A time-optimal motion planning method and system for display panel defect flyback detection

Abstract

The present application belongs to the technical field of automatic control and visual detection, and particularly relates to a time-optimal motion planning method and system for display panel defect flying shot detection. The present application firstly adopts a clustering algorithm to plan flying shot points, and combines multiple defects that can be covered in the same field of view into one detection unit, with the center of the detection unit being the flying shot point. The maximum allowed scanning speed when the platform uniformly passes through the flying shot point during the flying shot process is derived, and on this basis, an adaptive segmented speed planning strategy is designed. Then, displacement-time nonlinear mapping relationship functions of X-axis and Y-axis are respectively established, and the total motion time required to traverse all flying shot points is minimized as an optimization objective, so that the flying shot point access sequence is globally optimized to generate a time-optimal detection path. Finally, a prediction compensation model is established in combination with system communication delay and camera response time, the camera exposure is triggered in advance when the camera is about to reach the flying shot point, and high-quality images are collected.

Description

Technical Field

[0001] This invention belongs to the field of automatic control and visual inspection technology, and in particular relates to a time-optimal motion planning method and system for detecting defects in display panels by aerial photography. Background Technology

[0002] In the production of display panels, defect detection is one of the key processes to ensure panel yield. As the display industry moves towards higher generations and higher resolutions, the increasing size of panels places higher demands on inspection equipment. Current inspection processes typically employ a two-stage method: the first stage uses scanning imaging and comparison with the original design to identify suspected defects and their locations; the second stage uses a microscope camera to confirm the authenticity of each defect and classify them. After initially marking the locations of suspected defects, the microscope camera, driven by a high-precision gantry motion platform, moves sequentially to each suspected defect location for imaging. The platform's motion planning during this process determines the system's inspection efficiency and image acquisition quality.

[0003] While the industry has widely adopted aerial photography technology and the traveling salesman problem algorithm for path optimization, the following problems still exist when detecting multiple suspected defects on large-size panels: First, there is a lack of planning for the microscopic imaging field of view. When adjacent suspected defects are very close, the camera module needs to perform unnecessary point-to-point movements. Second, to ensure that the image does not produce ghosting, the current aerial photography technology generally keeps the platform moving at a constant speed throughout the process. However, this speed is limited by the camera exposure time and pixel resolution. In the long-stroke jumps of large-size panel detection, it is necessary to make full use of the platform's acceleration and deceleration capabilities to further improve detection efficiency. Third, existing path planning usually only considers the geometric shortest distance between points, ignoring the dynamic constraints of the moving platform and the requirements for imaging stability. It fails to consider the differences in acceleration and deceleration capabilities and maximum speeds between the X and Y axes in the gantry structure, making it difficult to reflect the actual movement time of the platform under real dynamic constraints. Summary of the Invention

[0004] The purpose of this invention is to propose a time-optimal motion planning method and system for detecting defects in display panels by taking photos, so as to solve one or more technical problems existing in the prior art, and at least provide a beneficial option or create conditions.

[0005] A time-optimal motion planning method for detecting defects in display panels by aerial photography, the method comprising the following steps: After receiving the coordinate information of suspected defects from the front-end detection device, the system performs data standardization on the coordinates, and then performs clustering on the suspected defect points on the display panel. Multiple suspected defects that can be covered within the same field of view are merged into one detection unit, with the center of the detection unit being the camera point.

[0006] Based on the imaging principle, the maximum allowable scanning speed of the platform passing through the shooting point at a constant speed during the shooting process is derived as a global unified constraint. Considering the discrete distribution characteristics of the shooting point on the panel, this invention designs an adaptive segmented speed planning strategy and determines the critical distance threshold of its adaptive speed planning according to the different dynamic parameters of the two axes.

[0007] Based on the aforementioned speed planning strategy and gantry platform, displacement-time nonlinear mapping relationships for the X and Y axes under different dynamic parameters are constructed respectively. With the goal of minimizing the total motion time required to traverse all shooting points, the optimal shooting point access order is generated. Furthermore, time parameterization is performed on each path segment to generate a continuous trajectory with dual-axis linkage, ensuring that both axes can reach the shooting point simultaneously.

