An optical element automatic assembling and adjusting method based on force-visual hybrid control

Abstract

The application discloses an optical element automatic assembling and adjusting method based on force-visual hybrid control, and belongs to the technical fields of precise optical assembly, robot precise operation, force control assembly and interference visual detection. The steps are as follows: first, assembling force signal acquisition and zero adjustment, and smooth assembling force is obtained; second, assembling force closed loop contact control, and controlled contact state is obtained; third, double angle shaft contact rough adjustment and leveling based on assembling force gradient and variable step length critical search, and the posture of the optical element to be assembled after contact rough adjustment and leveling is obtained; fourth, interference background reference extraction; fifth, interference fringe appearance judgment and pose solution; and sixth, robot angle precise compensation is executed according to the solution value. The application can complete target fringe analysis area selection and robot angle precise compensation by using frequency domain fringe analysis after effective interference fringes appear, forms optical element automatic assembling and adjusting connected by force feedback and visual feedback, and realizes automatic leveling of the optical element to be assembled relative to the reference optical surface.

Description

Technical Field

[0001] This invention belongs to the fields of precision optical assembly, robot precision operation, force-controlled assembly and interferometric vision inspection technology, and specifically relates to an automatic assembly and adjustment method for optical components based on force-vision hybrid control. Background Technology

[0002] In precision optical assembly, the relative attitude between the optical component to be assembled and the reference optical surface affects the optical path quality, assembly consistency, and system stability. Current optical component leveling processes still largely rely on manual observation of contact states, light spots, or interference fringes, with attitude correction achieved through repeated adjustments. When the initial tilt between the optical component to be assembled and the reference optical surface is significant, effective interference fringes for pose calculation may not yet have formed in the field of view, making it difficult to directly use interferometric visual information for precise leveling. Furthermore, a lack of stable control over the assembly force during the contact phase can cause fluctuations in the contact state, increasing the risk of uneven stress on the optical surface. Therefore, establishing controlled contact and completing initial leveling before the appearance of effective interference fringes, and further performing precise attitude compensation after the appearance of effective interference fringes, are problems that need to be solved in the automated assembly and leveling of optical components.

[0003] To address the relationship between assembly force and pose error during optical lens assembly, Chinese invention patent CN120429906A discloses a neural network-based force-position coordinated assembly and adjustment method for optical lenses. This method conducts cyclic simulations based on an optical system model and the assembly process to establish a dataset relating assembly force and pose error, and then uses this dataset to train an assembly error prediction model. During the assembly and adjustment process, the method predicts the pose error based on the current assembly force and determines the actual adjustment amount by combining this with the target misalignment. This method primarily solves the problem of predicting and compensating for optical lens pose errors caused by assembly force, providing a data-driven solution for modeling the correlation between mechanical state and pose error.

[0004] To address the issues of load-bearing and alignment during optical lens installation, Chinese invention patent CN121165278A discloses an optical lens installation device and method. The method involves using a frame support adjustment assembly and a lens support adjustment assembly to support the frame and optical lens respectively. Adjusting the height and relative position of the frame and lens moves the optical lens into its mounting slot on the frame. A laser reference can be used to assist in determining the installation position. This method primarily solves the problems of support, handling, and mechanical alignment during the process of inserting the optical lens into the frame mounting slot, helping to reduce the risks of collision, scratches, and contamination caused by manually holding the optical lens.

[0005] The aforementioned existing technologies have proposed solutions for predicting pose errors caused by assembly forces and for the mechanical support and installation of optical lenses, respectively. However, in the automatic leveling process of the optical element to be installed relative to the reference optical surface, it is still necessary to solve problems such as the lack of effective interference fringes in the initial stage, the difficulty in stabilizing the assembly force in the contact state, the difficulty in continuously connecting contact coarse leveling and non-contact interference precision measurement, and the need to reliably determine the effectiveness of interference fringes and the target analysis area.

[0006] Therefore, it is necessary to provide an automatic assembly and adjustment method for optical components based on force-vision hybrid control, which utilizes assembly force feedback to complete controlled contact and coarse angle leveling, and utilizes interference vision information to complete pose calculation and precise angle compensation after the appearance of effective interference fringes. Summary of the Invention

[0007] To address the problems encountered in precision optical assembly, such as reliance on manual experience for leveling optical components to be assembled relative to a reference optical surface, potential lack of effective interference fringes in the initial stage, difficulty in stabilizing contact force control, and susceptibility to misjudgment of noise fringes, this invention provides an automated optical component assembly and adjustment method based on force-vision hybrid control. The optical component to be assembled refers to the optical component that needs to be moved, its attitude adjusted, and ultimately establish a predetermined relative attitude relationship with the target position during the optical assembly process. The reference optical surface refers to the optical surface that is set relative to the optical component to be assembled and serves as the leveling reference. This invention achieves automated leveling of the optical component to be assembled relative to the reference optical surface through closed-loop contact control of assembly force, coarse leveling based on dual-angle axial contact according to assembly force changes, interference background reference extraction, interference fringe appearance determination and pose calculation, and precise robot angle compensation.

[0008] To achieve the above objectives, the technical solution adopted by the present invention is as follows: An automated assembly and adjustment method for optical components based on force-vision hybrid control includes the following steps: The first step is to acquire and zero the assembly force signal to obtain a smooth assembly force.

[0009] Assembly force refers to the force generated along the normal direction and measured by a force sensor during the contact or near-contact process between the optical element to be assembled and the reference optical surface. This step is used to obtain the assembly force signal that can be used for subsequent closed-loop contact control of assembly force and coarse leveling of dual-angle axis contact. Specifically, it includes acquiring the voltage signal of the force sensor, establishing a zero-point reference, converting the voltage signal into an assembly force signal, and filtering the assembly force signal to obtain a smooth assembly force.

[0010] Step 1.1: Acquire the voltage signal output by the force sensor, and record it as the force sensor output voltage. ,in Indicates time The force sensor output voltage, Indicates time. When the optical element to be installed has not yet made effective contact with the reference optical surface, the time is... Zero-point sampling is performed to obtain the zero-point voltage. ,in This indicates the voltage reference of the force sensor under zero assembly force reference conditions.

[0011] Step 1.2, based on the results obtained in Step 1.1 and Convert the force sensor voltage signal into an assembly force signal. : (1) in, Indicates time Assembly force signal, This indicates the calibration coefficient for the force sensor voltage to the assembly force.

[0012] Step 1.3, based on the results obtained in Step 1.2 The assembly force was smoothed by using a filtering method. : (2) in, Indicates time Smooth assembly force, This represents the filtering operator used for assembly force signals. The smoothing assembly force... It serves as the input for subsequent assembly force closed-loop contact control and dual-angle shaft contact coarse leveling.

[0013] The second step is to perform closed-loop contact control of the assembly force to obtain a controlled contact state.

[0014] This step is based on the smooth assembly force obtained in the first step. The robot is controlled to move along the normal direction between the optical element to be installed and the reference optical surface. Specifically, this includes setting the target assembly force, judging the contact state close to the target, calculating the assembly force error, calculating the normal displacement compensation amount, and executing the normal compensation movement to obtain the controlled contact state for subsequent coarse leveling of dual-angle axis contact.

[0015] Step 2.1: Set the target assembly force to be maintained during the assembly process. The robot is controlled to approach the reference optical surface along the normal direction, and the smooth assembly force obtained in the first step is read in real time. .

[0016] Step 2.2, based on the results obtained in Step 2.1 and Determine if the optical component to be installed is close to the target contact state. Set the closed-loop cutting force threshold. ,when When the closed-loop entry condition is not yet met, the robot approaches the reference optical surface with gradually decreasing normal displacement increments; when When the closed-loop entry condition is met, proceed to step 2.3 to start assembly force closed-loop control.

[0017] Step 2.3: Based on the position of the optical element to be installed in the near-target contact state obtained in Step 2.2, initiate closed-loop control of the assembly force. Define the time. Assembly force error : (3) Step 2.4, based on the results obtained in Step 2.3 Calculate the robot's normal displacement compensation amount : (4) in, Indicates time The normal displacement compensation amount, This represents the proportionality coefficient. Represents the integral coefficient. This represents the differential coefficient.

[0018] Step 2.5, based on the results obtained in Step 2.4 The robot is controlled to perform compensating motion along the normal direction to obtain the position of the optical element to be installed in the closed-loop contact state. : (5) in, Indicates the normal position at the next control moment. This indicates the normal position at the current control moment. This yields the controlled contact state used for subsequent coarse leveling of the dual-angle shaft contact.

[0019] The third step involves coarse leveling of the optical element to be assembled using dual-angle axis contact based on the assembly force gradient and variable step size critical search, to obtain the orientation of the optical element after coarse leveling.

[0020] Based on the controlled contact state obtained in the second step, the first angular axis Second angle axis The same assembly force gradient direction determination and variable step size critical search method are used for contact coarse leveling. Specifically, this includes establishing a set of angle axes to be adjusted and an angle step size sequence, collecting the average smooth assembly force before and after the positive trial, determining the current angle axis coarse leveling direction, performing critical search according to the angle step size sequence from large to small, and completing the coarse leveling of the two angle axes.

