC and estimate the new position of the target (Xt, Yt). Specifically, first perform fast Fourier transform FFT on each frame of the image, then convolve it with the cosine weight matrix, and then find the place with the largest response value after inverse Fourier transform iFFT, which is the predicted position of the new target.

[0043] 3) The target scale change is calculated by the regression model RT. Specifically, an image of size sP×sQ is intercepted near the target prediction position, all normalized to P×Q size, a target multi-scale pyramid is constructed, the target HOG features are extracted, and the regression model Rt is used to calculate the scale with the largest response value, which is the optimal scale.

[0044] 4) Compare the maximum response value of the previous step with the preset value. If the maximum response value is smaller, the tracking is considered unreliable. Start the algorithm target re-detection module and use the online random fern classifier Drf for detection. The positive and negative samples used in training are obtained by classification using the KNN classifier.

[0045] 5) Update relevant model parameters, including regression model RC, regression model RT, detector Drf, and conduct tracking algorithm testing in combination with the application target object and application environment.

[0046] In one of the embodiments, the maximum value of the response value is compared with the threshold value. If the maximum value of the response value is less than the threshold value, the target re-detection module is started, the regression model is updated using an online random fern classifier, and the position information feature is predicted based on the updated regression model.

[0047] In one embodiment, the three-axis servo linkage control module includes: a gyro IMU unit, a position encoding unit, a three-axis servo motor and a servo control unit. The gyro IMU unit is used to collect the original posture information of the target calibration object through a three-axis sensor, calibrate the original posture information by temperature segmentation, obtain the corrected posture information, and periodically send the posture information to the servo control unit. The position encoding unit is used to obtain the absolute position of the start-up time of the three-axis servo motor, start the inner loop stabilization loop control according to the absolute position control, and detect the device position information of the small optical module, and transmit the device position information to the servo control unit to adjust the small optical module to move to a specified position. The three-axis servo motor is used to receive the servo control signal output by the servo control unit to lock the posture and position of the unmanned platform in the rolling direction, azimuth direction and pitch direction according to the posture and position of the target calibration object. The servo control unit is used to use multi-closed-loop nested loop control to perform path optimization calculation on the received device position information, posture information and predicted position, generate a servo control signal, and the servo control signal is fed along the multi-closed-loop nested loop.

[0048] In one embodiment, the multi-closed-loop nested loop servo control system includes: an inner-loop stabilization loop control and an outer-loop tracking loop control. The three-axis servo linkage control module starts the multi-closed-loop nested loop servo control system according to the predicted position. In the outer-loop tracking loop control, the tracking performance error index of the target calibration object is solved according to the received device position, attitude information, predicted position and the preset correction network transfer function, and the tracking performance error index is fed back to the inner-loop stabilization loop control for signal calibration, and a locking tracking control signal is generated. The locking tracking control signal is fed to the position encoding unit, the gyro IMU unit and the three-axis servo motor respectively, so that when the unmanned platform controls its own attitude to be stable, a navigation path between its own position and the predicted position is generated.

[0049] It is worth noting that when performing forest fire monitoring tasks, due to smoke and terrain obstruction, GPS signals are interfered and communication links are unstable. The drone uses sensor modules to obtain images of the fire area, uses image detection models to identify the fire source and the affected area, locates the fire source through long-term tracking algorithms, generates navigation paths, and can achieve real-time monitoring of the fire and data feedback. In addition, in the earthquake-stricken area, a drone was sent to search for survivors. Due to the collapse of buildings caused by the earthquake, GPS signals were blocked and communication links were also affected. The drone uses sensor modules to obtain images of the disaster area, uses image detection models to identify survivors in the rubble, determines the location of the survivors through long-term tracking algorithms, and generates navigation paths to guide rescue teams to rescue.

[0050] In one embodiment, if Figure 3As shown in the figure, the electrical structure of the servo control system framework based on unmanned platform target tracking, the optical signals obtained by the front downward visible light detector and the front downward infrared detector are converted into electrical signals and sent directly to the image tracking board, the optical signals obtained by the downward visible light detector and the downward infrared detector are converted into electrical signals and sent to the imaging board for signal synthesis processing, and then transmitted to the image tracking board in two ways. The image tracking board performs small target detection calculation on the image electrical signal, and the calculation results are output to the image navigation board through the Ethernet and SDI interface. The image navigation board takes the detected small target image as the path end point, performs positioning path planning, and outputs the path planning scheme to the airborne flight control, data link, and feeds it back to the image tracking board, so that the image tracking board synchronizes the path to the servo control board. The servo control board synthesizes the small target detection image processing results and the path planning scheme, and generates attitude control signals for each servo unit and gyro board respectively.