[0008] During the uniform motion of aerial photography, this invention establishes a predictive compensation model to trigger the camera exposure in advance, ensuring that the platform completes the acquisition of clear images while in a stable uniform motion state.

[0009] Preferably, the method is based on a camera point planning module, a motion planning module, and a control and synchronization triggering module. After receiving the list of suspected defect coordinates from the previous detection output, the system transmits it to the camera point planning module. The camera point planning module uses a clustering algorithm to plan camera points according to the distribution of suspected defect points on the panel, merging multiple defects that can be covered within the same field of view into one detection unit. The center of the detection unit is the camera point.

[0010] Preferably, after the aerial shooting point is planned, the coordinate data is transmitted to the motion planning module. The motion planning module first determines the maximum permissible speed during aerial shooting, and based on the imaging principle, constructs a relationship function between speed, exposure time, and camera parameters to derive the maximum permissible speed during shooting. , These are the unified constraints that the platform must meet when passing through the aerial photography points. instantaneous velocity Based on the aforementioned constraints, a speed planning strategy is determined, according to the maximum permissible speed. A distance threshold is set based on the platform's own dynamic parameters. When the distance traveled is less than this threshold, a constant speed is maintained for jumping to avoid vibration caused by short-distance acceleration and deceleration. When the distance traveled is greater than this threshold, an asymmetric S-shaped acceleration and deceleration strategy is adopted. That is, a larger acceleration is used in the acceleration phase to quickly reach the platform's maximum speed to shorten the time, while a smaller acceleration is used in the deceleration phase to achieve a smooth transition to a constant speed shooting state, ensuring that there is no residual vibration of the camera module during shooting.

[0011] Preferably, based on the speed planning strategy and the measured dynamic performance of the gantry platform and measuring head, displacement-time nonlinear mapping models for the X and Y axes are established respectively. The displacement-distance nonlinear mapping functions for the X and Y axes are pre-calculated offline. Inputting any axial displacement quickly outputs the theoretical time required for that axis to complete the aforementioned motion strategy. Based on this, time-optimal multi-point global path planning is performed, with the objective function being minimizing the total time to traverse all shooting points. When calculating the movement time cost between any two shooting points, the actual time consumed by the path segment depends on the slower-moving axis. Through iterative optimization, the system outputs the optimal access sequence and estimated total movement time for all bidding points. Based on the generated access sequence and speed planning strategy, a time-parameterized dual-axis linkage trajectory is generated for each path segment to ensure that the X and Y axes can move to the next bidding point simultaneously. The previously calculated slow axis time is used as the reference time for this segment. For the faster-moving axis, time scaling is used to map it onto the reference time axis, thereby generating the trajectory sequence of the target position, instantaneous velocity, and acceleration command of the X and Y axes at each discrete moment.

[0012] Preferably, the generated trajectory sequence file is transmitted to the control module for execution. A high-real-time motion control card is used to perform high-frequency trajectory tracking and servo control after reading the trajectory sequence file. Simultaneously, a closed-loop control system is formed by combining high-precision grating ruler feedback to ensure that the actual motion trajectory accurately tracks the planned trajectory. During the system's uniform motion for aerial photography, the synchronization trigger module establishes a predictive compensation model based on the camera's real-time position and speed information, combined with system communication latency and camera response time. The system triggers camera exposure in advance just before reaching the aerial photography point, and the platform completes the acquisition of clear images while in a stable, uniform motion state. The acquired images are transmitted to a host computer via a high-speed network for storage and subsequent defect screening and automatic classification processing.

[0013] A time-optimal motion planning system for detecting defects in display panels by aerial photography is disclosed. The system is a high-precision OLED display panel re-inspection system, which includes a host computer, a high real-time motion control card, a precision gantry motion platform, a microscope camera and its trigger control device. The method is applied to the high-precision OLED display panel re-inspection system.

