Intelligent assembly method and system suitable for low-light environments

Abstract

The application provides a smart assembling method and system suitable for a low-illumination environment, and the method comprises the following steps: acquiring an image in the belly of an airplane, performing enhancement processing on the acquired image in the belly of the airplane to obtain an enhanced image; performing pose detection on an assembling hook and an assembling rack based on the enhanced image to obtain pose data; performing trajectory planning on the hanging missiles in the belly of the airplane based on the obtained pose data to output trajectory data; and controlling a platform to drive the assembling hook to hang the missile body on the assembling rack based on the output trajectory data.

Description

Technical Field

[0001] This invention belongs to the field of intelligent assembly technology, and particularly relates to intelligent assembly methods and systems suitable for low-light environments. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] Internal fuselage arming refers to the configuration of mounting missiles, bombs, or other weapons inside the fuselage of an aircraft. Currently, the operation of internal fuselage arming still mainly relies on manual visual inspection, which presents significant problems such as long loading time, high personnel requirements, and low operational efficiency.

[0004] With the continuous maturation of perception technologies such as machine vision, they are showing broad application prospects in the military field. By integrating visual positioning and motion control technologies, the automation level of the ordnance loading process can be significantly improved, reducing personnel workload, lowering risks, and increasing overall operational efficiency. However, the complex internal environment and harsh lighting conditions of the fuselage pose severe challenges to the stable identification and precise control of intelligent assembly systems, necessitating the development of an intelligent assembly solution with environmental adaptability, high identification accuracy, and high control precision.

[0005] Currently, the key challenges in creating a suitable environment for the aircraft belly include:

[0006] (1) Complex structure and difficult feature recognition. The internal structure of the fuselage is complex, and the bottom structure of the assembly frame is concealed. It is generally composed of two separate rectangular slots, and the structural features of the visual target are weak and the background interference is strong. This kind of environment poses a challenge to traditional image processing methods, which can easily lead to unstable feature recognition. Deep learning or point cloud recognition methods also often have false detections, making it difficult to achieve high-precision target positioning and alignment control.

[0007] (2) Poor lighting conditions limit image quality. The underside of the aircraft is often in low-light conditions, making it susceptible to interference from shadows, reflections, and other factors. Especially in actual use, the light source is strictly controlled, and it is impossible to improve image quality through supplementary lighting. This results in low contrast and high noise in the images acquired by the visual sensor, severely affecting the accuracy of target edge extraction and attitude estimation. Low-light environments refer to environments with low light intensity, below 30 lux. Dusk illumination: 10–100 lux. Furthermore, it should be noted that military scenarios strictly limit active lighting, fundamentally conflicting with the sensitivity requirements of visual sensors.

[0008] In summary, existing image enhancement algorithms focus on the overall effect of the image without specifically enhancing the characteristics of ArUco code recognition. Furthermore, current missile loading systems use template-based or deep learning-based recognition, which is not suitable for the complex environment inside the aircraft. Summary of the Invention

[0009] To overcome the shortcomings of the prior art, this invention provides an intelligent assembly method suitable for low-light environments, which realizes automated and intelligent operation of loading ammunition inside the fuselage, significantly reducing manpower input and safety hazards.

[0010] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions:

[0011] Firstly, a smart assembly method suitable for low-light environments is disclosed, including:

[0012] If the image inside the aircraft's belly is acquired under low light conditions and the ArUco code cannot be recognized, then the acquired image inside the aircraft's belly is enhanced to obtain an enhanced image that can recognize the ArUco code.

[0013] Pose data is obtained by detecting the pose of the assembly hook and the assembly rack based on the enhanced image with the ArUco code identified.

[0014] Based on the obtained pose data, the trajectory of the internally mounted missiles on the aircraft fuselage is planned and the trajectory data is output.

[0015] The control platform, based on the output trajectory data, drives the assembly hook to hang the projectile onto the assembly rack.

[0016] As a further technical solution, the acquired images of the aircraft's interior are enhanced to obtain enhanced images, specifically including:

[0017] The acquired images of the aircraft's belly were RGB images. The RGB images were converted to HSV space, and the V component was extracted as the main target for subsequent enhancement, while the H and S components remained unchanged.