[0051] It should be understood that although Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0052] In one embodiment, Figure 4 As shown, an unmanned platform autonomous navigation system based on a small optical module is provided, including: a target detection image acquisition module 402, a small target image extraction module 404, a predicted position acquisition module 406 and a navigation path generation module 408, wherein:

[0053] The target detection image acquisition module 402 is used to acquire a target detection image of the environment where the small target is located through a sensor module.

[0054] The small target image extraction module 404 is used to extract a plurality of small target calibration object images from the target detection image through the image detection model.

[0055] The predicted position acquisition module 406 is used to extract the position information features of the small target calibration object in the small target calibration object image by using a long-term tracking algorithm, and obtain the predicted position of the small target calibration object according to the position information features and a preset regression model.

[0056] The navigation path generation module 408 is used for the three-axis servo linkage control module to start the multi-closed-loop nested loop servo control system according to the predicted position, so as to generate a navigation path between its own position and the predicted position when the unmanned platform controls its own posture to be stable.

[0057] For the specific definition of the unmanned platform autonomous navigation system based on a small optical module, please refer to the definition of the unmanned platform autonomous navigation method based on a small optical module above, which will not be repeated here. Each module in the above-mentioned unmanned platform autonomous navigation system based on a small optical module can be implemented in whole or in part through software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0058] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 5 As shown. The computer device includes a processor, a memory, a network interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a communication network deployment method for emergency response is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0059] Those skilled in the art will understand that Figure 3-Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0060] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0061] The target detection image of the environment where the small target is located is obtained through the sensor module.

[0062] The target detection image extracts multiple small target calibration images through the image detection model.

[0063] A long-term tracking algorithm is used to extract the position information features of the small target calibration object in the small target calibration object image, and the predicted position of the small target calibration object is obtained based on the position information features and the preset regression model.

[0064] The three-axis servo linkage control module starts the multi-closed-loop nested loop servo control system according to the predicted position, so that the unmanned platform can generate a navigation path between its own position and the predicted position when controlling its own posture to be stable.

[0065] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided by the present invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0066] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0067] The above-mentioned embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. An autonomous navigation method for an unmanned platform based on a small optical module, characterized in that: Applied to an unmanned platform equipped with a small optical module, wherein the small optical module includes: a sensor module and a three-axis servo linkage control module; The method comprises: Acquire a target detection image of the environment where the small target is located through the sensor module; The target detection image extracts a plurality of small target calibration object images through an image detection model; A long-term tracking algorithm is used to extract the position information features of the small target calibration object in the small target calibration object image, and a predicted position of the small target calibration object is obtained according to the position information features and a preset regression model; The three-axis servo linkage control module starts a multi-closed-loop nested-loop servo control system according to the predicted position, so that when the unmanned platform controls its own posture to be stable, a navigation path between its own position and the predicted position is generated.

2. The method according to claim 1, characterized in that: The sensor module includes: a front downward-looking infrared camera, a downward-looking infrared camera, a front downward-looking visible light camera, and a downward-looking visible light camera; Acquiring a target detection image of the environment where the small target is located by the sensor module includes: The sensor module mounted on the unmanned platform obtains the target detection image in the environment where the small target is located: f D (x,y)=f T (x,y)+f B (x,y)+f N (x,y) Among them, f D (x,y) is the target detection image, f T (x,y) is a small target, f B (x,y) is the background image, f N (x,y) is the noise, (x,y) is the position of the pixel.

3. The method according to claim 1, characterized in that Before the step of extracting a plurality of small target calibration object images from the target detection image through the image detection model, the method further includes: Constructing a foreground target matrix of small target pixels in the target detection image based on a rank sparse matrix, and constructing an image detection model according to the foreground target matrix and a low-rank background matrix; The image detection model uses robust principal component analysis to separate the foreground and background of the target detection image. If the binarization of the target detection image is greater than a preset threshold, the foreground target image is separated; if the binarization of the target detection image is less than the preset threshold, the background image is separated.

4. The method according to claim 3, characterized in that The target detection image extracts a plurality of small target calibration object images through an image detection model, including: Pre-training the image detection model using a machine learning algorithm and standard sample data, inputting the foreground target image as a training sample data set into the pre-trained image detection model to extract the target position and pixel size of small target pixels in the foreground target image; Extracting features of the foreground target image in a deep learning network according to the target position, the pixel size and a preset feature extraction rule to obtain a small target feature map, and generating a candidate region in the small target feature map; Mark the relative position and size of the object real box and the candidate area in the small target feature map to obtain the candidate label value, input the candidate label value and the candidate area into the deep learning network for feature prediction, and output the small target calibration object image corresponding to the small target prediction value.