[0014] The beneficial effects of this invention are as follows: This invention systematically integrates a series of algorithms, including aerial shooting point planning, adaptive speed planning, time-optimal global path planning, dual-axis linkage trajectory generation, and predictive compensation triggered imaging, to achieve high-efficiency and high-quality aerial shooting detection of suspected defects in display panels. Through aerial shooting point planning based on density clustering algorithms, multiple suspected defects that can be covered by the same field of view in space are merged into a single detection unit, effectively reducing the number of camera jump shots and unnecessary movement, lowering path planning complexity, and improving overall motion efficiency. Furthermore, considering the motion stability requirements of aerial shooting, a maximum allowable speed constraint under uniform shooting conditions is constructed, and an adaptive segmented speed planning strategy is designed based on the platform's dynamic characteristics. This allows the platform to avoid mechanical disturbances caused by frequent acceleration and deceleration during short-distance movement, and to fully utilize acceleration and deceleration capabilities to shorten movement time during long-distance movement, thereby further improving detection efficiency while ensuring image quality. This invention establishes a nonlinear mapping relationship between the displacement and motion time of each motion axis, using the minimization of total detection time as the global path planning objective. This avoids the problem of poor actual performance in traditional geometric distance-based optimization methods that fail to consider performance differences between different axes, achieving time-optimal path planning that better suits real-world detection scenarios. Simultaneously, through time-parameterized dual-axis linkage trajectory generation and closed-loop motion control, each motion axis reaches the target position simultaneously under a unified time reference, ensuring motion continuity. Combined with a shooting trigger compensation mechanism, the camera can complete precise exposure while the platform is in a stable, uniform motion state, ensuring the clarity of the microscopic image. In summary, this invention effectively improves detection efficiency, motion stability, and imaging reliability without increasing hardware costs. It can adapt to display panels of different sizes and defect distribution scenarios, possessing significant engineering practical value. Attached Figure Description

[0015] The above and other features of the present invention will become more apparent from the detailed description of the embodiments shown in conjunction with the accompanying drawings. In the accompanying drawings, the same reference numerals denote the same or similar elements. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without any creative effort. In the drawings: Figure 1 This is an overall flowchart of a time-optimal motion planning method for aerial photography detection of defects in display panels. Figure 2 A schematic diagram of the camera point planning for a time-optimal motion planning method for camera detection of display panel defects; Figure 3 This is a schematic diagram of the motion planning velocity curve for a time-optimal motion planning method for detecting defects in display panels. Detailed Implementation

[0016] The following will provide a clear and complete description of the concept, specific structure, and technical effects of the present invention in conjunction with the embodiments and accompanying drawings, so as to fully understand the purpose, solution, and effects of the present invention. It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other.

[0017] In the description of this invention, "several" means one or more, "more than" means two or more, "greater than," "less than," and "exceeding" are understood to exclude the stated number, while "above," "below," and "within" are understood to include the stated number. The use of "first" and "second" in the description is merely for distinguishing technical features and should not be construed as indicating or implying relative importance, or implicitly indicating the number of indicated technical features, or implicitly indicating the order of the indicated technical features.

[0018] This embodiment operates in a high-precision OLED display panel re-inspection system, which includes a host computer (PC), a high real-time motion control card, a precision gantry motion platform, a microscope camera, and its trigger control device.

[0019] like Figure 1 As shown, a time-optimal motion planning method for detecting defects in display panels by aerial photography is provided. The method includes the following steps: After receiving the coordinate information of suspected defects from the front-end detection device, the system performs data standardization on the coordinates, and then performs clustering on the suspected defect points on the display panel. Multiple suspected defects that can be covered within the same field of view are merged into one detection unit, with the center of the detection unit being the camera point.

[0020] Based on the imaging principle, the maximum allowable scanning speed of the platform passing through the shooting point at a constant speed during the shooting process is derived as a global unified constraint. Considering the discrete distribution characteristics of the shooting point on the panel, this invention designs an adaptive segmented speed planning strategy and determines the critical distance threshold of its adaptive speed planning according to the different dynamic parameters of the two axes.

[0021] Based on the aforementioned speed planning strategy and gantry platform, displacement-time nonlinear mapping relationships for the X and Y axes under different dynamic parameters are constructed respectively. With the goal of minimizing the total motion time required to traverse all shooting points, the optimal shooting point access order is generated. Furthermore, time parameterization is performed on each path segment to generate a continuous trajectory with dual-axis linkage, ensuring that both axes can reach the shooting point simultaneously.