[0018] The extracted luminance V component is processed to enhance image brightness; the extracted luminance V component is processed using a Retinex algorithm based on guided filtering.

[0019] The enhanced luminance V component improves contrast and enhances the visibility of image details;

[0020] Optimize overall brightness distribution and adjust overall brightness;

[0021] The enhanced V component is combined with the original H and S components and then converted into an enhanced RGB image.

[0022] As a further technical solution, an image of the inside of the aircraft fuselage is obtained, and the image of the inside of the aircraft fuselage contains ArUco codes, which are affixed to the assembly hook and the assembly rack respectively.

[0023] The corner pixel coordinates and depth information are obtained by recognizing the ArUco code. After coordinate system transformation, the world coordinate system coordinates are obtained, and the relative pose relationship between the assembly hook and the assembly frame is obtained.

[0024] As a further technical solution, the corner pixel coordinates of the ArUco code are obtained through identification, specifically including:

[0025] The image is processed by grayscale conversion, image filtering, threshold segmentation, and edge detection.

[0026] The ArUco code outline is filtered, the ArUco code perspective transformation is performed, and the ArUco code encoding is identified, ultimately obtaining the pixel coordinates of the ArUco code corner points.

[0027] As a further technical solution, trajectory planning for internally mounted ordnance on the aircraft fuselage is performed based on the obtained pose data, and the trajectory data is output, specifically including:

[0028] Based on the relative pose relationship between the assembly hook and the assembly frame, a piecewise polynomial interpolation method in joint space is adopted. The joint positions of the starting point, path point and the ending point are input, and the velocity and acceleration of the joint movement are calculated and output through piecewise polynomial.

[0029] As a further technical solution, a trajectory optimization step is also included:

[0030] The goal of time optimization for a six-DOF platform trajectory is to minimize the motion time while adhering to velocity and acceleration constraints. The trajectory, after piecewise polynomial interpolation programming, is divided into three segments, each with a time interval of T1, T2, and T3. An improved particle swarm optimization algorithm is used to optimize the trajectory, finding the shortest time for each segment, thus obtaining the optimal trajectory motion time. The objective function is expressed as:

[0031] f(t)=min(T1+T2+T3)

[0032] This function, also known as the fitness value function, is used to measure the quality of a particle. The smaller the fitness value of a particle, that is, the smaller the objective function value, the better the particle is.

[0033] Secondly, an intelligent assembly system suitable for low-light environments is disclosed, including:

[0034] The image processing module is configured to: acquire an image of the interior of the aircraft's belly; if the image of the interior of the aircraft's belly is acquired under low light conditions and the ArUco code cannot be recognized, then perform enhancement processing on the acquired image of the interior of the aircraft's belly to obtain an enhanced image so that the ArUco code can be recognized.

[0035] The pose data detection module is configured to: obtain pose data by detecting the pose of the assembly hook and the assembly frame based on the enhanced image that has been identified with ArUco code;

[0036] The trajectory planning module is configured to: plan the trajectory of the aircraft's internal weapons loadout based on the obtained pose data and output the trajectory data.

[0037] The assembly module is configured such that the control platform, based on the output trajectory data, drives the assembly hook to hang the projectile onto the assembly frame.

[0038] It should be noted that if the image of the inside of the aircraft's belly is acquired under normal lighting conditions, no image enhancement processing is required.

[0039] The above one or more technical solutions have the following beneficial effects:

[0040] After acquiring an image of the aircraft's belly, the technical solution of this invention first determines whether the assembly hook and assembly rack are coded with ArUco codes. If not, it performs enhancement processing. Based on the enhanced image, it realizes the identification and attitude estimation between the assembly hook and assembly rack, which helps in the subsequent trajectory planning between the assembly hook and assembly rack.

