Micro device end face feature visual guidance high precision alignment system and method
By using an orthogonal vision architecture and interventional prism technology, the problems of blind spots and multi-source errors in the observation of micro-device end-face features are solved, achieving high-precision alignment of device end-face features and improving the assembly accuracy and efficiency of micro-device packaging.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SOUTH CHINA UNIV OF TECH
- Filing Date
- 2026-03-16
- Publication Date
- 2026-06-09
Smart Images
Figure CN122172499A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the fields of precision assembly and machine vision technology, and more specifically, to a high-precision alignment system and method guided by visual features of the end face of a micro-device. Background Technology
[0002] With the rapid development of microelectromechanical systems (MEMS), microfluidic chips, and photonic devices, the precision requirements for the packaging and assembly of micro-devices are increasing. The cross-sectional dimensions of fluid channels in microfluidic chips are typically around 10 μm or even smaller. Similarly, the feature dimensions of microelectrode contacts used for lateral interconnection between MEMS sensors and application-specific integrated circuits (ASICs) are also on the order of 10 μm. Furthermore, the mode field diameter of single-mode waveguides in photonic devices is typically less than 10 μm. Achieving precise alignment of the end-face features of two micro-devices—such as microchannel openings, microelectrode contacts, and waveguide ports—is crucial for ensuring leak-free fluid transmission, reliable electrical signal interconnection, and low-loss optical coupling. It is also a core technological bottleneck restricting device packaging yield and system performance.
[0003] Currently, the end-face assembly and alignment of micro-devices mainly rely on the following two mainstream technical methods.
[0004] The first type is active alignment or dielectric-assisted alignment. This method applies an excitation signal to the device during alignment, such as introducing fluid into a microchannel or applying a test signal to the electrode. It iteratively searches for the optimal alignment position by real-time monitoring of physical quantities such as flow rate, optical power, or electrical signal strength at the output. However, this method is essentially a multi-degree-of-freedom blind search process lacking direct visual position information. End-face alignment typically involves at least three translational degrees of freedom and one rotational degree of freedom. When the end-face feature size is on the order of 10 μm, the peak region of the signal response is narrow, requiring a correspondingly smaller search step size, leading to a significant increase in the number of iterations and a long alignment time per attempt. For multi-channel microchannel arrays or multi-contact electrode arrays, the coupling effect of rotational angle deviation can easily cause the system to get trapped in local optima, requiring repeated backtracking and re-searching, further reducing alignment efficiency. Furthermore, this method requires continuous application of fluid or electrical signal excitation during alignment, increasing process complexity and limiting its implementation in packaging scenarios with strict cleanliness or electrostatic discharge protection requirements. These factors make active alignment methods difficult to meet the efficiency demands of modern microdevice mass production and high-cycle automated packaging.
[0005] The second method is an indirect passive visual alignment method based on markings on the upper surface of the device. This method uses a microscope camera positioned directly above the device to observe alignment marks on the upper surface. The coordinates of the marks are determined through image recognition, and then the position of the end-face features is indirectly determined by calculating the coordinates based on a pre-defined spatial geometric relationship between the marks and the end-face features. A glass micro / nanofluidic chip alignment and assembly method and device (CN101691203A) mainly utilizes the semi-transparent nature of the glass micro / nanofluidic chip to obtain image information of the chip's internal channels or related structures from a top-down perspective, thereby achieving positioning and alignment control during chip assembly. This type of method is applicable to semi-transparent chips whose internal structures can be clearly observed from a top-down perspective. However, it is highly dependent on the device's light transmittance and top-down observability. When the device to be aligned does not have good light transmittance, or when the alignment channel cannot be clearly observed from a top-down perspective due to device structure, packaging, or clamping obstructions, this type of method will struggle to obtain effective alignment images, or even fail to achieve accurate alignment. Furthermore, this type of method does not directly observe the alignment features on the opposite end faces of the two devices, making it difficult to directly reflect the true relative positional relationship between the end face channels. However, the accuracy of this method is fundamentally limited by the stability of the spatial relationship between the upper surface markings and the end face features. In actual manufacturing, when chips are separated from wafers by dicing, factors such as blade wear, feed offset, and material breakage can cause random deviations of several micrometers to tens of micrometers relative to the internal microstructure of the device's physical edge. For microfluidic chips manufactured using multilayer bonding processes, the superposition of interlayer alignment errors and dicing errors further increases the uncertainty of the actual offset between the upper surface markings and the end face features. These multi-source cumulative errors mean that even if the top vision system can identify the marking position with sub-micrometer accuracy, the deviation introduced by coordinate calculation can easily exceed the alignment tolerance required for end face features on the order of 10μm. As a result, the surface markings of the two devices may appear aligned when viewed from the top, but the microchannel openings or microelectrode contacts on the end faces may actually be misaligned, leading to assembly defects such as fluid leakage or channel mismatch. This limitation in accuracy is a fundamental bottleneck in indirect calculation methods and cannot be eliminated simply by improving the resolution of the vision system or the accuracy of the motion platform.