[0022] During the uniform motion of aerial photography, this invention establishes a predictive compensation model to trigger the camera exposure in advance, ensuring that the platform completes the acquisition of clear images while in a stable uniform motion state.

[0023] like Figure 2As shown, the aerial photography site planning and implementation process The main task of aerial photography point planning is to perform spatial clustering on the original defect coordinates issued by the previous process, group defect points with similar physical locations into the same inspection field of view (FOV), find the best aerial photography position, and reduce the number of point movements of the motion platform.

[0024] The system first receives a list of suspected defect coordinates from the upstream inspection device. The logical coordinates based on the panel CAD drawings are mapped to the physical coordinates of the gantry motion platform using a pre-calibrated transformation matrix. Based on the requirements of the detection task, the current field-of-view parameters of the microscope camera are called, and its effective imaging range is set to... .

[0025] Based on standardized coordinate data, this invention employs an improved DBSCAN clustering algorithm to perform spatial clustering of suspected defect points. First, the neighborhood radius of the algorithm is set. To ensure that all points within a cluster can stably fall within the central region of the field of view during imaging, the short side of the field of view is reserved to a minimum of 40% of the field of view of the microscope camera.

[0026] The algorithm iterates through the coordinates and calculates the distances between them. Points within the range are divided into the same cluster. Isolated points that cannot be assigned to any cluster are marked as single-point tasks. To prevent defects located at the edge of the field of view from being missed, a field of view overlap coefficient is introduced. The value is 10% to ensure necessary pixel overlap between adjacent fields of view.

[0027] For each cluster of points generated by clustering ={ The system determines the capture point of the detection unit by calculating the geometric centroid of all points within the cluster. : ; After the above processing, the original n discrete defect points are transformed into m aerial photography points ( This module will generate a set of task points. The original defect information, along with its associated information, is transmitted to the motion planning module. , The displacement-time nonlinear mapping relationship models belonging to the X and Y axes respectively. , , The displacement-time nonlinear mapping relationship models belonging to the X and Y axes respectively. k is a point cluster The total number of elements.

[0028] Through this process, the system provides basic data for solving subsequent path planning while ensuring detection coverage.

[0029] like Figure 3 As shown, velocity planning and global path planning After completing the aerial photography point planning and obtaining the aerial photography task point set, the system enters the motion planning phase. Specifically, under the premise of meeting the dynamic constraints of the motion platform and the stability requirements of aerial photography imaging, an accurate evaluation model of the motion time between aerial photography points is established to determine the optimal access order of aerial photography points and generate the corresponding dual-axis linkage motion trajectory to minimize the total detection time.

[0030] First, determine the maximum permissible speed for aerial scanning. This serves as a uniform upper limit constraint for the platform when passing through all photography points, and as a boundary condition for subsequent speed planning and path optimization. The system is based on the pixel resolution of the current microscope camera. Exposure time that meets photosensitivity requirements and the maximum allowable pixel ghosting amount The uniform scanning speed of the computing platform as it passes through each aerial photography point satisfies: ; Faced with the random distribution of defects on large-size panels, characterized by localized clustering and global sparseness, if the distance between adjacent camera points is too short, the acceleration and deceleration phases of the S-curve require physical space. Performing a complete "acceleration-uniform speed-deceleration" process in this situation would force the platform to decelerate immediately before completing acceleration, easily triggering mechanical vibrations without buffer time, thus reducing imaging stability. Conversely, if the distance between camera points is large, maintaining a constant speed throughout the process... Low-speed operation wastes the high-speed motion performance of the gantry crane, increases non-productive time, and reduces overall motion efficiency. Therefore, this invention designs an adaptive speed planning strategy. The system executes different speed planning strategies based on different displacement conditions, such as... Figure 3 As shown, uniform motion is used between point 1 and point 2, while an asymmetric acceleration / deceleration strategy is used between point 2 and point 3. This strategy forms the basis for subsequent path planning and trajectory planning.

