Optical active assembly method and system based on event wavefront sensing

Abstract

The application discloses an optical active assembly method and system based on an event wavefront sensor, and belongs to the technical field of optical active assembly and precise optical adjustment. The system mainly comprises four main modules of an optical system to be assembled, an event wavefront sensor, an optical active assembly decision terminal and an assembly mechanical mechanism, is divided into two parts of a modulation stage and an assembly stage, the modulation stage captures the time when the first positive event of each pixel trigger of a focal spot is triggered by modulating incident light using an event camera, and then adopts an elliptical Gaussian model to parameterize and represent the corresponding focal spot state; in the assembly stage, the event camera is used to capture the slight change of the brightness of the focal spot in real time, and a local linear mapping is constructed in combination with a degree of freedom action tensor, and an incremental optimization strategy is adopted to determine the assembly degree of freedom adjustment amount at each moment until the ideal state is reached. The method can realize high-speed, stable and closed-loop automatic alignment and active assembly of optical components.

Description

Technical Field

[0001] This invention belongs to the field of active optical assembly and precision optical calibration technology, specifically relating to an active optical assembly method and system based on event wavefront sensing. Background Technology

[0002] In the manufacturing and assembly of precision optical systems such as biological microscopic imaging systems, freeform mirror arrays, photolithographic projection lenses, and space telescopes, positional and orientation errors of optical components directly lead to a decline in image quality. Therefore, how to perform high-precision, rapid, and stable active assembly and alignment of optical components has become a crucial issue in optical engineering.

[0003] Current optical assembly and alignment methods mainly include three categories: pure mechanical / geometric reference alignment, image quality evaluation-based alignment, and wavefront sensing-driven alignment. Among them, wavefront sensing-driven alignment often uses an interferometer or a Shaker-Hartmann wavefront sensor (SHWFS) to obtain the wavefront.

[0004] In recent years, event cameras, with their high temporal resolution, low redundancy sensing, and extremely high dynamic range, have replaced traditional frame cameras in many visual sensing and imaging fields, achieving better results. Similarly, the characteristics of event cameras can achieve superior wavefront sensing performance in the SHWFS field. However, current research mainly focuses on "wavefront reconstruction" and has not further explored how to apply event wavefront sensing to the active assembly and closed-loop adjustment of degrees of freedom in practical optical systems.

[0005] Existing active alignment (AA) systems generally follow a "one-step prediction" approach, which has the following fundamental drawbacks: First, the mapping relationship between optical state and mechanical degrees of freedom is highly nonlinear and the coupling between degrees of freedom is severe, causing the inverse kinematics process to exhibit ill-conditioned characteristics, resulting in unstable assembly commands. Second, traditional methods mostly rely on model-based predictive control, which is extremely sensitive to wavefront noise and model errors, and is prone to adjustment overshoot or oscillations. Third, the "one-step" adjustment strategy cannot take into account both mechanical travel limitations and mutual constraints between multiple degrees of freedom, and cannot achieve stable convergence in high-dimensional parameter spaces. Fourth, traditional imaging sensing methods lack real-time feedback with high temporal resolution, and cannot capture transient wavefront changes during the assembly process, making closed-loop, correctable, and highly dynamic active alignment almost impossible to achieve.

[0006] In summary, the traditional AA paradigm, which "inversely derives the adjustment amount of the ideal degrees of freedom from the initial optical state", is inherently highly dependent on the ill-conditioned inverse model, sensitive to noise, lacks real-time feedback, and has an uncontrollable mechanical execution process, making it difficult to achieve stable, high-speed, and robust active assembly in complex optical systems. Summary of the Invention

[0007] To address the aforementioned issues, this invention achieves high-speed, stable, and closed-loop automatic alignment and active assembly of optical components through event-driven wavefront information extraction, real-time estimation of wavefront change trends, and dynamic adjustment indication of optical degrees of freedom. This invention is applicable to various scenarios including optical system assembly and adjustment, free-space optical communication, laser coupling, precision optical instrument manufacturing, micro-optical component assembly, and collimation of biological microscopic imaging systems.