[0006] From the perspective of error propagation theory, directly extracting minute feature points from the end faces of two devices, using the actual positions of these feature points as alignment references, and establishing a real-time visual feedback closed loop is an effective way to eliminate accumulated errors in intermediate stages and achieve high alignment accuracy. In this mode, alignment accuracy depends only on the vision system's ability to locate the end face features and the positioning accuracy of the motion platform, and is no longer affected by the superposition of external uncertainties such as cutting tolerances and interlayer deviations. Simultaneously, the vision system can directly obtain the relative positional deviation information of the two end face features, allowing the motion control system to implement deterministic closed-loop corrections without the need for multi-dimensional iterative searches, thus significantly improving alignment efficiency.
[0007] However, direct visual observation of minute features on the 10μm scale faces stringent physical limitations in engineering practice. Clearly distinguishing and locating 10μm-scale end-face features requires a high-magnification microscopic optical system; however, high-magnification microscope objectives inevitably come with inherent constraints such as short working distance and shallow depth of field. In actual docking operations, the gap between the end faces of two devices is typically only a few hundred micrometers or even smaller, making it difficult for conventional microscope lenses to penetrate such narrow spaces due to their physical size limitations. Even when observing from a distance at the side, the line of sight is easily obstructed by the device body and clamping mechanism, creating blind spots. Simultaneously, illumination light cannot effectively enter the narrow gap, resulting in insufficient image contrast and signal-to-noise ratio, making it difficult to meet the requirements for feature extraction. These limitations, determined by the laws of optical physics and geometric spatial conditions, constitute a long-standing technical barrier restricting direct visual observation and closed-loop alignment of micro-features on end faces.
[0008] Therefore, there is an urgent need for a system and method that can overcome the blind zone of end face observation, achieve high-resolution direct imaging of minute features on the 10μm scale in the narrow gap of device docking, establish a unified high-precision measurement benchmark, and realize a high-precision alignment system and method for visual guidance of end face features of micro-devices based on visual feedback closed-loop control of end face feature points throughout the entire process. Summary of the Invention
[0009] The purpose of this invention is to provide a high-precision alignment system and method for visual guidance of micro-device end-face features. By combining orthogonal vision architecture with interventional prism technology, it solves the technical problems of blind spots in the observation of micro-device end-face features and the difficulty in achieving multi-channel feature matching.
[0010] The present invention is achieved by at least one of the following technical solutions.
[0011] A high-precision alignment system guided by visual features of the end face of a micro-device includes:
[0012] Precision alignment platform module: includes a six-degree-of-freedom precision motion stage and a fixed support stage. The six-degree-of-freedom precision motion stage is used to adsorb and clamp the first micro device on the motion side. The fixed support stage is set on the right side to adsorb and clamp the second micro device on the fixed side. The first micro device and the second micro device are arranged opposite each other, with their end faces to be aligned facing each other, and multiple end face features with consistent spacing are distributed on the end faces. Top vision imaging unit: includes a top industrial camera arranged vertically downwards along the Z-axis and a motorized zoom lens mounted on the industrial camera. The top vision imaging unit features both a low-magnification, large-field-of-view imaging mode and a high-magnification, microscopic imaging mode; Side vision imaging unit: includes a side imaging camera arranged horizontally along the Y-axis. The optical axis of the side imaging camera is orthogonal to the optical axis of the top vision imaging unit, and is used to acquire side projection images of the micro device. Interventional end-face reflection module: includes a right-angle reflecting prism disposed below the alignment gap between the first micro-device and the second micro-device, the right-angle reflecting prism being driven to rise and fall in the vertical direction by a prism lifting mechanism.