[0031] First, the system will calculate the critical distance threshold for switching motion modes based on the dynamic parameters of each axis of the platform. Since the dynamic parameters of the two axes on the gantry are different, their critical thresholds need to be calculated separately, but the steps are the same. Here, we take the X-axis as an example. First, let's set the maximum allowable speed of the X-axis motor. Maximum acceleration The acceleration during the acceleration phase is The acceleration during the deceleration phase is The start and end speeds are both the scanning speed of the flying camera. The core issue lies in solving for the scanning speed of the flying camera. Accelerate to the platform's maximum speed slow down and return The minimum required physical displacement, which serves as the critical distance threshold. .

[0032] Specifically, the first step is to calculate the scanning speed from the drone. to maximum speed The change in velocity between ; After learning the speed increment In this case, it is necessary to further determine whether the maximum acceleration can be reached. The judgment condition is: ; When the above conditions are met, the acceleration process adopts an S-shaped velocity curve with limited jerk and saturated acceleration. The acceleration process consists of three segments: the jerk increase segment, the uniform acceleration segment, and the jerk decrease segment.

[0033] The displacement during the acceleration phase can be expressed as: ; Similarly, the motion process of the deceleration phase is similar to that of the acceleration phase, but different jerk parameters are used to make the motion curve smoother. The displacement of the deceleration phase can be expressed as: ; The minimum displacement required for both acceleration and deceleration phases is defined as follows: ; When the displacement of adjacent shooting points in the X-axis direction satisfies: ; If the platform is unable to complete the full acceleration and deceleration process, the system automatically adopts a constant-speed flying camera motion strategy to complete the jump. When: ; The system adopts an asymmetric acceleration and deceleration planning strategy to make full use of the platform's high-speed motion capability and ensure imaging stability.

[0034] After determining the critical distance threshold of the platform under given dynamic constraints, the motion process between adjacent camera points can be divided into different motion modes. To quickly evaluate the motion time between any two camera points during the path planning phase, this invention further establishes a nonlinear mapping relationship between axial displacement and the time required to complete that displacement. This mapping relationship is pre-built using offline analytical modeling and is used for subsequent calculation of the time cost of global path planning. The following explanation still uses the X-axis as an example.

[0035] Within a short distance, the system employs a constant-speed scanning strategy, with the platform consistently maintaining a scanning speed. Complete this displacement segment. In this case, the displacement D has a linear relationship with time T, and its displacement-time mapping function is expressed as: ; In long-distance intervals, the platform movement process includes from Accelerate to acceleration phase, with The middle section of uniform motion, from Decelerate The deceleration phase consists of three parts. Therefore, the displacement-time mapping function over a long distance can be expressed as: ; in The time required for the acceleration phase, This is the time required for the deceleration phase.

[0036] Since the X and Y axes of a gantry platform are usually not completely consistent in terms of dynamic parameters and maximum speed, this invention establishes the above-mentioned displacement-time mapping functions for the X and Y axes respectively: ; For any two given flying camera points and The axial displacements of its two shafts can be calculated as follows: ; After modeling the dual-axis displacement-time mapping function, the system can transform the traditional path planning problem based on geometric distance into an optimization problem based on actual motion time. Since both axes need to reach the camera point simultaneously during the platform's coordinated motion, the actual completion time of a path segment is determined by the slower axis. Therefore, the path is defined as starting from the camera point... Move to the drone shooting point The time cost is: ; The set of task points generated by the above aerial photography point planning is as follows This invention constructs a time-optimal path planning model with the constraint of traversing all shooting points without revisiting them, and the optimization objective of minimizing the total movement time. The objective function is: ; in An arrangement representing the order in which aerial photography sites are visited.

[0037] Since this type of time-optimal path planning problem is a typical nondeterministic polynomial complexity problem when there are a large number of points, it is difficult to solve directly using analytical methods. Heuristic algorithms such as genetic algorithms and ant colony algorithms can be used to solve the above model. Through any of the above solution methods, the system finally outputs the time-optimal or near-time-optimal access order of all shooting points, as well as the corresponding path segment time information, providing basic input for subsequent dual-axis linkage trajectory generation and motion control.