[0021] Step 3.1, establish the set of angle axes to be adjusted First angular axis Second angle axis These represent the small rotational degrees of freedom of the robot's end effector about two mutually orthogonal attitude adjustment axes. An angle step sequence is established. , Indicates the first Angle step size, Indicates the angle step number. Indicates the number of angle step increments, and Follow The values ​​increase sequentially and then decrease. Establish a sequence of assembly force increment thresholds. ,in Indicates the first Angle step The corresponding assembly force increment threshold. Assembly force increment threshold sequence. The assembly force increment threshold can be determined based on the target assembly force, assembly force noise level, allowable contact force variation range of the optical component to be assembled, and expected contact state changes corresponding to each angle step. Based on the target assembly force The amplitude of the assembly force signal noise and the allowable range of contact force variation are determined. In one optional method, Target assembly force can be taken 20% to 50%.

[0022] Step 3.2, for Any angle axis currently undergoing coarse leveling. Its average smooth assembly force before positive testing is collected. : (6) in, The average smooth assembly force before positive testing. For mean calculation, This is the sampling time window for assembly force.

[0023] Step 3.3, based on the results obtained in Step 3.2 Control the robot along the current angle axis Execute the first angle step in the positive direction. The angular position after the positive trial is obtained, and the average smooth assembly force after the positive trial is collected. : (7) Step 3.4, based on the results obtained in step 3.2 And obtained in step 3.3 Calculate the increment of positive trial assembly force : (8) Step 3.5, based on the results obtained in step 3.4 Compared with the first-level assembly force increment threshold Determine the current angle axis coarse adjustment direction .like Greater than Then determine the current angle axis. The positive direction causes the contact state to tend to increase force, controlling the robot to retreat from the positive probing angle, and... Set to the negative direction; if Not greater than Then determine the current angle axis. The positive direction did not cause a significant increase in force, and Set as the positive direction. Among them, Indicates the current angle axis The coarse adjustment direction, Indicates the first-level assembly force increment threshold. Step 3.6, based on the results obtained in step 3.5 For the current angle axis Angle step sequence Perform critical searches sequentially. In the... With each angle step, control the robot to move along... The direction is adjusted by the angle, and the average smooth assembly force before and after the angle adjustment is collected respectively. The first step is to calculate the... Adjusting the gear angle corresponds to the increase in assembly force. : (9) in, For the first Average smooth assembly force after gear angle adjustment For the first Average smooth assembly force before gear angle adjustment.

[0024] Step 3.7, based on the results obtained in Step 3.6 Determine the current angle axis Has the first level been reached? The critical position under the angle step size. If Greater than Then control the robot along Reverse the current angle step by one increment in the opposite direction and switch to the next smaller angle step; if Not greater than Then continue along The direction is used to perform the current angle step search.

[0025] Step 3.8: Following steps 3.2 to 3.7, assemble the set of angle axes to be adjusted. In and Coarse leveling was performed to obtain the orientation of the optical element to be installed after coarse leveling. This orientation serves as the initial orientation for subsequent interference background extraction.

[0026] The fourth step is to extract the interference background reference.

[0027] The interference background reference refers to the frequency domain reference information acquired and processed when the optical element to be installed and the reference optical surface are in a non-contact background state. This step, based on the orientation of the optical element to be installed after coarse leveling in step three, closes the assembly force closed-loop contact control and controls the robot to retreat along the normal direction. Specifically, this includes stopping the normal closed-loop compensation, executing normal retreat, acquiring the background image, determining the fringe analysis region set based on the visible interference fringe distribution, and calculating the background spectral amplitude of each fringe analysis region to obtain the interference background reference used for subsequent interference fringe appearance determination and pose calculation. The visible interference fringes refer to alternating bright and dark fringes or local fringe fragments that can be acquired by the camera in the imaging field of view, generated by the gap variation between the optical element to be installed and the reference optical surface. Due to the influence of the shape of the optical element to be installed, the clamping structure, or field-of-view occlusion, visible interference fringes may only be distributed in a local area of ​​the image, and are not required to appear continuously and completely in the entire image.

[0028] Step 4.1: Based on the posture of the optical component to be assembled after contact coarse leveling obtained in step 3, stop the assembly force closed-loop contact control in step 2, so that the robot no longer performs normal closed-loop compensation based on the assembly force error, that is, obtain the stopped normal closed-loop compensation state.

[0029] Step 4.2: Based on the stopping normal closed-loop compensation state obtained in Step 4.1, control the robot to move a distance away along the normal direction away from the reference optical surface. ,in The normal retraction distance used to form a non-contact background state. A non-contact background state is obtained when the controlled contact between the optical element to be mounted and the reference optical surface is released and the element is in a relative position where a background image can be acquired.

[0030] Step 4.3: Based on the non-contact background state obtained in Step 4.2, acquire image coordinates. Background image at the location ,in, Represents the horizontal coordinate of the image. Represents the vertical coordinate of the image.

[0031] Step 4.4, based on the background image obtained in step 4.3 Based on the distribution, continuity, and effective imaging range of visible interference fringes in the image, the background image is divided into... Each stripe analysis area ,in This indicates the total number of stripe analysis regions. Indicates the index of the stripe analysis region. , Indicates the first Each stripe analysis region. The stripe analysis region This refers to the left, right, upper, lower, or ring-shaped local areas of the image, or other local image regions defined by the distribution of visible stripes.

[0032] Step 4.5, analyze each stripe region separately. The background image in the image is subjected to frequency domain transformation to obtain the first... Each fringe analysis region in the frequency domain coordinates Background spectral amplitude at the location : (10) in, Represents the frequency domain transform operator. Represents the horizontal coordinate in the frequency domain. Represents the vertical coordinate in the frequency domain. Indicates the first Background area image.

[0033] Background spectral amplitude of each stripe analysis region Together, they constitute the interference background reference for subsequent determination of interference fringe appearance and pose calculation.

[0034] The fifth step is to determine the appearance of interference fringes and calculate their pose.

[0035] The determination of interference fringe appearance refers to judging whether there are effective wedge interference fringes in the current image that can be used for pose calculation based on the main peak of the spectrum, the spectral background, and the spatial frequency range. This step is based on the interference background reference obtained in step four. The robot is controlled to gradually approach the reference optical surface along the normal. First, the current interference image is acquired. The fringe analysis region in the current interference image is divided according to the distribution of visible interference fringes. Then, frequency domain transformation and background subtraction are performed on each fringe analysis region in sequence. Candidate main peaks of the spectrum of each fringe analysis region are extracted. The fringe quality evaluation results are calculated. The effective fringe analysis regions are selected. Finally, the fringe spacing, total tilt angle, first angular axis compensation amount and second angular axis compensation amount are calculated based on the selected effective analysis results.

[0036] Step 5.1: Based on the non-contact background state obtained in step 4, control the robot to perform a search displacement along the normal direction close to the reference optical surface. Move gradually, among which This represents the normal search displacement used to search for interference fringes.

[0037] Step 5.2: Based on each normal search position obtained in Step 5.1, acquire image coordinates. Current interferometric image at the location Following the region division rules in step 4.4, or updating the region boundaries accordingly based on the current visible interference fringe distribution, the current interference image is divided into... Current stripe analysis area ,in Indicates the first [number] in the current interferometric image Each stripe analysis area.

[0038] Step 5.3, for each current fringe analysis region obtained in step 5.2 Perform a frequency domain transformation to obtain the current spectral amplitude. And compared with the background spectrum amplitude obtained in step 4.5 Background subtraction is performed to obtain the effective spectral amplitude. : (11) in, Indicates the first Each fringe analysis region in the frequency domain coordinates The effective spectral amplitude at that location, Indicates the first Each fringe analysis region in the frequency domain coordinates The current spectral amplitude at that location, This indicates that the background weight has been removed. Indicates the first Each fringe analysis region in the frequency domain coordinates The background spectral amplitude at that location.

[0039] Step 5.4, based on the effective spectral amplitude obtained in step 5.3 Search within the preset effective spatial frequency range for the first The candidate spectral main peak of the first stripe analysis region is obtained. Candidate spectral peak amplitude of each stripe analysis region , No. Spectral noise floor of each stripe analysis region , No. The transverse frequency components of the candidate main peak in each stripe analysis region , No. The longitudinal frequency components of the main peak in the stripe analysis region are candidate. and the Significance of the main peak in each stripe analysis region candidate .

[0040] Step 5.5, based on the amplitude of the main peak of the candidate spectrum obtained in step 5.4 and spectral noise floor Calculate the first The signal-to-noise ratio of the main peak in each stripe analysis region : (12) in, Indicates the first The signal-to-noise ratio of the main peak in each stripe analysis region. This indicates a positive number used to avoid a denominator of zero.

[0041] Step 5.6: Based on the candidate spectral main peaks and fringe quality evaluation results of each fringe analysis region obtained in Steps 5.4 and 5.5, the first... The presence of interference fringes in each fringe analysis region is determined based on the transverse frequency component of the main peak in the candidate spectrum. and longitudinal frequency components of the main peak of the candidate spectrum Calculate the first Candidate spectral main peak spatial frequency of each stripe analysis region : (13) Pre-set the main peak signal-to-noise ratio threshold Threshold for determining the significance of the main peak and effective spatial frequency range The thresholds for determining the signal-to-noise ratio of the main peak, the threshold for determining the salience of the main peak, and the effective spatial frequency range are determined based on the interferometric background reference, imaging system parameters, historical calibration data, or preset judgment rules. This indicates the lower limit of the signal-to-noise ratio of the main peak used to distinguish between the main peak of a valid candidate spectrum and spectral noise. This represents the lower limit of significance of the main peak used to distinguish the candidate spectral main peak from its neighboring spectral components. and These represent the lower and upper limits of the effective spatial frequency range, respectively.