[0013] Furthermore, the interventional end-face reflection module has two states: a working position and a retracted position. When in the working position, the right-angle reflecting prism rises between the end faces of the first and second micro-devices, with the right-angle prism facing upward and perpendicular to the optical axis of the top visual imaging unit, to reflect the images of the two opposite end faces into the high-magnification field of view of the top industrial camera. When in the retraction position, the right-angle reflecting prism descends away from the alignment axis, allowing the first and second micro-devices to have the spatial conditions for physical contact or proximity.
[0014] Furthermore, the right-angled edge of the right-angle reflecting prism serves as an absolute spatial common reference in the high-magnification imaging of the top visual imaging unit, used to segment the end face image of the first micro-device and the end face image of the second micro-device, and to establish the mapping relationship between physical spatial coordinates and image pixel coordinates.
[0015] Furthermore, the prism lifting mechanism adopts a linear slide rail mechanism based on stick-slip drive, including a slide rail body, a slider disposed on the slide rail body, a prism mounting base fixedly connected to the slider, and a stick-slip drive unit that is drively connected to the slider. The right-angle reflecting prism is fixedly installed on the prism mounting base. During operation, the stick-slip drive unit drives the slider to move linearly along the slide rail body in the vertical direction, thereby driving the right-angle reflecting prism to switch between the working position and the retraction position, so as to realize the function of the prism entering and exiting the optical path.
[0016] The method for implementing the aforementioned high-precision alignment system guided by visual features of a micro-device end face includes the following steps: Step S1: Control the prism lifting mechanism to drive the prism to the retracted position, activate the side imaging camera to acquire the side projection image, calculate the height difference and pitch angle deviation of the first micro-device relative to the second micro-device; activate the top industrial camera in low magnification mode to acquire the top contour image, calculate the yaw angle deviation; drive the six-degree-of-freedom precision motion stage to perform attitude correction, so that the axes of the first micro-device and the second micro-device are collinear, and complete the coarse adjustment of the spatial attitude of the first micro-device and the second micro-device. Step S2: Control the prism lifting mechanism to drive the right-angle reflecting prism into the working position. The top industrial camera, in conjunction with the electric zoom lens, switches to high magnification mode to acquire alignment images including the end face features of the first micro-device, the end face features of the second micro-device, and the prism edge of the right-angle reflecting prism. Step S3: Identify the prism edge line and establish a symmetrical reference coordinate system by using an image processing algorithm based on Canny edge detection and least squares line fitting. Use a target detection algorithm to obtain the target area on the end face and combine it with a sub-pixel fitting algorithm to calculate the true geometric center of the target. In this way, extract the key feature point coordinates of the end face of the second micro device on the fixed carrier stage and the feature point coordinates of the end face of the first micro device on the six-degree-of-freedom precision motion stage. Step S4: Calculate the roll angle deviation of the feature line connecting the two end faces based on the extracted feature point coordinates; calculate the distance deviation of each feature point pair in the vertical height mapping direction and the position deviation along the horizontal line direction based on the optical mapping principle of the right-angle reflecting prism. Step S5: Generate a pose compensation signal based on the deviation data calculated in step S4, drive the six-degree-of-freedom precision motion stage to eliminate the roll angle deviation, and make the feature points of the first micro-device and the feature points of the second micro-device precisely coincide in the spatial coordinates to achieve fine-tuning alignment of visual closed-loop feedback. Step S6: Control the prism lifting mechanism to reset to the return position, and reactivate the side imaging camera to monitor the physical gap between the two device end faces in real time, driving the first micro device to feed along the alignment axis until physical docking is completed.