5. The method according to claim 4, characterized in that A long-term tracking algorithm is used to extract the position information features of the small target calibration object in the small target calibration object image, and a predicted position of the small target calibration object is obtained according to the position information features and a preset regression model, including: A long-term tracking algorithm is used to intercept the search window to extract the position information features of the small target calibration object in the small target calibration object image, and the position information features are input into a preset regression model for position prediction. After each frame of the small target calibration object image is fast Fourier transformed by the RC regression model in the regression model, it is convolved with the cosine weight matrix to obtain the image features of the small target calibration object to be predicted, and the image features of the small target calibration object to be predicted are inversely Fourier transformed to obtain a response value, and the pixel position corresponding to the maximum value of the response value is used as the predicted position of the small target calibration object to be confirmed; A target multi-scale pyramid is constructed near the predicted position to be confirmed by the RT (target regressor: used for scale estimation) regression model in the regression model, and the optimal scale corresponding to the maximum value of the response value is calculated according to the target multi-scale pyramid to extract a small target directional gradient histogram, and the predicted position of the small target calibration object in the small target directional gradient histogram is output.

6. The method according to claim 5, characterized in that Before the step of extracting the small target directional gradient histogram and outputting the predicted position of the small target calibration object in the small target directional gradient histogram, the method further includes: The maximum value of the response value is compared with the threshold. If the maximum value of the response value is less than the threshold, the target re-detection module is started, the regression model is detected and updated using an online random fern classifier, and the position information feature is predicted based on the updated regression model.

7. The method according to any one of claims 1 to 6, characterized in that: The three-axis servo linkage control module includes: a gyro IMU unit, a position encoding unit, a three-axis servo motor and a servo control unit; The gyro IMU unit is used to collect the original posture information of the target calibration object through a three-axis sensor, calibrate the original posture information by temperature segmentation, obtain the corrected posture information, and periodically send the posture information to the servo control unit; The position encoding unit is used to obtain the absolute position of the start-up time of the three-axis servo motor, start the inner loop stabilization loop control according to the absolute position control, and detect the device position information of the small optical module, and transmit the device position information to the servo control unit to adjust the small optical module to move to a specified position; The three-axis servo motor is used to receive the servo control signal output by the servo control unit to lock the posture and position of the unmanned platform in the rolling direction, azimuth direction and pitch direction according to the posture and position of the target calibration object; The servo control unit is used to perform path optimization calculation on the received device position information, posture information and predicted position using multi-closed-loop nested loop control to generate a servo control signal, and the servo control signal is fed along the multi-closed-loop nested loop.

8. The method according to claim 7, characterized in that The multi-closed-loop nested loop servo control system includes: an inner loop stabilization loop control and an outer loop tracking loop control; The three-axis servo linkage control module starts a multi-closed-loop nested loop servo control system according to the predicted position, so that when the unmanned platform controls its own posture to be stable, a navigation path between its own position and the predicted position is generated, including: The three-axis servo linkage control module starts a multi-closed-loop nested loop servo control system according to the predicted position. In the outer-loop tracking loop control, the tracking performance error index of the target calibration object is solved according to the received device position, the attitude information, the predicted position and the preset correction network transfer function, and the tracking performance error index is fed back to the inner-loop stabilization loop control for signal calibration to generate a locking tracking control signal, which is fed to the position encoding unit, the gyro IMU unit and the three-axis servo motor respectively, so that when the unmanned platform controls its own attitude to be stable, a navigation path between its own position and the predicted position is generated.

9. An unmanned platform autonomous navigation system based on a small optical module, characterized in that: The system comprises: A target detection image acquisition module, used to acquire a target detection image of the environment where the small target is located through the sensor module; A small target image extraction module, used for extracting a plurality of small target calibration object images from the target detection image through an image detection model; A predicted position acquisition module, used to extract the position information features of the small target calibration object in the small target calibration object image by using a long-term tracking algorithm, and obtain the predicted position of the small target calibration object according to the position information features and a preset regression model; The navigation path generation module is used for the three-axis servo linkage control module to start the multi-closed-loop nested loop servo control system according to the predicted position, so that the unmanned platform can control its own posture to be stable and generate a navigation path between its own position and the predicted position.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.

Images