[0041] This invention's technical solution, in enhancing low-light images, utilizes an improved Retinex image enhancement algorithm based on guided filtering. It performs illumination estimation and enhancement processing on the V channel in the HSV space, achieving highly robust target recognition and pose estimation under complex lighting and structural backgrounds. This effectively improves the success rate of ArUco code recognition in low-light conditions and enhances the reliability of system operation in dark environments. This embodiment's sub-technical solution uses an improved guided filtering approach, focusing on improving ArUco code recognition performance after enhancement. The system uses a depth camera to recognize ArUco codes, demonstrating strong robustness.

[0042] This invention improves the trajectory stability and execution efficiency of the assembly platform. Closed-loop control based on visual feedback significantly enhances loading accuracy and reduces the risk of mis-docking. This invention achieves automated and intelligent operation of loading ammunition inside the fuselage through visual positioning and trajectory planning, significantly reducing manpower and improving operational efficiency.

[0043] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0044] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0045] Figure 1 This embodiment shows a schematic diagram of the sub-camera installation location and working area;

[0046] Figure 2 This example illustrates the sub-camera field of view and the use of ArUco codes.

[0047] Figure 3 ArUco code recognition process diagram;

[0048] Figure 4 Schematic diagram of coordinate system transformation;

[0049] Figure 5 Schematic diagram of low-light image enhancement based on improved Retinex;

[0050] Figure 6 Flowchart of the improved particle swarm optimization algorithm;

[0051] Figure 7 This is a schematic diagram of the overall principle of this embodiment;

[0052] Figure 8 This is a schematic diagram of the assembly of the sub-method in this embodiment;

[0053] Figure 9 The following is a simulation diagram of a 3-5-3 piecewise polynomial trajectory planning, where (a) is the position change curve; (b) is the velocity change curve; and (c) is the acceleration change curve.

[0054] Figure 10 This diagram illustrates the usage of the ArUco code and the movement of the ammunition hook.

[0055] Figure 11 This diagram illustrates how changes in the platform angle lead to changes in the position of the bomb hook. Detailed Implementation

[0056] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0057] It should be noted that the terminology used herein is for the purpose of describing particular implementations only and is not intended to limit the exemplary implementations of the present invention.

[0058] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0059] Terminology Explanation:

[0060] ArUco code: A square binary marker image based on Hamming code, used for visual localization and pose detection.

[0061] Retinex algorithm: an image enhancement algorithm based on the human visual system, used for image processing in low-light environments.

[0062] 3-5-3 Polynomial Interpolation: A piecewise polynomial interpolation method used to improve trajectory smoothness.

[0063] Particle Swarm Optimization (PSO): A swarm intelligence-based optimization algorithm used to solve complex optimization problems.

[0064] Example 1

[0065] Current methods for internal munition loading operations in low-light environments lack intelligent assembly strategies adapted to the low-light conditions inside the aircraft fuselage. There are also issues with insufficient accuracy in visual recognition and detection under complex environments, and the inability to recognize ArUco codes in low-light conditions. Furthermore, the current six-degree-of-freedom platform, which supports the munition body, with the assembly hook attached to the body and the assembly rack inside the fuselage, suffers from low trajectory planning efficiency.

[0066] This embodiment discloses an intelligent assembly method suitable for low-light environments. Using this method, joint space trajectory planning is performed on the motors of a six-degree-of-freedom platform. The motors move according to the planned speed and acceleration. ArUco codes are affixed near the assembly hooks and assembly racks. The overall steps are detailed in the appendix. Figure 7 As shown, it includes:

[0067] Step 1: Acquire a low-light image of the aircraft's interior under RGB lighting conditions. If the ArUco code cannot be recognized in the low-light image, enhancement processing is performed. If the ArUco code can be recognized, the pose between the assembly hook and the assembly rack is directly detected. The depth camera is mounted on the platform body of the six-DOF platform. See the attached diagram for camera installation instructions. Figure 1 As shown, the camera's field of view is shown in the appendix. Figure 2 As shown, this is to ensure that the viewpoint including the assembly hook and assembly rack can be captured, and that the ArUco code can be photographed.

[0068] In this embodiment of the patent, when the ArUco code cannot be identified and enhancement processing is required, an improved Retinex image enhancement algorithm based on guided filtering is introduced for low-light environments to perform illumination estimation and enhancement processing on the V channel in the HSV space. Combined with CLAHE and Gamma correction, image contrast is further improved, enhancing target visibility.