[0017] Furthermore, the deviation in step S4 includes: Roll angle deviation The difference in tilt angle between the line connecting the moving side feature points and the line connecting the fixed side feature points relative to the image coordinate system is calculated using the arctangent function. The formula is as follows:
[0018] in, and These are the coordinates of two feature points extracted from the end face of the first micro-device. and These are the coordinates of two feature points extracted from the end face of the second micro-device. Vertical height deviation : Calculate the motion side The horizontal distance from the feature point to the edge line and the fixed side of the first feature point are related. The difference in horizontal distances from each feature point to the edge is expressed by the formula:
[0019] in To set the x-coordinate of the edge of the right-angle reflecting prism in the image coordinate system; Horizontal deviation along the line: Calculate the first deviation on the moving side The ordinate of the first feature point and the fixed side The difference in the ordinates of the feature points.
[0020] Furthermore, in step S6, the gap distance between the end faces of the two devices is calculated in real time using a side imaging camera; when the gap distance is less than a preset safety threshold, the feed speed of the first micro-device is automatically reduced to a micro-stepping mode until the gap distance reaches a preset docking distance or the first micro-device moves to a preset position.
[0021] Furthermore, the gap distance between the end faces of the two devices is calculated as follows: Edge detection and line fitting are used to extract the boundary positions of the two device end faces respectively, and the gap distance between the two device end faces is calculated based on the distance between the two end face boundaries obtained by fitting.
[0022] A computer device according to the present invention includes a memory and a processor, the memory being electrically connected to the processor, the memory storing a computer program, which, when executed by the processor, causes the processor to implement the method described herein.
[0023] The present invention provides a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor implements the method described herein.
[0024] Compared with the prior art, the present invention has the following advantages: 1. Overcoming the blind spot in end-face observation and establishing an absolute measurement benchmark. Addressing the problem in existing technologies where the top camera cannot directly observe the side, i.e., end-face features of a device, this invention uses a right-angle reflecting prism 41 in an intervening end-face reflection module to fold the end-face features into the high-magnification field of view at the top. Simultaneously, by utilizing the straight line formed by the physical edge of the prism in the image as a common measurement benchmark, calibration errors in traditional multi-camera alignment systems are effectively eliminated, ensuring high-precision alignment of the left and right devices in a unified coordinate system.
[0025] 2. Achieving feature-based direct visual closed-loop feedback. Addressing the issue of traditional methods relying on surface marker point calculations and lacking direct end-face feedback, this invention directly extracts the image coordinates of end-face functional features such as microchannel openings and calculates the actual positional deviation between feature points. By driving a six-degree-of-freedom precision motion stage to eliminate this deviation, the system can compensate for the effects of device cutting tolerances or processing errors in real time, ensuring strict matching and precise alignment of multi-channel end-face features, significantly improving assembly accuracy and yield. Attached Figure Description
[0026] Figure 1 This is a schematic diagram of the structure of a high-precision alignment system for visual guidance based on the end-face features of a micro-device, as illustrated in this embodiment. Figure 2 This is a schematic diagram of the optical path structure of the right-angle reflecting prism in an embodiment; Figure 3 A schematic diagram of feature point coordinates for different feature forms during high-magnification imaging of the top visual imaging unit; Figure 4 This is a flowchart illustrating a high-precision alignment method using visual guidance based on the end-face features of a micro-device. Detailed Implementation
[0027] The present invention will now be described in further detail with reference to the accompanying drawings.
[0028] like Figure 1 As shown, the high-precision alignment system for the end face features of a micro-device according to this embodiment includes: The precision alignment platform module includes a six-degree-of-freedom precision motion stage 11 and a fixed support stage 12. The six-degree-of-freedom precision motion stage 11 is used to adsorb and clamp a first micro-device 101 (e.g., the input end of a microfluidic chip) on the moving side. The fixed support stage 12 is provided on the right side to adsorb and clamp a second micro-device 102 (e.g., the output end of a microfluidic chip) on the fixed side. The first micro-device 101 and the second micro-device 102 are arranged facing each other, with their end faces to be aligned facing each other. Both end faces have multiple end face features (e.g., microchannel openings) with consistent spacing, defining the alignment axis of the system. The alignment axis is a reference axis used during the alignment process to characterize the angular relationship between the end face channels of the first micro-device 101 and the second micro-device 102, used to determine the angular deviation between the end face channels and to perform angle correction accordingly.