[0042] Then the first The criteria for determining effective interference fringes in each fringe analysis region are: (14) in, Indicates the first The signal-to-noise ratio of the main peak in each stripe analysis region. Indicates the first The significance of the main peak in each stripe analysis region.

[0043] when , and When all three conditions shown in formula (14) are met simultaneously, the first condition is determined. If a valid interference fringe exists in a fringe analysis region, the region is deemed to lack a valid interference fringe if any of the criteria is not met.

[0044] Step 5.7: Based on the interference fringe occurrence status of each fringe analysis region obtained in Step 5.6, a set of effective fringe analysis regions is formed. ,in This represents the set of fringe analysis region indices indicating the presence of valid interference fringes. If... If it is an empty set, return to step 5.1 to continue the normal search; if If it is not an empty set, proceed to step 5.8.

[0045] Step 5.8: Based on the candidate spectral main peaks and fringe quality evaluation results of each fringe analysis region obtained in steps 5.4 to 5.6, calculate the... Quality evaluation value of each stripe analysis area and from the effective stripe analysis region set Select the target stripe analysis area : (15) (16) in, This indicates the index of the target fringe analysis region selected for pose calculation. From the main peak signal-to-noise ratio Significance of the main peak Main peak amplitude It is determined jointly by one or more of the frequency range constraints.

[0046] Step 5.9, based on the target stripe analysis region obtained in Step 5.8 Determine the transverse frequency components of the main peak of the target spectrum. and the longitudinal frequency component of the main peak of the target spectrum ,in Specifically, it refers to the spatial frequency component corresponding to the main peak of the target spectrum in the horizontal direction of the image. Specifically, it refers to the spatial frequency component corresponding to the main peak of the target spectrum in the vertical direction of the image. Calculate the spatial frequency of the main peak of the target spectrum. : (17) Step 5.10, based on the target spectrum main peak spatial frequency obtained in step 5.9 Calculate the stripe pixel spacing And according to pixel size Calculate the physical spacing of the stripes : (18) Step 5.11, based on the stripe physical spacing obtained in step 5.10 The total tilt angle between the optical element to be installed and the reference optical surface is calculated using wedge interference relations. : (19) in, This indicates the wavelength of the light source used to form the wedge interference.

[0047] Step 5.12, based on the transverse frequency components of the main peak of the target spectrum obtained in step 5.9. and the longitudinal frequency component of the main peak of the target spectrum Calculate the direction angle of the main peak of the target spectrum. : (20) in, This represents the arctangent function in the four quadrants.

[0048] Step 5.13, based on the total tilt angle between the optical element to be installed and the reference optical surface obtained in step 5.11. and the target spectrum main peak direction angle obtained in step 5.12 Based on the projection relationship of the main peak direction of the target spectrum onto two mutually orthogonal angular axes, the total tilt angle is... Decomposed into the absolute value of the first angular axis compensation amount |C RX |The absolute value of the second angle axis compensation|C RY |: (twenty one) Among them, |C RX | represents the absolute value of the compensation amount for the first angular axis, |C RY | indicates the absolute value of the compensation amount for the second angular axis; This represents the total tilt angle between the optical element to be installed and the reference optical surface obtained in step 5.11; This represents the orientation angle of the main peak of the target spectrum obtained in step 5.12; and These represent the projection coefficients of the total tilt angle in the first and second angular axis directions, respectively. Step 5.13 only determines the magnitude of the compensation amounts for the two angular axes, without specifying their positive or negative directions.

[0049] Step 5.14, based on the target stripe analysis region obtained in step 5.8 The location attributes and the transverse frequency components of the main peak of the target spectrum obtained in step 5.9. and the longitudinal frequency component of the main peak of the target spectrum The sign relationship, and the pre-defined mapping relationship between the image tilt direction and the robot's angular axis movement direction, are used to determine the first angular axis compensation amount. Second angle axis compensation amount Determine the sign; assign the determined sign to the value obtained in step 5.13. and The first angular axis compensation amount with the sign is obtained. Second angle axis compensation amount .

[0050] The target stripe analysis area The positional attribute represents its relative position in the image coordinate system; the sign relationship of the main peak frequency components of the target spectrum represents... and The positive and negative combinations within the specified spectral half-plane; the mapping relationship between the image tilt direction and the robot's angular axis movement direction, representing the positional attributes of the target fringe analysis region and the sign combination of the target spectral main peak frequency components, and the first angular axis. Second angle axis The correspondence between positive and negative motion directions. This mapping relationship is established before automatic assembly by controlling the robot to move along the first angular axis. Second angle axis The process involves performing a preset small positive angle motion and recording the changes in the fringe analysis region and the frequency components of the target spectral main peak in the interferometric image. For paired conjugate main peaks in the real-valued image spectrum, a predefined spectral half-plane selection rule is used to determine the target spectral main peak for sign determination, thus maintaining consistency in the sign determination results.

[0051] Step 6, according to , The calculated values ​​are used to perform precise angle compensation for the robot.

[0052] This step is based on the signed first angle axis compensation amount obtained in step five. Second angle axis Compensation amount To control the robot to perform precise angle compensation, specifically including determining the compensation amount. Does the safety constraint meet the requirements? Is the first angular axis being executed? Compensation, execution of the second angle axis Compensation and confirmation of the leveling results based on the compensated interferometric image.

[0053] Step 6.1, based on the results obtained in step 5 and ,judge and Whether the angle compensation safety constraints are met. The angle compensation safety constraints include single-time angle compensation amplitude constraints and robot angle axis motion range constraints; wherein, the single-time angle compensation amplitude constraint is used to limit the maximum compensation amplitude of the first and second angle axes in a single precision compensation, and the robot angle axis motion range constraint is used to limit the compensated angle position from exceeding the robot's preset motion range. and If the angle compensation safety constraint is met, proceed to step 6.2; if any compensation amount does not meet the angle compensation safety constraint, terminate the angle precision compensation and output an abnormal status.

[0054] Step 6.2, based on the determination in step 6.1 that the safety constraints are met. Control the robot to rotate around the first angular axis Execution angle compensation: (twenty two) in, Indicates the first angular axis The compensated angle position Indicates the first angular axis The angle position before compensation, Indicates the first angular axis Compensation amount.

[0055] Step 6.3, based on the determination in step 6.1 that the safety constraints are met. Control the robot to rotate around the second angular axis Execution angle compensation: (twenty three) in, Indicates the second angular axis The compensated angle position Indicates the second angular axis The angle position before compensation, Indicates the second angular axis Compensation amount.

[0056] Step 6.4: Based on the compensated angular positions obtained in steps 6.2 and 6.3, re-acquire the interferometric image and perform fringe appearance determination and pose calculation according to the method in step 5. If the calculated residual attitude deviation meets the preset leveling criterion, it is determined that the optical element to be installed has completed automatic leveling relative to the reference optical surface; if the preset leveling criterion is not met, compensation can continue to be performed or a re-inspection status can be output according to the process requirements.

[0057] The beneficial effects of this invention are as follows: (1) This invention combines assembly force control in the contact stage with interference vision detection in the non-contact stage to sequentially complete assembly force signal acquisition and zeroing, assembly force closed-loop contact control, coarse leveling of dual-angle axis contact based on assembly force gradient and variable step size critical search, interference background reference extraction, interference fringe appearance determination and pose calculation, and robot angle precision compensation. This enables the optical element to be assembled to maintain controlled contact and complete coarse leveling of the first angle axis RX and the second angle axis RY when the optical element has not yet formed effective interference fringes in the initial stage, thereby reducing the limitation of the initial fringe state on the automatic assembly process.

[0058] (2) Simultaneously, this invention performs critical search using multiple decreasing angle step sizes, enabling the angle adjustment process to gradually transition from a large-range search to a smaller-range correction. By dividing the current interferometric image into multiple fringe analysis regions and combining the interference background benchmark, main peak signal-to-noise ratio, main peak salience, spatial frequency range, and fringe quality evaluation value to complete the determination of effective interference fringes and the selection of target fringe analysis regions, the influence of background interference, local occlusion, and invalid spectral main peaks on the pose calculation results can be reduced.

[0059] (3) Furthermore, the present invention calculates the total tilt angle and direction angle based on the main peak of the target spectrum and decomposes them into compensation amounts of two angular axes to control the robot to perform precise angular compensation, thereby forming an automatic assembly and adjustment process that transitions from coarse leveling by assembly force contact to precise compensation by interference vision. This helps to reduce the dependence on manual adjustment experience and improve the repeatability, process observability and controllability of the leveling process of optical components.

[0060] In summary, this invention can complete controlled contact and coarse leveling of dual-angle axis contact using assembly force feedback before the effective interference fringes appear, and complete the selection of target fringe analysis area and precise compensation of robot angle using frequency domain fringe analysis after the effective interference fringes appear, thereby forming an automatic assembly and adjustment process for optical components that is continuously connected from force feedback to visual feedback. Attached Figure Description

[0061] Figure 1 This is the overall flowchart of an automatic assembly and adjustment method for optical components based on force-vision hybrid control.

[0062] Figure 2 It is a flowchart of coarse leveling with multiple steps based on force changes.

[0063] Figure 3 This is a schematic diagram of the optical element to be installed and its assembly contact surface in an embodiment of the present invention.