[0069] See appendix Figure 5 As shown, the basic steps of the low-light image enhancement algorithm based on the improved Retinex are as follows:

[0070] Step (1-1) Image color space conversion. Convert the RGB image to HSV space, extract the V component as the main target for subsequent enhancement, and leave the H and S components unchanged.

[0071] Step (1-2) Luminance component enhancement. An improved Retinex algorithm is applied to the extracted luminance V component, and guided filtering is used to improve the illumination image estimation and enhance the image brightness.

[0072] It should be noted that the Retinex algorithm internally uses Gaussian filtering. Here, guided filtering is used to improve the detection of ArUco codes. Guided filtering is an edge-preserving filtering method based on a local linear model. Compared with Gaussian filtering, guided filtering has better performance in preserving image edges and details, and is more suitable for the recognition of images inside the fuselage of an aircraft in the low-light scene of this example.

[0073] Steps (1-3) Contrast Enhancement. The enhanced luminance V component is further enhanced using the CLAHE algorithm to improve contrast and improve the visibility of image details.

[0074] Steps (1-4) Overall brightness adjustment. Gamma correction is used to further optimize the overall brightness distribution, ensuring clear details in shadows while avoiding over-enhancement issues that may result from CLAHE.

[0075] Steps (1-5) Image Reconstruction. The enhanced V component is synthesized with the original H and S components, and then converted into an enhanced RGB image.

[0076] It should be noted that using CLAHE and Gamma correction to further enhance image contrast is an existing correction technique, and will not be elaborated on here.

[0077] Step 2: Visual localization of the enhanced image: Obtain the corner pixel coordinates and depth information through ArUco code recognition, and calculate its world coordinate system coordinates after coordinate system transformation.

[0078] ArUco codes are used for pose detection of the assembly hook and the assembly frame, and depth information is combined to optimize the accuracy of target 3D position and attitude estimation. The relative pose relationship between the assembly hook and the assembly frame is obtained through coordinate system transformation and hand-eye calibration, i.e., the output of the coordinate transformation.

[0079] ArUco code recognition yielded the corner pixel coordinates; see appendix. Figure 3 As shown, coordinate system transformation is described in the appendix. Figure 4 As shown, the relevant formula is:

[0080]

[0081] f in the above formula x and f y These represent the camera focal length in rows and columns of pixels, respectively. K1 represents the camera's intrinsic parameter matrix, containing information such as the optical center position and focal length. K2 represents the camera's extrinsic parameter matrix, where R is the rotation matrix and T is the translation vector, obtained through hand-eye calibration. (u,v) are pixel coordinates, Z... c This is the depth value of the ArUco code corner point in the camera coordinate system, obtained by the depth camera. (x) w y w , z w ) is the coordinate of the ArUco code corner point in the world coordinate system.

[0082] Step 3: Trajectory Planning: A 3-5-3 piecewise polynomial interpolation method in joint space is employed. Based on the pose, the joint positions of the input start point, path point, and end point are determined. The velocity and acceleration of the joint motion are calculated and output using piecewise polynomials to ensure that the assembly platform trajectory is continuous and smooth in terms of position, velocity, and acceleration, and meets the platform's dynamic constraints. See the appendix for details on the 3-5-3 piecewise polynomial trajectory planning simulation. Figure 9 As shown.

[0083] Step 4: Trajectory Optimization: An improved particle swarm optimization algorithm is introduced to optimize the trajectory execution time. An adaptive inertia weight update strategy is adopted to accelerate the convergence speed and improve trajectory efficiency.

[0084] The improved particle swarm optimization algorithm flowchart is attached. Figure 6 As shown.

[0085] To improve the search performance and convergence speed of particle swarm optimization, this paper studies an adaptive inertia weight adjustment method based on the change of optimal fitness value. The specific formula is as follows:

[0086]

[0087] In the formula, ω(t) is the inertia weight at the current time t; ω max It is the maximum inertial weight at the initial moment; ωmin It is the minimum inertia weight; f best f(t) is the optimal fitness value at time t; best (0) is the optimal fitness value at the initial moment.