[0029] The top vision imaging unit, vertically positioned directly above the alignment area, includes a top industrial camera 21 arranged vertically downwards along the Z-axis and an electrically powered zoom lens 22 mounted on the industrial camera 21. This unit has two operating modes: a low-magnification mode for observing the overall macroscopic outline of the top of the device to assist in coarse alignment; and a high-magnification mode for observing micrometer-level end-face features reflected by a prism to perform fine alignment.
[0030] The side vision imaging unit is horizontally positioned to the side of the alignment area and includes a side imaging camera 31 (with a telecentric lens) arranged horizontally along the Y-axis. Its optical axis is perpendicular to the alignment axis (X-axis) of the system and forms an orthogonal viewing angle with the top vision imaging unit. It is used to acquire side projection images of the micro device to provide a side projection view.
[0031] The interventional end-face reflection module includes a right-angle reflecting prism 41, which is installed directly below the alignment gap between the first micro-device 101 and the second micro-device 102. The right-angle prism 41 is arranged with its right-angle edges facing upwards, used to simultaneously reflect the images of the two opposite end faces into the high-magnification field of view of the top visual imaging unit. The right-angle prism 41 serves as an absolute spatial common reference in the high-magnification imaging of the top visual imaging unit, used to segment the end-face images of the first and second micro-devices and establish a mapping relationship between physical spatial coordinates and image pixel coordinates.
[0032] The right-angle reflecting prism 41 is driven by a prism lifting mechanism 42, enabling it to move vertically (Z-axis) and switch between a "working position" and a "returning position". In one embodiment, the prism lifting mechanism 42 employs a linear slide rail mechanism based on stick-slip drive, including a slide rail body, a slider mounted on the slide rail body, a prism mounting base fixedly connected to the slider, and a stick-slip drive unit driven by the slider. The right-angle reflecting prism 41 is fixedly mounted on the prism mounting base. During operation, the stick-slip drive unit drives the slider to move linearly along the slide rail body in the vertical direction, thereby causing the right-angle reflecting prism 41 to move vertically and switch between the working position and the return position, thus realizing the function of the prism entering and exiting the optical path.
[0033] When in the working position, the right-angle reflecting prism 41 rises between the end faces of the first micro-device 101 and the second micro-device 102, with the right-angle prism line facing upward and perpendicular to the optical axis of the top visual imaging unit, so as to reflect the images of the two opposite end faces into the high magnification field of view of the top industrial camera 21 respectively. When in the retraction position, the right-angle reflecting prism 41 descends away from the alignment axis, so that the first micro-device 101 and the second micro-device 102 have the spatial conditions for physical contact or proximity.
[0034] like Figure 4 As shown, this embodiment implements an alignment method for a high-precision alignment system with visual guidance based on the end-face features of the micro-device, comprising the following steps: Step S1: Activate the side vision imaging unit and the top vision imaging unit to correct the posture and complete coarse alignment. System initialization: Control the prism lifting mechanism 42 to drive the prism to the retracted position (lowering state).
[0035] Specifically, firstly, the side imaging camera 31 is activated to acquire side projection images of the first micro-device 101 and the second micro-device 102. The side contours of the devices are extracted using Canny edge detection, and the equations of the side edges of the two devices are obtained by least-squares linear fitting. Based on this, the height difference (Z-axis deviation) and pitch angle deviation of the first micro-device 101 relative to the second micro-device 102 are calculated. The six-degree-of-freedom precision motion stage 11 is then driven to eliminate the height difference and level the pitch angle. Subsequently, the top industrial camera 21 is switched to low-magnification mode. Canny edge detection and least-squares linear fitting are also used to extract the top outer contours of the two devices, and the yaw angle deviation is calculated. The six-degree-of-freedom precision motion stage 11 is then driven to rotate, making the axes of the first micro-device 101 and the second micro-device 102 collinear, thus completing the initial alignment of their spatial attitude.
[0036] Step S2: The prism is inserted into the optical path.