[0064] Figure 4 This is a schematic diagram of the image division of the stripe analysis region in an embodiment of the present invention.

[0065] Figure 5 This is a schematic diagram of the main peak and frequency components of the interference spectrum in an embodiment of the present invention. Detailed Implementation

[0066] The present invention will be further described below with reference to specific embodiments.

[0067] This embodiment combines Figures 1 to 5 Please provide an explanation. Figure 1 The overall flow of an automated assembly and adjustment method for optical components based on force-vision hybrid control is shown. The overall flow executes sequentially as follows: assembly force signal acquisition and zeroing, assembly force closed-loop contact control, coarse leveling of dual-angle axis contact based on assembly force gradient and variable step-size critical search, interference background reference extraction, interference fringe appearance determination and pose calculation, and precise robot angle compensation. The automated assembly and adjustment method for optical components based on force-vision hybrid control includes the following steps: The first step is to acquire and zero the assembly force signal.

[0068] Step 1.1: Acquire the voltage signal output by the force sensor, and record it as the force sensor output voltage. ,in Indicates time The force sensor output voltage, Indicates time. When the optical element to be installed has not yet made effective contact with the reference optical surface, the time is... Zero-point sampling is performed to obtain the zero-point voltage. ,in This indicates the voltage reference of the force sensor under zero assembly force reference conditions.

[0069] Step 1.2, based on the results obtained in Step 1.1 and Convert the force sensor voltage signal into an assembly force signal. : (1) in, Indicates time Assembly force signal, This indicates the calibration coefficient for the force sensor voltage to the assembly force.

[0070] Step 1.3, based on the results obtained in Step 1.2 The assembly force was smoothed by using a filtering method. : (2) in, Indicates time Smooth assembly force, This represents the filtering operator used for assembly force signals. The smoothing assembly force... It serves as the input for subsequent assembly force closed-loop contact control and dual-angle shaft contact coarse leveling.

[0071] In this embodiment, the voltage-assembly force calibration coefficient of the assembly force acquisition channel is set to 6727.895 mN / V. During zeroing, 20 voltage samples are continuously acquired without the optical component to be assembled making effective contact with the reference optical surface. The interval between adjacent samples is 5 ms, and the average value of the 20 samples is taken as the zero-point voltage. After zeroing, the filter history is cleared to avoid the data before zeroing affecting subsequent assembly force calculations.

[0072] In this embodiment, the assembly force processing cycle is 20ms. The converted assembly force signal is input into a two-stage filter to obtain a smooth assembly force with an update frequency of 50Hz. After zeroing and filtering, the assembly force display under no-load conditions returns to near zero, and instantaneous sampling spikes are suppressed, thus providing a stable input for closed-loop contact control of the assembly force and comparison of the average assembly force before and after angle adjustment.

[0073] The second step is to perform closed-loop contact control of the assembly force to obtain a controlled contact state.

[0074] Based on the smooth assembly force obtained in the first step The robot is controlled to move along the normal direction between the optical element to be assembled and the reference optical surface. Specifically, this includes setting the target assembly force, determining the proximity to the target contact state, calculating the assembly force error, calculating the normal displacement compensation amount, and executing the normal compensation movement to obtain a controlled contact state for subsequent coarse leveling of the dual-angle axis contact. In this embodiment, Figure 2 The process of coarse leveling with multiple steps based on force variation is shown.

[0075] Step 2.1: Set the target assembly force to be maintained during the assembly process. The robot is controlled to approach the reference optical surface along the normal direction, and the smooth assembly force obtained in the first step is read in real time. .

[0076] Step 2.2, based on the results obtained in Step 2.1 and Determine if the optical component to be installed is close to the target contact state. Set the closed-loop cutting force threshold. ,when When the closed-loop entry condition is not yet met, the robot approaches the reference optical surface with gradually decreasing normal displacement increments; when When the closed-loop entry condition is met, proceed to step 2.3 to start assembly force closed-loop control.

[0077] Step 2.3: Based on the position of the optical element to be installed in the near-target contact state obtained in Step 2.2, initiate closed-loop control of the assembly force. Define the time. Assembly force error : (3) in, Indicates time Assembly force error. Step 2.4, based on the results obtained in step 2.3. Calculate the robot's normal displacement compensation amount : (4) in, Indicates time The normal displacement compensation amount, This represents the proportionality coefficient. Represents the integral coefficient. This represents the differential coefficient.

[0078] Step 2.5, based on the results obtained in Step 2.4 The robot is controlled to perform compensating motion along the normal direction to obtain the position of the optical element to be installed in the closed-loop contact state. : (5) in, Indicates the normal position at the next control moment. This indicates the normal position at the current control moment. This yields the controlled contact state used for subsequent coarse leveling of the dual-angle shaft contact.

[0079] In this embodiment, the target assembly force is set to 300 mN, and the proportional, integral, and derivative coefficients are set to 0.65, 0.10, and 0.01, respectively. The closed-loop control cycle is set to 20 ms. The robot first moves to a preset safe approach position. Before the smooth assembly force reaches the closed-loop entry condition, it approaches the reference optical surface with normal displacement increments of 0.0006 mm, 0.00016 mm, and 0.00003 mm, respectively, based on the ratio of the current assembly force to the target assembly force. When the smooth assembly force reaches 200 mN, the assembly force closed-loop control is initiated.

[0080] To limit the single-cycle normal adjustment amplitude during the closed-loop control process, this embodiment restricts the PID output to the range of -0.000012mm to 0.000012mm. When the assembly force significantly exceeds the target assembly force, the robot retreats along the normal direction away from the reference optical surface; when the assembly force returns to the vicinity of the target assembly force, the PID recalculates the normal displacement compensation. This control process can continue to correct the normal position after executing the first angular axis RX or the second angular axis RY angular movement, maintaining a controlled contact state during coarse leveling.

[0081] The third step involves coarse leveling of the optical element to be assembled using dual-angle axis contact based on the assembly force gradient and variable step size critical search, to obtain the orientation of the optical element after coarse leveling.

[0082] Step 3.1, establish the set of angle axes to be adjusted First angular axis Second angle axis These represent the small rotational degrees of freedom of the robot's end effector about two mutually orthogonal attitude adjustment axes. An angle step sequence is established. , Indicates the first Angle step size, Indicates the angle step number. Indicates the number of angle step increments, and Follow The values ​​increase sequentially and then decrease. Establish a sequence of assembly force increment thresholds. ,in Indicates the first Angle step The corresponding assembly force increment threshold. Assembly force increment threshold sequence. It can be determined based on the target assembly force, assembly force noise level, allowable contact force variation range of the optical element to be assembled, and expected contact state changes corresponding to each angle step.

[0083] Figure 2 The multi-step coarse leveling process used in this embodiment is illustrated. The first angle axis RX and the second angle axis RY use the same processing logic: first, a positive probe is performed with the first angle step size, and the coarse adjustment direction is determined based on the average smooth assembly force increment before and after the probe. Then, a critical search is performed along the determined direction. When the assembly force increment is detected to exceed the current threshold, the current angle step size is reversed by one step, and the search is switched to the next smaller angle step size.

[0084] In this embodiment, the angle step sequence is set sequentially as follows: Establish assembly force increment threshold sequence Ten smooth assembly force samples were collected before and after each angle adjustment, with a sample interval of 20ms. The assembly force increment was calculated based on the average value of the samples. The cumulative angle search range for a single angle axis was limited to 1.0°.

[0085] Step 3.2, for Any angle axis currently undergoing coarse leveling. Its average smooth assembly force before positive testing is collected. : (6) in, This represents the average smooth assembly force before the positive trial. This represents the mean operation. This indicates the time window for sampling assembly force.

[0086] Step 3.3, based on the results obtained in Step 3.2 Control the robot along the current angle axis Execute the first angle step in the positive direction. The angular position after the positive trial is obtained, and the average smooth assembly force after the positive trial is collected. : (7) Step 3.4, based on the results obtained in step 3.2 And obtained in step 3.3 Calculate the increment of positive trial assembly force : (8) Step 3.5, based on the results obtained in step 3.4 Compared with the first-level assembly force increment threshold Determine the current angle axis coarse adjustment direction .like Greater than Then determine the current angle axis. The positive direction causes the contact state to tend to increase force, controlling the robot to retreat from the positive probing angle, and... Set to the negative direction; if Not greater than Then determine the current angle axis. The positive direction did not cause a significant increase in force, and Set as the positive direction. Among them, Indicates the current angle axis The coarse adjustment direction, This indicates the threshold for the first-level assembly force increment. Based on the target assembly force The amplitude of the assembly force signal noise and the allowable range of contact force variation are determined. In one optional method, Target assembly force can be taken of .

[0087] Step 3.6, based on the results obtained in step 3.5 For the current angle axis Angle step sequence Perform critical searches sequentially. In the... With each angle step, control the robot to move along... The direction is adjusted by the angle, and the average smooth assembly force before and after the angle adjustment is collected respectively. The first step is to calculate the... Adjusting the gear angle corresponds to the increase in assembly force. : (9) in, Indicates the first Average smooth assembly force after gear angle adjustment Indicates the first Average smooth assembly force before gear angle adjustment.

[0088] Step 3.7, based on the results obtained in Step 3.6 Determine the current angle axis Has the first level been reached? The critical position under the angle step size. If Greater than Then control the robot along Reverse the current angle step by one increment in the opposite direction and switch to the next smaller angle step; if Not greater than Then continue along The direction is used to perform the current angle step search.