[0038] Dual-axis linkage trajectory generation and execution After completing the optimal access sequence planning for multiple locations, the system can determine the positions of two adjacent aerial photography points. for and Therefore, the corresponding axial displacement can be obtained: ; Based on the aforementioned displacement-time nonlinear mapping function, the system calculates the theoretical time required for the X-axis and Y-axis to complete motion under the current displacement conditions: ; The time function used The above derivation has been carried out, taking into account the maximum allowable speed of the drone, the maximum speed of the platform, acceleration and jerk constraints, and satisfying the above adaptive piecewise speed planning strategy.

[0039] Due to the different moments of inertia and dynamic characteristics of the two axes of the gantry platform, the time required for each axis to complete its respective displacement within the same time frame is usually unequal. To ensure accurate alignment of the platform at the shooting point, this invention uses the axis with the longer execution time as the time reference axis for that segment of motion, defining the uniform execution time for this path segment as: ; In determining a unified time benchmark Then, the system performs time scaling on the faster-moving axis so that it can complete the target displacement in the same time as the slower axis without violating its own dynamic constraints.

[0040] Taking the X-axis as the fast axis as an example, when the X-axis's motion time... At this time, a time scaling factor is introduced: ; The velocity, acceleration, and jerk commands on the X-axis are scaled according to the following relationship: ; in , , Plan curves for the original velocity, acceleration, and jerk. This is the actual execution curve after time scaling.

[0041] After the above processing, both the X-axis and Y-axis are within the same time interval. Each component completes its own displacement within the time frame. After achieving dual-axis time unification, the system discretizes the trajectory according to the controller's interpolation period. Let the control system interpolation period be... Then at each discrete moment: ; The system calculates and outputs the corresponding trajectory instructions: ; The aforementioned trajectory data constitutes a complete dual-axis linkage trajectory segment file, which is sequentially spliced ​​according to the path planning order to form a continuous motion trajectory covering all shooting points. The generated trajectory file is transmitted to the motion controller, which performs real-time trajectory tracking and execution at a fixed interpolation frequency (e.g., 1 kHz). The controller sends the target position and velocity commands at each moment to the X-axis and Y-axis servo drives, while simultaneously receiving actual position data from a high-precision grating ruler, forming a fully closed-loop control. During the control process, the controller continuously compares the target position with the feedback position error and corrects it through cascaded control of the position loop, velocity loop, and current loop, ensuring that the actual motion trajectory follows the planned trajectory with high precision, thereby achieving stable and continuous dual-axis motion throughout the entire shooting process.

[0042] Synchronous Triggering and Image Acquisition Due to time delays in platform motion control, communication, and camera exposure, this invention proposes a trigger timing calculation model based on motion prediction. Let the total system delay be: ; in For communication delay time, For camera exposure delay time, The delay time for the control system execution.

[0043] The system's real-time prediction platform is in The system detects the future location and sends a trigger signal in advance when the distance between the predicted location and the next target point is within a set range, so that the actual exposure occurs at the shooting point.

[0044] The system uses an EtherCAT real-time bus for high-speed communication. The main control card sends position feedback packets at a 1 kHz control cycle, and the trigger module performs timing alignment through a timestamp synchronization mechanism. After exposure and shooting, the camera sends the acquired image frames and their timestamps to the host computer via Gigabit Ethernet.

[0045] Although the invention has been described in considerable detail and particularly with regard to several of the described embodiments, it is not intended to limit itself to any of these details or embodiments or any particular embodiment, thereby effectively covering the intended scope of the invention. Furthermore, the invention has been described above with respect to embodiments foreseeable by the inventors in order to provide a useful description, and non-substantial modifications to the invention that have not yet been foreseen may still represent equivalent modifications.

Claims

1. A time-optimal motion planning method for detecting defects in display panels using aerial photography, characterized in that, The method includes the following steps: The coordinates of suspected defects output by the front-end detection device are used to plan the shooting points using a density clustering algorithm; The maximum permissible speed for uniform shooting is determined based on aerial imaging constraints, and the critical distance threshold for adaptive speed planning is determined based on the dynamic performance of each axis. Global path planning is achieved by constructing nonlinear displacement-time mapping relationships along the X and Y axes respectively and minimizing the total time for traversing all shooting points. The path is parameterized by time to generate a continuous trajectory with dual-axis linkage. This is combined with a high real-time motion control card and a high-precision grating ruler to achieve closed-loop control. A predictive compensation model is established to address system latency, thereby triggering the camera exposure platform to complete image acquisition in advance under a uniform and stable state.