[0037] After completing the previous stage, the prism lifting mechanism 42 drives the right-angle reflecting prism 41 to rise to the working position, allowing the right-angle reflecting prism 41 to enter the alignment gap. The "alignment gap" refers to the small distance reserved between the end faces of the first micro-device 101 and the second micro-device 102 for subsequent docking. This distance allows the right-angle reflecting prism to be inserted and complete the optical path refracting, specifically as follows... Figure 2 As shown. The right-angle reflecting prism 41 is a roof-structure reflector, comprising two reflecting surfaces at 90° to each other. Its function is equivalent to two right-angled plane mirrors, reflecting the images of the end faces of the devices on both sides into the upper field of view. The top industrial camera 21, in conjunction with the motorized zoom lens 22, switches to high-magnification mode. At this time, the right-angle reflecting prism 41 simultaneously reflects the end face images of the first micro-device 101 and the second micro-device 102 into the camera's field of view. Figure 3 As shown, a clear black straight line appears in the center of the camera's field of view, which is the physical edge of the right-angle reflecting prism 41; the end face features of the first micro-device 101 are shown on the left side of the field of view, and the end face features of the second micro-device 102 are shown on the right side of the field of view.
[0038] Step S3: Using the prism edge as a reference, extract the image coordinates of feature points on both sides. The image processing unit is an existing industrial vision processing module, which can be implemented by an industrial control computer in conjunction with the OpenCV machine vision library. The image processing unit establishes a symmetrical reference coordinate system using the prism edge as the absolute spatial common reference. The horizontal coordinate of the edge in the image coordinate system is set as... The system employs an image processing algorithm based on Canny edge detection and least squares line fitting to identify prism edges and establish a symmetrical reference coordinate system. It uses a target detection algorithm to obtain the target area on the end face and combines it with a sub-pixel fitting algorithm to calculate the true geometric center of the target. This allows the extraction of the key feature point coordinates on the end face of the second micro-device on the fixed support stage 12 and the feature point coordinates on the end face of the first micro-device on the six-degree-of-freedom precision motion stage 11.
[0039] In one embodiment, taking an end face containing two microchannel features as an example: (e.g.) Figure 3 As shown, on the fixed side (right side): the coordinates of two feature points on the end face of the second micro-device 102 are extracted. and ;Motion side (left side): Extract the coordinates of two corresponding feature points on the end face of the first micro-device 101. and , i and j They are 1 and 2 respectively. Figure 3 The imaging and feature point coordinate extraction results under two feature forms are illustrated. It should be noted that the feature forms are not limited to the two shown in the figure. Any identifiable geometric feature or grayscale feature that can be used for end face alignment can be applied to this step.
[0040] Step S4: Calculate the roll angle deviation and the coordinate deviation of each feature point pair. Based on the extracted coordinate data, the system performs the following deviation calculation: (1) Roll angle deviation: The tilt angle of the line connecting the feature points on both sides relative to the image coordinate system is calculated using the arctangent function. Fixed tilt angle. The formula is calculated based on the feature point coordinates of the second micro-device 102:
[0041] Roll angle The formula is calculated based on the coordinates of the original feature points of the first micro-device 101:
[0042] Deviation calculation: Roll angle deviation
[0043] The calculation is as follows:
[0044] The system will drive the six-DOF precision motion stage 11 to rotate around the alignment axis according to this deviation, until... Approaching zero.
[0045] (2) Feature point position deviation: For the two pairs of homologous feature points extracted in step S3 (i.e. and The deviations are calculated using the optical mapping principle of prism reflection: Vertical height deviation (Z-axis mapping): Calculates the distance from the feature point on the moving side to the edge. The horizontal distance and the x-coordinate of the corresponding feature point on the fixed side to the edge. The difference in horizontal distance. For the first pair of feature points:
[0046] For the second pair of feature points:
[0047] Horizontal deviation along the line (Y-axis): Calculate the difference between the ordinate of the feature point on the moving side and the ordinate of the corresponding feature point on the fixed side. For the first pair of feature points:
[0048] For the second pair of feature points:
[0049] Step S5: Drive the motion stage to compensate for deviations and achieve feature point matching and precise overlap. The system generates control commands based on the deviations calculated in step S4, driving the six-degree-of-freedom precision motion stage 11 to perform pose compensation: First, correct the angle: drive the six-degree-of-freedom precision motion stage 11 to rotate around the alignment axis (X-axis), making the feature connection line on the moving side symmetrical with the feature connection line on the fixed side about the edge line, eliminating the roll angle deviation. At this time, the arrangement directions of the features on both end faces are parallel to each other. Then, correct the position: drive the six-degree-of-freedom precision motion stage 11 to make small movements along the vertical direction (Z-axis) and the horizontal direction (Y-axis), so that for each pair of feature points, the aforementioned height deviation and horizontal deviation approach zero. Alignment result: After the deviation is eliminated, the feature points of the first micro-device 101... Feature points of the second micro-device 102 Precise spatial alignment; due to pre-corrected angles and consistent feature spacing, the second pair of feature points... and They will inevitably overlap, thus achieving strict matching and precise alignment of multi-channel end face features.