[0088] The corresponding particle velocity update formula is improved as follows:

[0089]

[0090] In the formula, each particle i represents a possible solution, and its position is represented by the decision variable. Let k represent the number of updates, and d represent the d-th dimension of the search space. Each particle has a velocity vector. This represents the direction and step size of its movement in the d-th dimension of the search space. In each iteration, the particle needs to update its individual optimal position P. id And the global optimal position P of the entire population gd ω represents the inertia weight, which is dynamically variable and usually ranges from (0.4 to 0.9). It represents the particle's ability to maintain its current velocity, and the smaller the value, the stronger the individual's optimization ability. c1 and c2 represent learning factors, which are usually both 2, representing the degree of dependence of the particle on the individual optimal position and the group optimal position, respectively. r1 and r2 represent random factors, which are random numbers between [0 and 1], used to increase the randomness of the search.

[0091] In this method, the inertia weight is not a fixed value, but is dynamically adjusted according to the current fitness changes of the particle swarm. This adaptive adjustment can better balance global and local search. Maintaining a larger inertia weight in the early stages of the search enhances the particles' global exploration ability and avoids premature entrapment in local optima. In the later stages of the optimization process, as the optimal fitness value increases, the inertia weight gradually decreases, prompting the particle swarm to concentrate more closely around the current optimal solution for a more refined search, effectively improving the algorithm's convergence speed and accuracy.

[0092] The goal of time optimization for a six-DOF platform trajectory is to minimize the motion time while adhering to velocity and acceleration constraints. After 3-5-3 piecewise polynomial interpolation programming, the trajectory is divided into three segments, each with a time interval of T1, T2, and T3. An improved particle swarm optimization algorithm is used to optimize the trajectory, finding the shortest time for each segment, thus obtaining the optimal trajectory motion time. The objective function is expressed as:

[0093] f(t)=min(T1+T2+T3)

[0094] This function, also known as the fitness value function, is used to measure the quality of a particle. The smaller the fitness value of a particle (the smaller the objective function), the better the particle is.

[0095] Step 5: Control and Feedback: The system uses visual detection results as feedback. When the system moves to the path point, it performs visual positioning again, provides feedback on the current pose, dynamically corrects assembly posture errors, realizes multi-stage segmented control, and improves docking success rate and operational stability.

[0096] During assembly, for the assembly hook to dock with the assembly frame, multi-stage segmented control involves setting a secondary visual positioning position, acquiring the poses of both, and moving the assembly hook to make its pose the same as the pose of the assembly frame, thus completing the docking.

[0097] Instructions for using ArUco codes:

[0098] A 6×6 ArUco code is attached to each end of the ammunition rack to obtain the coordinates of the center points of the two slots on the rack. A 7×7 ArUco code is attached to each end of the ammunition hook to obtain the coordinates of the center points of the two lugs on the hook. The center points of the two slots on the rack are j1 and j2, and the center points of the two lugs on the hook are g1 and g2. Their three-dimensional coordinates in their respective nearest ArUco code coordinate systems are measured. After the camera detects the pose of the ArUco codes, the transformation matrix between the ArUco code coordinate system {M} and the world coordinate system {B} is obtained using the coordinate transformation relationship. The coordinates of points j1, j2, g1, and g2 in the world coordinate system can be solved as follows:

[0099]

[0100] Based on the principle that two points form a line, a line (j1, j2) with J as the midpoint and a line (g1, g2) with G as the midpoint can be obtained in the world coordinate system. The difference in attitude between the two lines and the difference in coordinates of the midpoints represent the amount of movement required for the grenades to be hooked. The usage of ArUco codes and a diagram illustrating the movement of the grenades are shown below. Figure 10 As shown.