2. The time-optimal motion planning method for detecting defects in a display panel according to claim 1, characterized in that, The method is based on a camera point planning module, a motion planning module, and a control and synchronization triggering module. After receiving the list of suspected defect coordinates from the previous detection output, the system transmits it to the camera point planning module. The camera point planning module uses a clustering algorithm to plan camera points according to the distribution of suspected defect points on the panel, merging multiple defects that can be covered within the same field of view into one detection unit. The center of the detection unit is the camera point.

3. The time-optimal motion planning method for detecting defects in a display panel by aerial photography according to claim 2, characterized in that, After the aerial photography point is planned, the coordinate data is transmitted to the motion planning module. The motion planning module first determines the maximum permissible speed during aerial photography, and based on imaging principles, constructs a relationship function between speed, exposure time, and camera parameters to derive the maximum permissible speed during shooting. , These are the unified constraints that the platform must meet when passing through the aerial photography points. instantaneous velocity ; Based on the aforementioned constraints, a speed planning strategy is determined, according to the maximum permissible speed. A distance threshold is set based on the platform's own dynamic parameters. When the distance traveled is less than this threshold, a constant speed is maintained for jumping to avoid vibration caused by short-distance acceleration and deceleration. When the distance traveled is greater than this threshold, an asymmetric S-shaped acceleration and deceleration strategy is adopted. That is, a larger acceleration is used in the acceleration phase to quickly reach the platform's maximum speed to shorten the time, while a smaller acceleration is used in the deceleration phase to achieve a smooth transition to a constant speed shooting state, ensuring that there is no residual vibration of the camera module during shooting.

4. The time-optimal motion planning method for detecting defects in a display panel by aerial photography according to claim 1, characterized in that, Based on the velocity planning strategy and the measured dynamic performance of the gantry platform and measuring head, displacement-time nonlinear mapping models for the X and Y axes are established respectively. The displacement-distance nonlinear mapping functions for the X and Y axes are pre-calculated offline. By inputting any axial displacement, the theoretical time required for that axis to complete the aforementioned motion strategy can be quickly output. Based on this, time-optimal multi-point global path planning is performed, with the objective function being to minimize the total time to traverse all shooting points. When calculating the movement time cost between any two shooting points, the actual time consumption of the path segment depends on the slower-moving axis. Through iterative optimization, the system outputs the optimal access sequence and estimated total movement time for all bidding points. Based on the generated access sequence and speed planning strategy, a time-parameterized dual-axis linkage trajectory is generated for each path segment to ensure that the X and Y axes can move to the next bidding point simultaneously. The previously calculated slow axis time is used as the reference time for this segment. For the faster-moving axis, time scaling is used to map it onto the reference time axis, thereby generating the trajectory sequence of the target position, instantaneous velocity, and acceleration command of the X and Y axes at each discrete moment.

5. The time-optimal motion planning method for detecting defects in a display panel by aerial photography according to claim 1, characterized in that, The generated trajectory sequence file is transmitted to the control module for execution. A high-real-time motion control card is used to read the trajectory sequence file and perform high-frequency trajectory tracking and servo control. Simultaneously, a closed-loop control system is formed by combining high-precision grating ruler feedback to ensure that the actual motion trajectory accurately tracks the planned trajectory. During the system's uniform motion for aerial photography, the synchronization trigger module establishes a predictive compensation model based on the camera's real-time position and speed information, combined with system communication latency and camera response time. The system triggers camera exposure in advance just before reaching the photography point, and the platform completes the acquisition of clear images while in a stable, uniform motion state. The acquired images are transmitted via a high-speed network to a host computer for storage, subsequent defect screening, and automatic classification processing.

6. A time-optimal motion planning system for detecting defects in display panels using aerial photography, characterized in that, The system is a high-precision OLED display panel re-inspection system, which includes a host computer, a high real-time motion control card, a precision gantry motion platform, a microscope camera and its trigger control device. The time-optimal motion planning method for display panel defect aerial photography detection according to claims 1-5 is implemented in the high-precision OLED display panel re-inspection system.

Images