[0050] Step S6: The prism is removed from the optical path. After visual alignment is completed and the gap is monitored from the side, the prism lifting mechanism 42 is controlled to drive the prism down to the return position. The side imaging camera 31 is activated again to monitor the physical gap between the two device end faces in real time, guiding the first micro-device 101 to feed along the alignment axis until the docking is completed. The six-degree-of-freedom precision motion stage 11 is driven to feed to the right along the alignment axis (X-axis); when the physical gap is less than the preset safety threshold (e.g., 10 micrometers), the control system automatically switches to micro-stepping mode until the preset position is reached, completing the final physical docking.
[0051] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, enabling those skilled in the art to better understand and utilize the invention.
Claims
1. A high-precision alignment system guided by visual characteristics of the end face features of a micro-device, characterized in that, include: Precision alignment platform module: includes a six-degree-of-freedom precision motion stage and a fixed support stage, wherein the six-degree-of-freedom precision motion stage is used to adsorb and clamp the first micro device on the motion side; A fixed support platform is set on the right side for adsorbing and clamping the second micro device on the fixed side; the first micro device and the second micro device are arranged opposite each other, with their end faces to be aligned facing each other, and multiple end face features with consistent spacing are distributed on their end faces. Top vision imaging unit: includes a top industrial camera arranged vertically downward along the Z-axis and an electric zoom lens mounted on the industrial camera; the top vision imaging unit has a low magnification large field of view imaging mode and a high magnification microscopic imaging mode. Side vision imaging unit: includes a side imaging camera arranged horizontally along the Y-axis. The optical axis of the side imaging camera is orthogonal to the optical axis of the top vision imaging unit, and is used to acquire side projection images of the micro device. Interventional end-face reflection module: includes a right-angle reflecting prism disposed below the alignment gap between the first micro-device and the second micro-device, the right-angle reflecting prism being driven to rise and fall in the vertical direction by a prism lifting mechanism.
2. The high-precision alignment system for micro-device end-face features as described in claim 1, characterized in that: The interventional end-face reflection module has two states: a working position and a retracted position. When in the working position, the right-angle reflecting prism rises between the end faces of the first and second micro-devices, with the right-angle prism facing upward and perpendicular to the optical axis of the top visual imaging unit, to reflect the images of the two opposite end faces into the high-magnification field of view of the top industrial camera. When in the retraction position, the right-angle reflecting prism descends away from the alignment axis, allowing the first and second micro-devices to have the spatial conditions for physical contact or proximity.
3. The high-precision alignment system for micro-device end-face features visually guided according to claim 2, characterized in that: The right-angled edge of the right-angle reflecting prism serves as an absolute spatial common reference in the high-magnification imaging of the top visual imaging unit. It is used to segment the end face image of the first micro-device and the end face image of the second micro-device, and to establish the mapping relationship between physical space coordinates and image pixel coordinates.
4. The high-precision alignment system for micro-device end-face features visually guided according to claim 1, characterized in that: The prism lifting mechanism adopts a linear slide rail mechanism based on stick-slip drive, including a slide rail body, a slider set on the slide rail body, a prism mounting base fixedly connected to the slider, and a stick-slip drive unit connected to the slider for transmission. The right-angle reflecting prism is fixedly installed on the prism mounting base. During operation, the stick-slip drive unit drives the slider to move linearly along the slide rail body in the vertical direction, thereby driving the right-angle reflecting prism to switch between the working position and the retraction position, so as to realize the function of the prism entering and exiting the optical path.