[0101] In this implementation example, visual localization is performed again to provide feedback on the current pose, i.e., secondary pose detection and adjustment:

[0102] Once the ArUco code poses of the ammunition rack and the ammunition hook are detected, the required movement of the ammunition hook can be obtained. This movement is achieved by the motion of the six-degree-of-freedom platform. Because there is an offset between the ammunition hook and the end of the six-degree-of-freedom platform, when the angle of the six-degree-of-freedom platform changes, not only will the angle of the ammunition hook change, but its position will also change. A schematic diagram illustrating the positional change of the ammunition hook caused by the platform's angle change is shown below. Figure 11 As shown.

[0103] Taking the change in rotation angle as an example, assuming the perpendicular distance between the hook and the rotation axis of the six-degree-of-freedom platform is d, when the platform drives the hook to rotate at an angle α, the resulting change in position of the hook is:

[0104]

[0105] Due to differences in the projectile body and deviations in the release position each time, the offset of the loading hook and the end of the six-degree-of-freedom platform is not fixed. Furthermore, the movement angle during loading is small, resulting in minimal positional change. Therefore, this embodiment incorporates a secondary detection and attitude adjustment during automatic loading. A path point is set directly below the loading rack, where a second attitude detection is performed. At this point, the angle adjustment is complete, and the six-degree-of-freedom platform moves according to the new positional difference. This secondary detection and attitude adjustment not only avoids complex positional change calculations but also effectively eliminates various systematic errors, improving operational accuracy.

[0106] In some implementation examples, the ArUco code can be replaced by other visual tagging systems such as AprilTag and ARTag.

[0107] In addition, the image enhancement algorithm can employ deep learning methods or Retinex variant models as needed. The trajectory optimization section can select other intelligent optimization methods such as genetic algorithms or immune algorithms, depending on the platform's performance. For specific assembly instructions, please refer to the appendix. Figure 9 As shown.

[0108] Example 2

[0109] The purpose of this embodiment is to provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the above-described method.

[0110] Example 3

[0111] The purpose of this embodiment is to provide a computer-readable storage medium.

[0112] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the steps of the above method.

[0113] Example 4

[0114] The purpose of this embodiment is to provide an intelligent assembly system suitable for low-light environments, including:

[0115] The image processing module is configured to: acquire an image of the interior of the aircraft's fuselage, and perform enhancement processing on the acquired image of the interior of the aircraft's fuselage to obtain an enhanced image;

[0116] The pose data detection module is configured to: obtain pose data by detecting the pose of the assembly hook and the assembly frame based on the enhanced image;

[0117] The trajectory planning module is configured to: plan the trajectory of the aircraft's internal weapons loadout based on the obtained pose data and output the trajectory data.

[0118] The assembly module is configured such that the control platform, based on the output trajectory data, drives the assembly hook to hang the projectile onto the assembly frame.

[0119] Example 5

[0120] The purpose of this embodiment is to provide a computer program product containing instructions that, when run on a computer, cause the computer to perform the methods and functions involved in any of the above embodiments.

[0121] The steps and methods involved in the apparatus of the above embodiments correspond to those in Embodiment 1. For specific implementation details, please refer to the relevant description section of Embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood as including any medium capable of storing, encoding, or carrying an instruction set for execution by a processor and enabling the processor to perform any of the methods in this invention.

[0122] Those skilled in the art will understand that the modules or steps of the present invention described above can be implemented using general-purpose computer devices. Optionally, they can be implemented using computer-executable program code, thereby allowing them to be stored in a storage device for execution by a computer device, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. The present invention is not limited to any particular combination of hardware and software.

[0123] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. An intelligent assembly method suitable for low-light environments, characterized in that, include: If an image of the aircraft's interior is acquired under low light conditions and the ArUco code cannot be identified, the acquired image is enhanced to obtain an enhanced image that can identify the ArUco code. Specifically, the acquired image of the aircraft's interior is an RGB image; the RGB image is converted to HSV space, and the V component is extracted while the H and S components remain unchanged. The extracted brightness V component is processed using a Retinex algorithm based on guided filtering, combined with CLAHE and Gamma correction to improve image contrast. Pose data is obtained by detecting the pose of the assembly hook and the assembly rack based on the enhanced image with the ArUco code identified. Based on the obtained pose data, the trajectory of the internally mounted missiles on the aircraft fuselage is planned and the trajectory data is output. Based on the output trajectory data, the control platform drives the assembly hook to hang the projectile onto the assembly rack; The trajectory is optimized using an improved particle swarm optimization algorithm, and an adaptive inertia weight update strategy is adopted to accelerate the convergence speed and improve trajectory efficiency. The formula is as follows: In the formula, It is the inertia weight at the current time t; It is the maximum inertial weight at the initial moment; It is the minimum inertia weight; It is the optimal fitness value at the current time t; It is the optimal fitness value at the initial moment; During automatic ammunition loading operations, a secondary detection and attitude adjustment is designed. A path point is set directly below the ammunition rack, and a second attitude detection is performed at this point. At this time, the angle adjustment has been completed, and the six-degree-of-freedom platform moves according to the new position difference.