[0089] Step 3.8: Following steps 3.2 to 3.7, assemble the set of angle axes to be adjusted. In and Coarse leveling was performed to obtain the orientation of the optical element to be installed after coarse leveling. This orientation serves as the initial orientation for subsequent interference background extraction.

[0090] In the coarse adjustment process of this embodiment, if the average smooth assembly force increment after a positive trial is greater than the first-level assembly force increment threshold, it indicates that the positive rotation increases the local contact pressure, and the robot cancels the positive trial and searches in the negative direction; otherwise, the positive direction is maintained as the coarse adjustment direction. The three-level step size search allows the robot to first determine the range of the critical position using a larger angle step size, and then gradually narrow the adjustment range by using critical backtracking and smaller angle step sizes. After the RX axis is completed, the same process is repeated for the RY axis to obtain the contact coarse adjustment plane attitude of the optical element to be installed relative to the reference optical surface.

[0091] The fourth step is to extract the interference background reference.

[0092] Step 4.1: Based on the posture of the optical component to be assembled after contact coarse leveling obtained in step 3, stop the assembly force closed-loop contact control in step 2, so that the robot no longer performs normal closed-loop compensation based on the assembly force error, that is, obtain the stopped normal closed-loop compensation state.

[0093] Step 4.2: Based on the stopping normal closed-loop compensation state obtained in Step 4.1, control the robot to move a distance away along the normal direction away from the reference optical surface. ,in The normal retraction distance is used to form a non-contact background state. A non-contact background state is achieved when the controlled contact between the optical element to be mounted and the reference optical surface is released, and the element is in a relative position where a background image can be acquired. In this embodiment, the normal retraction distance is set to 0.015 mm. After the robot completes the retraction, it waits 600 ms to allow motion vibrations and assembly force changes to decay before acquiring a background image that does not contain effective contact stripes. This operation converts the force-controlled contact stage into the non-contact background state required for interferometric visual inspection.

[0094] Step 4.3: Based on the non-contact background state obtained in Step 4.2, acquire image coordinates. Background image at the location ,in Represents the horizontal coordinate of the image. Represents the vertical coordinate of the image.

[0095] Step 4.4, based on the background image obtained in step 4.3 Based on the distribution, continuity, and effective imaging range of visible interference fringes in the image, the background image is divided into... Each stripe analysis area ,in This indicates the total number of stripe analysis regions. Indicates the index of the stripe analysis region. , Indicates the first Each stripe analysis region. The stripe analysis region This refers to the left, right, upper, lower, or ring-shaped local areas of the image, or other local image regions defined by the distribution of visible stripes.

[0096] In this embodiment, the optical element to be installed is as follows: Figure 3 As shown, the assembly contact surface refers to the optical surface on the optical element to be assembled, which is used to establish contact or near-contact with the reference optical surface. The relative tilt state between the assembly contact surface and the reference optical surface constitutes the attitude deviation that needs to be corrected by the automatic assembly and adjustment method in this embodiment. Figure 4 The image segmentation method for the stripe analysis region in this embodiment is illustrated. R1 represents the stripe analysis region on the left side of the image, and R2 represents the stripe analysis region on the right side of the image. Let the image width be W and the height be H. Then R1 is the region on the left side of the image with a width of 0.3W and a height of H, and R2 is the region on the right side of the image with a width of 0.3W and a height of H. Therefore, in this embodiment, K=2. Figure 4The left and right 30% regions correspond to R1 and R2, respectively. The central region of the image is not included in the fringe analysis region of this embodiment. R1 and R2 are used for both background spectrum amplitude extraction in the background image and for determining the effective interference fringes and calculating the pose in the current interference image.

[0097] The aforementioned left and right region division is the result of region division determined by this embodiment based on the actual visible fringe distribution, used to retain effective interference information in the crescent-shaped visible regions on both sides. For the shape of other optical elements or imaging field of view, the number K of fringe analysis regions and the boundaries of each region can be adjusted in accordance with the method described in the invention.

[0098] Step 4.5, analyze each stripe region separately. The background image in the image is subjected to frequency domain transformation to obtain the first... Each fringe analysis region in the frequency domain coordinates Background spectral amplitude at the location : (10) in, Represents the frequency domain transform operator. Represents the horizontal coordinate in the frequency domain. Represents the vertical coordinate in the frequency domain. Indicates the first Background area image.

[0099] Background spectral amplitude of each stripe analysis region Together, they constitute the interference background reference for subsequent determination of interference fringe appearance and pose calculation.

[0100] In this embodiment, the exposure time of the interferometric camera is set to 20000 μs, and the gain is set to 1.0 dB. Frequency domain transformations are performed on the background grayscale images in R1 and R2 respectively, and the background spectral amplitudes are saved to form the respective interference background references for the left and right fringe analysis regions. By establishing separate background references, the differences in imaging brightness and fixed structural textures between the left and right regions can be avoided from being mixed into the same spectrum.

[0101] The fifth step is to determine the appearance of interference fringes and calculate their pose.

[0102] Step 5.1: Based on the non-contact background state obtained in step 4, control the robot to perform a search displacement along the normal direction close to the reference optical surface. Move gradually, among which This represents the normal search displacement used to search for interference fringes.

[0103] Step 5.2: Based on each normal search position obtained in Step 5.1, acquire image coordinates. Current interferometric image at the location Following the region division rules in step 4.4, or updating the region boundaries accordingly based on the current visible interference fringe distribution, the current interference image is divided into... Current stripe analysis area ,in Indicates the first [number] in the current interferometric image Each stripe analysis area.

[0104] In this embodiment, the normal search displacement is set to 0.005 mm, and the maximum number of searches is set to 20. After each normal search movement is completed, a 350 ms wait is waited before acquiring the current interferometric image. The current interferometric image is reused. Figure 4 The region division method shown is still set to K=2, and the left 30% region R1 and the right 30% region R2 are extracted respectively to ensure that the corresponding regions in the current image and the background image have the same size and position.

[0105] Step 5.3, for each current fringe analysis region obtained in step 5.2 Perform a frequency domain transformation to obtain the current spectral amplitude. And compared with the background spectrum amplitude obtained in step 4.5 Background subtraction is performed to obtain the effective spectral amplitude. : (11) in, Indicates the first Each fringe analysis region in the frequency domain coordinates The effective spectral amplitude at that location, Indicates the first Each fringe analysis region in the frequency domain coordinates The current spectral amplitude at that location, This indicates that the background weight has been removed. Indicates the first Each fringe analysis region in the frequency domain coordinates The background spectral amplitude at that location.

[0106] Step 5.4, based on the effective spectral amplitude obtained in step 5.3 Search within the preset effective spatial frequency range for the first The candidate spectral main peak of the first stripe analysis region is obtained. Candidate spectral peak amplitude of each stripe analysis region , No. Spectral noise floor of each stripe analysis region , No. The transverse frequency components of the candidate main peak in each stripe analysis region , No. The longitudinal frequency components of the main peak in the stripe analysis region are candidate. and the Significance of the main peak in each stripe analysis region candidate .

[0107] Step 5.5, based on the amplitude of the main peak of the candidate spectrum obtained in step 5.4 and spectral noise floor Calculate the first The signal-to-noise ratio of the main peak in each stripe analysis region : (12) in, Indicates the first The signal-to-noise ratio of the main peak in each stripe analysis region. This indicates a positive number used to avoid a denominator of zero.

[0108] Step 5.6: Based on the candidate spectral main peaks and fringe quality evaluation results of each fringe analysis region obtained in Steps 5.4 and 5.5, the first... The presence of interference fringes in each fringe analysis region is determined based on the transverse frequency component of the main peak in the candidate spectrum. and longitudinal frequency components of the main peak of the candidate spectrum Calculate the first Candidate spectral main peak spatial frequency of each stripe analysis region : (13) Pre-set the main peak signal-to-noise ratio threshold Threshold for determining the significance of the main peak and effective spatial frequency range The thresholds for determining the signal-to-noise ratio of the main peak, the threshold for determining the salience of the main peak, and the effective spatial frequency range are determined based on the interferometric background reference, imaging system parameters, historical calibration data, or preset judgment rules. This indicates the lower limit of the signal-to-noise ratio of the main peak used to distinguish between the main peak of a valid candidate spectrum and spectral noise. This represents the lower limit of significance of the main peak used to distinguish the candidate spectral main peak from its neighboring spectral components. and These represent the lower and upper limits of the effective spatial frequency range, respectively.

[0109] Then the first The criteria for determining effective interference fringes in each fringe analysis region are: (14) in, Indicates the first The signal-to-noise ratio of the main peak in each stripe analysis region. Indicates the first The significance of the main peak in each stripe analysis region.

[0110] when , and When all three conditions shown in formula (14) are met simultaneously, the first condition is determined. If a valid interference fringe exists in a fringe analysis region, the region is deemed to lack a valid interference fringe if any of the criteria is not met.

[0111] In this embodiment, three candidate spectral peaks with high energy are extracted from each fringe analysis region. The signal-to-noise ratio threshold for the main peak is set to 46.0, the significance threshold for the main peak is set to 0.55, and the effective spatial frequency range is set to 0.0025 cycles / pixel to 0.12 cycles / pixel. Only when the candidate spectral peaks simultaneously meet the above three conditions are the corresponding regions added to the effective fringe analysis region set, thereby suppressing non-target spectral peaks formed by low-frequency illumination variations, high-frequency noise, and fixed structural edges.