5. A method for implementing a high-precision alignment system for the end-face features of a micro-device as described in claim 1, characterized in that, Includes the following steps: Step S1: Control the prism lifting mechanism to drive the prism to the retracted position, activate the side imaging camera to acquire the side projection image, and calculate the height difference and pitch angle deviation of the first micro-device relative to the second micro-device. Use the top industrial camera in low-magnification mode to acquire a top contour image and calculate the sway angle deviation; The six-degree-of-freedom precision motion stage is driven to perform attitude correction, so that the axes of the first and second micro-devices are collinear, thus completing the coarse spatial attitude adjustment of the first and second micro-devices. Step S2: Control the prism lifting mechanism to drive the right-angle reflecting prism into the working position. The top industrial camera, in conjunction with the electric zoom lens, switches to high magnification mode to acquire alignment images including the end face features of the first micro-device, the end face features of the second micro-device, and the prism edge of the right-angle reflecting prism. Step S3: Identify the prism edge line and establish a symmetrical reference coordinate system by using an image processing algorithm based on Canny edge detection and least squares line fitting. Use a target detection algorithm to obtain the target area on the end face and combine it with a sub-pixel fitting algorithm to calculate the true geometric center of the target. In this way, extract the key feature point coordinates of the end face of the second micro device on the fixed carrier stage and the feature point coordinates of the end face of the first micro device on the six-degree-of-freedom precision motion stage. Step S4: Calculate the roll angle deviation of the feature line connecting the two end faces based on the extracted feature point coordinates; calculate the distance deviation of each feature point pair in the vertical height mapping direction and the position deviation along the horizontal line direction based on the optical mapping principle of the right-angle reflecting prism. Step S5: Generate a pose compensation signal based on the deviation data calculated in step S4, drive the six-degree-of-freedom precision motion stage to eliminate the roll angle deviation, and make the feature points of the first micro-device and the feature points of the second micro-device precisely coincide in the spatial coordinates to achieve fine-tuning alignment of visual closed-loop feedback. Step S6: Control the prism lifting mechanism to reset to the return position, and reactivate the side imaging camera to monitor the physical gap between the two device end faces in real time, driving the first micro device to feed along the alignment axis until physical docking is completed.
6. The method for high-precision alignment of micro-device end-face features guided by vision according to claim 5, characterized in that, The deviation in step S4 includes: Roll angle deviation The difference in tilt angle between the line connecting the moving side feature points and the line connecting the fixed side feature points relative to the image coordinate system is calculated using the arctangent function. The formula is as follows: in, and These are the coordinates of two feature points extracted from the end face of the first micro-device. and These are the coordinates of two feature points extracted from the end face of the second micro-device. Vertical height deviation : Calculate the motion side The horizontal distance from the feature point to the edge line and the fixed side of the first feature point are related. The difference in horizontal distances from each feature point to the edge is expressed by the formula: in To set the x-coordinate of the edge of the right-angle reflecting prism in the image coordinate system; Horizontal deviation along the line: Calculate the first deviation on the moving side The ordinate of the first feature point and the fixed side The difference in the ordinates of the feature points.
7. The method for high-precision alignment of micro-device end-face features guided by vision according to claim 5, characterized in that, In step S6, the gap distance between the end faces of the two devices is calculated in real time using a side imaging camera; when the gap distance is less than a preset safety threshold, the feed speed of the first micro-device is automatically reduced to micro-stepping mode until the gap distance reaches the preset docking distance or the first micro-device moves to a preset position.
8. The method for high-precision alignment of micro-device end-face features guided by vision according to claim 7, characterized in that, The gap distance between the end faces of the two devices is calculated as follows: Edge detection and line fitting are used to extract the boundary positions of the two device end faces respectively, and the gap distance between the two device end faces is calculated based on the distance between the two end face boundaries obtained by fitting.
9. A computer device comprising a memory and a processor, the memory being electrically connected to the processor, the memory storing a computer program, characterized in that: When the computer program is executed by the processor, it causes the processor to implement the method as described in any one of claims 5 to 8.
10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the processor implements the method as described in any one of claims 5 to 8.