2. The intelligent assembly method for low-light environments as described in claim 1, characterized in that, The process of enhancing images acquired from inside the aircraft fuselage to obtain enhanced images also includes: The enhanced luminance V component improves contrast and enhances the visibility of image details; Optimize overall brightness distribution and adjust overall brightness; The enhanced V component is combined with the original H and S components and then converted into an enhanced RGB image.

3. The intelligent assembly method for low-light environments as described in claim 1, characterized in that, Acquire images of the interior of the aircraft fuselage. These images contain ArUco codes, which are affixed to the assembly hooks and assembly racks. The corner pixel coordinates and depth information are obtained by recognizing the ArUco code. After coordinate system transformation, the world coordinate system coordinates are obtained, and the relative pose relationship between the assembly hook and the assembly frame is obtained.

4. The intelligent assembly method for low-light environments as described in claim 1, characterized in that, The corner pixel coordinates are obtained by identifying the ArUco code, specifically including: The image is processed by grayscale conversion, image filtering, threshold segmentation, and edge detection. The ArUco code outline is filtered, the ArUco code perspective transformation is performed, and the ArUco code encoding is identified, ultimately obtaining the pixel coordinates of the ArUco code corner points.

5. The intelligent assembly method for low-light environments as described in claim 1, characterized in that, Based on the obtained pose data, trajectory planning is performed for the internal weapons mounted on the aircraft fuselage, and the trajectory data is output, specifically including: Based on the relative pose relationship between the assembly hook and the assembly frame, a piecewise polynomial interpolation method in joint space is adopted. The joint positions of the starting point, path point and the ending point are input, and the velocity and acceleration of the joint movement are calculated and output through piecewise polynomial.

6. The intelligent assembly method for low-light environments as described in claim 1, characterized in that, It also includes trajectory optimization steps: The goal of time optimization for a six-DOF platform trajectory is to minimize the motion time while adhering to velocity and acceleration constraints. The trajectory, after piecewise polynomial interpolation programming, is divided into three segments, each requiring a specific time. , , An improved particle swarm optimization algorithm is used to optimize the trajectory, finding the shortest time for each segment of the trajectory, thus obtaining the optimal time for trajectory motion. The objective function is expressed as: This function, also known as the fitness value function, is used to measure the quality of a particle. The smaller the fitness value of a particle, that is, the smaller the objective function value, the better the particle is.

7. An intelligent assembly system suitable for low-light environments, employing the intelligent assembly method for low-light environments as described in any one of claims 1-6, characterized in that, include: The image processing module is configured to: acquire an image of the interior of the aircraft's belly; if the image of the interior of the aircraft's belly is acquired under low light conditions and the ArUco code cannot be recognized, then perform enhancement processing on the acquired image of the interior of the aircraft's belly to obtain an enhanced image so that the ArUco code can be recognized. The pose data detection module is configured to: obtain pose data by detecting the pose of the assembly hook and the assembly frame based on the enhanced image that has been identified with ArUco code; The trajectory planning module is configured to: plan the trajectory of the aircraft's internal weapons loadout based on the obtained pose data and output the trajectory data. The assembly module is configured such that the control platform, based on the output trajectory data, drives the assembly hook to hang the projectile onto the assembly frame.

8. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 6.

9. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method described in any one of claims 1-6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it performs the steps of the method described in any one of claims 1-6 above.

Images