[0112] Step 5.7: Based on the interference fringe occurrence status of each fringe analysis region obtained in Step 5.6, a set of effective fringe analysis regions is formed. ,in This represents the set of fringe analysis region indices indicating the presence of valid interference fringes. If... If it is an empty set, return to step 5.1 to continue the normal search; if If it is not an empty set, proceed to step 5.8.

[0113] Step 5.8: Based on the candidate spectral main peaks and fringe quality evaluation results of each fringe analysis region obtained in steps 5.4 to 5.6, calculate the... Quality evaluation value of each stripe analysis area and from the effective stripe analysis region set Select the target stripe analysis area : (15) (16) in, This indicates the index of the target fringe analysis region selected for pose calculation. From the main peak signal-to-noise ratio Significance of the main peak Main peak amplitude It is determined jointly by one or more of the frequency range constraints.

[0114] Step 5.9, based on the target stripe analysis region obtained in Step 5.8 Determine the transverse frequency components of the main peak of the target spectrum. and the longitudinal frequency component of the main peak of the target spectrum ,in Specifically, it refers to the spatial frequency component corresponding to the main peak of the target spectrum in the horizontal direction of the image. Specifically, it refers to the spatial frequency component corresponding to the main peak of the target spectrum in the vertical direction of the image. Calculate the spatial frequency of the main peak of the target spectrum. : (17) Figure 5 The meaning of the main peak and frequency components of the interference spectrum in this embodiment is shown. Figure 5 The spectral center in the image represents the null space frequency location, and the horizontal frequency axis... The horizontal direction of the image corresponds to the spatial frequency direction, and the vertical frequency axis represents the frequency direction. The vertical direction of the image represents the spatial frequency direction; the main peak of the target spectrum represents the effective main peak of the spectrum selected from the target fringe analysis region, and the horizontal frequency component of the main peak of the target spectrum... and the longitudinal frequency component of the main peak of the target spectrum These represent the projections of the target spectral peak onto the horizontal and vertical frequency axes, respectively. Target spectral peak spatial frequency. The target spectrum main peak direction angle is obtained by synthesizing two frequency components. Indicates the direction of the main peak of the target spectrum relative to the transverse frequency axis The directional relationship.

[0115] Step 5.10, based on the target spectrum main peak spatial frequency obtained in step 5.9 Calculate the stripe pixel spacing And according to pixel size Calculate the physical spacing of the stripes : (18) Step 5.11, based on the stripe physical spacing obtained in step 5.10 The total tilt angle between the optical element to be installed and the reference optical surface is calculated using wedge interference relations. : (19) in, This indicates the wavelength of the light source used to form the wedge interference.

[0116] In this embodiment, a light source with a wavelength of 0.000625 mm is used to form wedge interference, and the physical size corresponding to the image pixel is set to 0.0125 mm. The fringe pixel spacing is obtained based on the spatial frequency of the main peak of the target spectrum, and then converted to obtain the fringe physical spacing. Based on this, the total tilt angle between the optical element to be installed and the reference optical surface is calculated. The farther the main peak of the target spectrum is from the center of the spectrum, the higher the fringe spatial frequency, the smaller the fringe spacing, and the larger the corresponding total tilt angle.

[0117] Step 5.12, based on the transverse frequency components of the main peak of the target spectrum obtained in step 5.9. and the longitudinal frequency component of the main peak of the target spectrum Calculate the direction angle of the main peak of the target spectrum. : (20) in, This represents the arctangent function in the four quadrants.

[0118] Step 5.13, based on the total tilt angle between the optical element to be installed and the reference optical surface obtained in step 5.11. and the target spectrum main peak direction angle obtained in step 5.12 Based on the projection relationship of the main peak direction of the target spectrum onto two mutually orthogonal angular axes, the total tilt angle is... Decomposed into the absolute value of the first angular axis compensation amount |C RX |The absolute value of the second angle axis compensation|C RY |: (twenty one) Among them, |C RX | represents the absolute value of the compensation amount for the first angular axis, |C RY | indicates the absolute value of the compensation amount for the second angular axis; This represents the total tilt angle between the optical element to be installed and the reference optical surface obtained in step 5.11; This represents the orientation angle of the main peak of the target spectrum obtained in step 5.12; and These represent the projection coefficients of the total tilt angle in the first and second angular axis directions, respectively. Step 5.13 only determines the magnitude of the compensation amounts for the two angular axes, without specifying their positive or negative directions.

[0119] Step 5.14, based on the target stripe analysis region obtained in step 5.8 The location attributes and the transverse frequency components of the main peak of the target spectrum obtained in step 5.9. and the longitudinal frequency component of the main peak of the target spectrum The sign relationship, and the pre-defined mapping relationship between the image tilt direction and the robot's angular axis movement direction, are used to determine the first angular axis compensation amount. Second angle axis compensation amount Determine the sign; assign the determined sign to the value obtained in step 5.13. and The first angular axis compensation amount with the sign is obtained. Second angle axis compensation amount .

[0120] The target stripe analysis area The positional attribute represents its relative position in the image coordinate system; the sign relationship of the main peak frequency components of the target spectrum represents... and The positive and negative combinations within the specified spectral half-plane; the mapping relationship between the image tilt direction and the robot's angular axis movement direction, representing the positional attributes of the target fringe analysis region and the sign combination of the target spectral main peak frequency components, and the first angular axis. Second angle axis The correspondence between positive and negative motion directions. This mapping relationship is established before automatic assembly by controlling the robot to move along the first angular axis. Second angle axis The process involves performing a preset small positive angle motion and recording the changes in the fringe analysis region and the frequency components of the target spectral main peak in the interferometric image. For paired conjugate main peaks in the real-valued image spectrum, a predefined spectral half-plane selection rule is used to determine the target spectral main peak for sign determination, thus maintaining consistency in the sign determination results.

[0121] Step 6, according to , The calculated values ​​are used to perform precise angle compensation for the robot.

[0122] Based on the signed first angular axis compensation amount obtained in step 5 Second angle axis Compensation amount To control the robot to perform precise angle compensation, specifically including determining the compensation amount. Does the safety constraint meet the requirements? Is the first angular axis being executed? Compensation, execution of the second angle axis Compensation and confirmation of the leveling results based on the compensated interferometric image.

[0123] Step 6.1, based on the results obtained in step 5 and ,judge and Whether the angle compensation safety constraints are met. The angle compensation safety constraints include single-time angle compensation amplitude constraints and robot angle axis motion range constraints; wherein, the single-time angle compensation amplitude constraint is used to limit the maximum compensation amplitude of the first and second angle axes in a single precision compensation, and the robot angle axis motion range constraint is used to limit the compensated angle position from exceeding the robot's preset motion range. and If the angle compensation safety constraint is met, proceed to step 6.2; if any compensation amount does not meet the angle compensation safety constraint, terminate the angle precision compensation and output an abnormal status.

[0124] Step 6.2, based on the determination in step 6.1 that the safety constraints are met. Control the robot to rotate around the first angular axis Execution angle compensation: (twenty two) in, Indicates the first angular axis The compensated angle position Indicates the first angular axis The angle position before compensation, Indicates the first angular axis Compensation amount.

[0125] Step 6.3, based on the determination in step 6.1 that the safety constraints are met. Control the robot to rotate around the second angular axis Execution angle compensation: (twenty three) in, Indicates the second angular axis The compensated angle position Indicates the second angular axis The angle position before compensation, Indicates the second angular axis Compensation amount.

[0126] Step 6.4: Based on the compensated angular positions obtained in steps 6.2 and 6.3, re-acquire the interferometric image and perform fringe appearance determination and pose calculation according to the method in step 5. If the calculated residual attitude deviation meets the preset leveling criterion, it is determined that the optical element to be installed has completed automatic leveling relative to the reference optical surface; if the preset leveling criterion is not met, compensation can continue to be performed or a re-inspection status can be output according to the process requirements.

[0127] In this embodiment, the absolute values ​​of the first and second angular axis compensation amounts are both limited to within 1.0°. When the safety constraints are met, the robot sequentially performs RX axis compensation and RY axis compensation; after compensation, the interference image is reacquired and the fringe determination and pose calculation in step five are repeated. If the effective fringes can still be stably identified and the calculated residual tilt angle meets the preset leveling criterion, the automatic leveling completion status is output; if the compensation amount exceeds the safety limit, no effective fringes are detected, or the residual tilt angle does not meet the criterion, an abnormal status or a re-inspection status is output.

[0128] The operational results of this embodiment demonstrate that the assembly force closed-loop contact control can continuously correct the normal position during the RX and RY angle search process; multiple decreasing angle steps can gradually obtain the contact coarse leveling attitude through force increase judgment, critical retreat, and step size reduction. After stopping force control and retracting, the left and right fringe analysis regions can respectively complete background subtraction and effective fringe determination, and select the target spectrum main peak from the effective region to complete the angle calculation. This realizes a continuous automatic assembly process from contact force feedback coarse leveling to interferometric vision precision compensation.

[0129] The above embodiments are merely illustrative of the implementation methods of the present invention, but should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the protection scope of the present invention.

Claims

1. An automatic assembly and adjustment method for optical components based on force-vision hybrid control, characterized in that, The automatic assembly and adjustment method for optical components includes the following steps: The first step is to acquire and zero the assembly force signal to obtain a smooth assembly force; Specifically, this includes acquiring the voltage signal from the force sensor, establishing a zero-point reference, converting the voltage signal into an assembly force signal, and filtering the assembly force signal to obtain a smooth assembly force, which is used for subsequent closed-loop contact control of the assembly force and coarse leveling of the dual-angle shaft contact. The second step is to perform closed-loop contact control of the assembly force to obtain a controlled contact state. Based on the smooth assembly force obtained in the first step, the robot is controlled to move along the normal direction between the optical element to be assembled and the reference optical surface. This includes setting the target assembly force, judging the contact state close to the target, calculating the assembly force error, calculating the normal displacement compensation amount, and executing the normal compensation movement to obtain the controlled contact state for subsequent coarse leveling of dual-angle axis contact. The third step is to perform coarse leveling of the optical element to be assembled based on the assembly force gradient and variable step size critical search of the dual-angle axis contact, and obtain the attitude of the optical element to be assembled after coarse leveling of the contact. Based on the controlled contact state, the first angular axis Second angle axis The same assembly force gradient direction determination and variable step size critical search method are used for contact coarse leveling, including establishing a set of angle axes to be adjusted and an angle step size sequence, collecting the average smooth assembly force before and after positive trial, determining the coarse adjustment direction of the current angle axis, performing critical search according to the angle step size sequence from large to small, and completing coarse leveling of dual angle axes. Step 4: Extraction of the interference background reference; Based on the posture of the optical component to be assembled, the assembly force closed-loop contact control is closed and the robot is controlled to retreat along the normal direction. This includes stopping the normal closed-loop compensation, performing normal retreat, acquiring background images, determining the fringe analysis region set according to the distribution of visible interference fringes, and calculating the background spectrum amplitude of each fringe analysis region to obtain the interference background reference. Step 5: Determine the appearance of interference fringes and calculate the pose; Based on the interference background reference, the robot is controlled to gradually approach the reference optical surface along the normal direction. First, the current interference image is acquired, and the fringe analysis region in the current interference image is divided according to the distribution of visible interference fringes. Then, frequency domain transformation and background subtraction are performed on each fringe analysis region, candidate spectral main peaks of each fringe analysis region are extracted, fringe quality evaluation results are calculated, and effective fringe analysis regions are selected. Finally, based on the selected effective analysis results, the fringe spacing, total tilt angle, first angular axis compensation amount and second angular axis compensation amount are calculated. The sixth step is to perform precise compensation of the robot angle based on the calculated values; Based on the signed first and second angle axis compensation values ​​obtained in step 5, the robot is controlled to perform precise angle compensation, including determining whether the compensation value meets the safety constraints, performing first angle axis compensation, performing second angle axis compensation, and confirming the leveling result based on the compensated interference image.

2. The automatic assembly and adjustment method for optical components based on force-vision hybrid control according to claim 1, characterized in that, The first step is specifically as follows: Step 1.1, Data Acquisition Time Voltage signal output by the force sensor ; When the optical element to be installed does not make effective contact with the reference optical surface, Zero-point sampling is performed to obtain the zero-point voltage. ; Step 1.2, based on the results obtained in Step 1.1 and Convert the force sensor voltage signal into an assembly force signal. : (1) in, Indicates time Assembly force signal, The calibration coefficient representing the force sensor voltage to the assembly force; Step 1.3, based on the results obtained in Step 1.2 The assembly force was smoothed by using a filtering method. : (2) in, Indicates time The smooth assembly force is used as an assembly force signal; This represents the filtering operator used for assembly force signals.

3. The automatic assembly and adjustment method for optical components based on force-vision hybrid control according to claim 2, characterized in that, The second step is as follows: Step 2.1: Set the target assembly force to be maintained during the assembly process. The robot is controlled to approach the reference optical surface along the normal direction, and the smooth assembly force obtained in the first step is read in real time. ; Step 2.2, based on and To determine whether the optical element to be installed is close to the target contact state; Set closed-loop cut-in force threshold ,when When the closed-loop entry condition is not yet met, the robot approaches the reference optical surface with gradually decreasing normal displacement increments; when When the closed-loop entry condition is met, proceed to step 2.3 to start assembly force closed-loop control; Step 2.3: Based on the position of the optical element to be installed in the near-target contact state obtained in Step 2.2, initiate closed-loop control of the assembly force; define the time. Assembly force error : (3) Step 2.4, based on the results obtained in Step 2.3 Calculate the robot's normal displacement compensation amount : (4) in, Indicates time The normal displacement compensation amount, Represents the proportionality coefficient. Represents the integral coefficient. Represents the differential coefficient; Step 2.5, based on the results obtained in Step 2.4 The robot is controlled to perform compensating motion along the normal direction to obtain the position of the optical element to be installed in the closed-loop contact state. : (5) in, Indicates the normal position at the next control moment. This indicates the normal position at the current control moment; thus, the controlled contact state is obtained for subsequent coarse leveling of the dual-angle shaft contact.

4. The automatic assembly and adjustment method for optical components based on force-vision hybrid control according to claim 3, characterized in that, The third step specifically involves: Step 3.1, establish the set of angle axes to be adjusted First angular axis Second angle axis These represent the small rotational degrees of freedom of the robot's end effector about two mutually orthogonal attitude adjustment axes; an angle step sequence is established. ,in Indicates the first Angle step size, Indicates the angle step number. Indicates the number of angle step increments, and Follow The increase is sequentially decreased; an assembly force increment threshold sequence is established. ,in Indicates the first Angle step size The corresponding assembly force increment threshold; assembly force increment threshold sequence The determination is based on the target assembly force, assembly force noise level, allowable contact force variation range of the optical element to be assembled, and expected contact state changes corresponding to each angle step. Step 3.2, for Any angle axis currently undergoing coarse leveling. Its average smooth assembly force before positive testing is collected. : (6) in, The average smooth assembly force before positive testing. For mean calculation, The sampling time window for assembly force; Step 3.3, based on the results obtained in Step 3.2 Control the robot along the current angle axis Execute the first angle step in the positive direction. The angular position after the positive trial is obtained, and the average smooth assembly force after the positive trial is collected. : (7) Step 3.4, based on the results obtained in step 3.2 And obtained in step 3.3 Calculate the increment of positive trial assembly force : (8) Step 3.5, based on the results obtained in step 3.4 Compared with the first-level assembly force increment threshold Determine the current angle axis coarse adjustment direction ;like Greater than Then determine the current angle axis. The positive direction causes the contact state to tend to increase force, controlling the robot to retreat from the positive probing angle, and... Set to the negative direction; if Not greater than Then determine the current angle axis. The positive direction did not cause a significant increase in force, and Set as positive direction; where, Indicates the current angle axis The coarse adjustment direction, This indicates the threshold for the first-level assembly force increment. Based on the target assembly force Determine the amplitude of assembly force signal noise and the allowable range of contact force variation; Step 3.6, based on the results obtained in Step 3.5 For the current angle axis Angle step sequence Perform critical searches sequentially; in the... With each angle step, control the robot to move along... The direction is adjusted by the angle, and the average smooth assembly force before and after the angle adjustment is collected respectively. The first step is to calculate the... Adjusting the gear angle corresponds to the increase in assembly force. : (9) in, For the first Average smooth assembly force after gear angle adjustment For the first Average smooth assembly force before gear angle adjustment; Step 3.7, based on the results obtained in Step 3.6 Determine the current angle axis Has the first stage been reached? Critical position under the angle step size; if Greater than Then control the robot along Reverse the current angle step by one increment in the opposite direction and switch to the next smaller angle step; if Not greater than Then continue along The direction is used to perform the current angle step search; Step 3.8, following steps 3.2 to 3.7, assemble the set of angle axes to be adjusted. In and After coarse leveling is performed, the orientation of the optical element to be installed is obtained after contact coarse leveling.

5. The automatic assembly and adjustment method for optical components based on force-vision hybrid control according to claim 4, characterized in that, In step 3.5, Take the target assembly force 20% to 50%.

6. The automatic assembly and adjustment method for optical components based on force-vision hybrid control according to claim 5, characterized in that, The fourth step is specifically as follows: Step 4.1: Based on the posture of the optical component to be assembled after contact coarse leveling obtained in the third step, stop the assembly force closed-loop contact control in the second step, so that the robot no longer performs normal closed-loop compensation according to the assembly force error, and obtain the stopped normal closed-loop compensation state. Step 4.2: Based on the stopping normal closed-loop compensation state obtained in Step 4.1, control the robot to move a distance away along the normal direction away from the reference optical surface. When the controlled contact between the optical element to be installed and the reference optical surface is released and the element is in a relative position where a background image can be acquired, a non-contact background state is obtained. Step 4.3: Based on the non-contact background state obtained in Step 4.2, acquire image coordinates. Background image at the location ,in, Represents the horizontal coordinate of the image. Represents the vertical coordinates of the image; Step 4.4, based on the background image obtained in step 4.3 Based on the distribution, continuity, and effective imaging range of visible interference fringes in the image, the background image is divided into... Each stripe analysis area ,in This indicates the total number of stripe analysis regions. Indicates the index of the stripe analysis region. , Indicates the first Each stripe analysis region; stripe analysis region The left, right, upper, lower, and ring-shaped local areas of the image, or other local image areas defined by the distribution of visible fringes; visible interference fringes refer to alternating bright and dark fringes or local fringe segments that can be captured by the camera in the imaging field of view and are generated by the change in the gap between the optical element to be mounted and the reference optical surface. Step 4.5, analyze each stripe region separately. The background image in the image is subjected to frequency domain transformation to obtain the first... Each fringe analysis region in the frequency domain coordinates Background spectral amplitude at the location : (10) in, Represents the frequency domain transform operator. Represents the horizontal coordinate in the frequency domain. Represents the vertical coordinate in the frequency domain. Indicates the first Image of a background area; Background spectral amplitude of each stripe analysis region Together they constitute the interference background benchmark.

7. The automatic assembly and adjustment method for optical components based on force-vision hybrid control according to claim 6, characterized in that, The fifth step is specifically as follows: Step 5.1: Based on the non-contact background state obtained in step 4, control the robot to perform a search displacement along the normal direction close to the reference optical surface. Move gradually; Step 5.2: Acquire image coordinates based on each normal search position. Current interferometric image at the location ; Following the region division rules in step 4.4, or updating the region boundaries accordingly based on the current visible interference fringe distribution, the current interference image is divided into... Current stripe analysis area ,in Indicates the first [number] in the current interferometric image Each stripe analysis area; Step 5.3, for each current fringe analysis region Perform a frequency domain transformation to obtain the current spectral amplitude. And compared with the background spectrum amplitude obtained in step 4.5 Background subtraction is performed to obtain the effective spectral amplitude. : (11) in, Indicates the first Each fringe analysis region in the frequency domain coordinates The effective spectral amplitude at that location, Indicates the first Each fringe analysis region in the frequency domain coordinates The current spectral amplitude at that location, This indicates that the background weight has been removed. Indicates the first Each fringe analysis region in the frequency domain coordinates Background spectral amplitude at that location; Step 5.4, based on the effective spectral amplitude obtained in step 5.3 Search within the preset effective spatial frequency range for the first The candidate spectral main peak of the first stripe analysis region is obtained. Candidate spectral peak amplitude of each stripe analysis region , No. Spectral noise floor of each stripe analysis region , No. The transverse frequency components of the candidate main peak in each stripe analysis region , No. The longitudinal frequency components of the main peak in the stripe analysis region are candidate. and the Significance of the main peak in each stripe analysis region ; Step 5.5, based on the amplitude of the main peak of the candidate spectrum obtained in step 5.4 and spectral noise floor Calculate the first The signal-to-noise ratio of the main peak in each stripe analysis region : (12) in, Indicates the first The signal-to-noise ratio of the main peak in each stripe analysis region. This indicates a positive number used to avoid a denominator of zero; Step 5.6: Based on the candidate spectral main peaks and fringe quality evaluation results of each fringe analysis region obtained in Steps 5.4 and 5.5, the first... The state of interference fringes in each fringe analysis region is determined; based on the transverse frequency component of the main peak of the candidate spectrum... and longitudinal frequency components of the main peak of the candidate spectrum Calculate the first Candidate spectral main peak spatial frequency of each stripe analysis region : (13) Pre-set the main peak signal-to-noise ratio threshold Threshold for determining the significance of the main peak and effective spatial frequency range Among them, the main peak signal-to-noise ratio judgment threshold, the main peak significance judgment threshold, and the effective spatial frequency range are determined based on the interferometric background reference, imaging system parameters, historical calibration data, or preset judgment rules. and These represent the lower and upper limits of the effective spatial frequency range, respectively. Then the first The criteria for determining valid interference fringes in each fringe analysis region are: (14) in, Indicates the first The signal-to-noise ratio of the main peak in each stripe analysis region. Indicates the first The significance of the main peak in each stripe analysis region; when , and When all three conditions shown in formula (14) are met simultaneously, the first condition is determined. If a valid interference fringe exists in a fringe analysis region, it is determined that no valid interference fringe exists in that fringe analysis region if any judgment condition is not met. Step 5.7: Based on the interference fringe occurrence status of each fringe analysis region obtained in Step 5.6, a set of effective fringe analysis regions is formed. ,in This represents the set of fringe analysis region indices indicating the presence of valid interference fringes; if If it is an empty set, return to step 5.1 to continue the normal search; if If it is not an empty set, proceed to step 5.8; Step 5.8: Based on the candidate spectral main peaks and fringe quality evaluation results of each fringe analysis region obtained in steps 5.4 to 5.6, calculate the... Quality evaluation value of each stripe analysis area and from the effective stripe analysis region set Select the target stripe analysis area : (15) (16) in, Indicates the index of the target fringe analysis region selected for pose calculation; From the main peak signal-to-noise ratio Significance of the main peak Main peak amplitude Determined jointly by one or more of the frequency range constraints; Step 5.9, based on the target stripe analysis region obtained in Step 5.8 Determine the transverse frequency components of the main peak of the target spectrum. and the longitudinal frequency component of the main peak of the target spectrum Calculate the spatial frequency of the main peak of the target spectrum. : (17) Step 5.10, based on the target spectrum main peak spatial frequency obtained in step 5.9 Calculate the stripe pixel spacing And according to pixel size Calculate the physical spacing of the stripes : (18) Step 5.11, based on the stripe physical spacing obtained in step 5.10 The total tilt angle between the optical element to be installed and the reference optical surface is calculated using wedge interference relations. : (19) in, Indicates the wavelength of the light source used to form wedge interference; Step 5.12, based on the transverse frequency components of the main peak of the target spectrum obtained in step 5.

9. and the longitudinal frequency component of the main peak of the target spectrum Calculate the direction angle of the main peak of the target spectrum. : (20) in, Represents the arctangent function in the four quadrants; Step 5.13, based on the total tilt angle between the optical element to be installed and the reference optical surface obtained in step 5.

11. and the target spectrum main peak direction angle obtained in step 5.12 Based on the projection relationship of the main peak direction of the target spectrum onto two mutually orthogonal angular axes, the total tilt angle is... Decomposed into the absolute value of the first angular axis compensation amount |C RX |Absolute value of the second angle axis compensation |C RY |: (21) Among them, |C RX | represents the absolute value of the compensation amount for the first angular axis, |C RY | indicates the absolute value of the compensation amount for the second angular axis; This represents the total tilt angle between the optical element to be installed and the reference optical surface obtained in step 5.11; This represents the orientation angle of the main peak of the target spectrum obtained in step 5.12; Step 5.14, based on the target stripe analysis region obtained in step 5.8 The location attributes and the transverse frequency components of the main peak of the target spectrum obtained in step 5.

9. and the longitudinal frequency component of the main peak of the target spectrum The sign relationship, and the pre-defined mapping relationship between the image tilt direction and the robot's angular axis movement direction, are used to determine the first angular axis compensation amount. Second angle axis compensation amount Determine the sign; assign the determined sign to the value obtained in step 5.

13. and The first angular axis compensation amount with the sign is obtained. Second angle axis compensation amount .

8. The automatic assembly and adjustment method for optical components based on force-vision hybrid control according to claim 7, characterized in that, In the fifth step, the target stripe analysis area The positional attribute represents its relative position in the image coordinate system; the sign relationship of the main peak frequency components of the target spectrum represents... and The positive and negative combinations within the specified spectral half-plane; the mapping relationship between the image tilt direction and the robot's angular axis motion direction; representing the positional attributes of the target fringe analysis region and the sign combination of the target spectral main peak frequency components, and the first angular axis. Second angle axis The correspondence between positive and negative motion directions; the mapping relationship is established before automatic assembly by controlling the robot to move along the first angular axis respectively. Second angle axis The process involves performing a preset small positive angle motion and recording the changes in the fringe analysis region and the frequency components of the target spectral main peak in the interferometric image. For paired conjugate main peaks in the real-valued image spectrum, a predefined spectral half-plane selection rule is used to determine the target spectral main peak for sign determination, in order to maintain the consistency of the sign determination results.

9. The automatic assembly and adjustment method for optical components based on force-vision hybrid control according to claim 8, characterized in that, The sixth step is specifically as follows: Step 6.1, based on the results obtained in step 5 and ,judge and Whether the angle compensation safety constraints are met; the angle compensation safety constraints include single angle compensation amplitude constraints and robot angle axis motion range constraints; wherein, the single angle compensation amplitude constraint is used to limit the maximum compensation amplitude of the first and second angle axes in a single precision compensation, and the robot angle axis motion range constraint is used to limit the compensated angle position from not exceeding the robot's preset motion range; if and If the angle compensation safety constraint is met, proceed to step 6.2; if any compensation amount does not meet the angle compensation safety constraint, terminate the angle precision compensation and output an abnormal status. Step 6.2, based on the determination in step 6.1 that the safety constraints are met. Control the robot to rotate around the first angular axis Execution angle compensation: (22) in, Indicates the first angular axis The compensated angle position Indicates the first angular axis The angle position before compensation Indicates the first angular axis Compensation amount; Step 6.3, based on the determination in step 6.1 that the safety constraints are met. Control the robot to rotate around the second angular axis Execution angle compensation: (23) in, Indicates the second angular axis The compensated angle position Indicates the second angular axis The angle position before compensation Indicates the second angular axis Compensation amount; Step 6.4: Based on the compensated angle position obtained in Steps 6.2 and 6.3, re-acquire the interference image and perform fringe appearance determination and pose calculation according to the method in Step 5; if the calculated residual attitude deviation meets the preset leveling criterion, it is determined that the optical element to be installed has completed automatic leveling relative to the reference optical surface; if the preset leveling criterion is not met, compensation is continued or the re-inspection status is output according to the process requirements.

Images