[0008] In a first aspect, this invention proposes an optical active assembly system based on event wavefront sensing. The system includes: The optical system module to be assembled consists of a set of fixed mirrors and a set of adjustable mirrors, used to receive modulated incident light and obtain the outgoing wavefront. During the modulation stage, the positions of both the fixed mirrors and the adjustable mirrors remain unchanged. During the assembly stage, the position of the fixed mirrors remains fixed, while the adjustable mirrors are adjusted by the assembly mechanical mechanism module. The event wavefront sensor module consists of a microlens array and an event camera. It is used to receive the outgoing wavefront of the optical system module to be assembled. After being decomposed by the microlens array, the outgoing wavefront forms multiple focal spots on the image plane of the event camera that correspond one-to-one with the sub-apertures. During the modulation stage, the event wavefront sensor module outputs the time when each pixel on the image plane triggers the first positive event. During the assembly stage, the event wavefront sensor module outputs the total number of positive events and the total number of negative events triggered by each pixel on the image plane. The optical active assembly decision terminal module is used to modulate the incident light during the modulation stage to obtain the modulated incident light, and to receive the moment when each pixel triggers the first positive event transmitted by the event wavefront sensor module, and finally obtain the initial state vector of the focal spot corresponding to each sub-aperture. During the assembly phase, the optical active assembly decision terminal module receives the initial state vector of the focal spot corresponding to each sub-aperture, all degrees of freedom physical quantities of the assembly mechanical mechanism module, and the total number of positive events and the total number of negative events triggered by each pixel transmitted by the event wavefront sensor module. It then generates the degree-of-freedom action tensor, the current sub-aperture focal spot state vector, and the degree-of-freedom adjustment amount for the next moment. It also determines whether the current global focal spot state vector satisfies the ideal assembly state. If it does, the assembly terminates; otherwise, the assembly continues. The global focal spot state vector is composed of the sub-aperture focal spot state vectors of all sub-apertures. The assembly mechanical mechanism module is used to receive the next-moment degree of freedom adjustment amount transmitted by the optical active assembly decision terminal module, and to adjust the degree of freedom of the adjustment lens group according to the degree of freedom adjustment decision.

[0009] Furthermore, during the modulation phase, the incident light intensity is driven to increase monotonically by the optical active assembly decision terminal module; during the assembly phase, the incident light intensity remains unchanged at the maximum intensity during the modulation phase.

[0010] Furthermore, the initial state vector of the focal spot corresponding to each sub-aperture is specifically calculated as follows: Based on the time-delay-intensity mapping principle, the moment when each pixel on the focal spot triggers its first positive event is converted into the pixel's initial relative intensity: ; in, For pixels The moment when the first positive event is triggered, where T is the total duration of the modulation phase. For pixels The initial relative intensity; The initial relative intensity of all focal spot pixels is calculated by iterating through them to obtain the initial focal spot intensity map of the outgoing wavefront; The focal spot state of each sub-aperture corresponding to the focal spot is parameterized by using an elliptical Gaussian model to obtain the focal spot state vector; Based on the focal spot state of each sub-aperture corresponding to the initial focal spot intensity map, substitute it into the focal spot state vector to obtain the initial state vector of the focal spot corresponding to each sub-aperture.

[0011] Furthermore, the specific expression for the focal spot state vector is as follows: ; in, Let i be the focal spot state vector corresponding to the i-th sub-aperture. Let be the coordinates of the centroid of the focal spot corresponding to the i-th sub-aperture. Let be the focal spot energy corresponding to the i-th sub-aperture. The focal spot corresponding to the i-th sub-aperture is in and Dimension of directional expansion The correlation coefficient of the elliptic Gaussian model is shown in the upper right corner. This indicates transpose.

[0012] Furthermore, the current sub-aperture focal spot state vector is specifically calculated as follows: The brightness increment value of each pixel in the current event stream is calculated based on the total number of positive events and the total number of negative events triggered by each pixel in the current event stream transmitted by the event wavefront sensor module. A first-order Taylor expansion of the elliptical Gaussian model yields a linear relationship between the brightness increment of each pixel and the focal spot state increment corresponding to the sub-aperture. Substitute the brightness increment value of each pixel in the current event stream into the linear relationship to obtain the focal spot state increment corresponding to each sub-aperture in the current event stream; The current event flow sub-aperture focal spot state increment is added to the sub-aperture focal spot state vector of the previous event flow to calculate the current event flow sub-aperture focal spot state vector; this process continues until all event flows at the current time are traversed to obtain the final event flow sub-aperture focal spot state vector. Exponential smoothing is introduced to obtain the current focal spot state vector of the sub-aperture based on the previous moment's focal spot state vector and the final event flow's focal spot state vector. The calculation formula is as follows: ; in, For smoothing coefficients, The smoothed version of the current time t. Individual aperture focal spot state vectors Let be the smoothed state vector of the i-th sub-aperture focal spot at time t-1. Let be the state vector of the i-th sub-aperture focal spot under the final event flow.

[0013] Furthermore, the moment includes multiple event streams, each of which is transmitted sequentially to the optical active assembly decision terminal module according to its trigger time.

[0014] Furthermore, the determination of whether the global focal spot state vector at the current moment satisfies the ideal assembly state specifically includes: Determine whether the current global focal spot state vector meets the first determination requirement. If it meets the first determination requirement, the assembly terminates. If it does not meet the first determination requirement, proceed to the next step of judgment. Determine whether the current global focal spot state increment meets the second determination requirement. If it does, update all sub-aperture focal spot state vectors in the current global focal spot state vector to the sub-aperture focal spot state vectors of the next moment and end the calculation. If it does not meet the requirement, proceed to the next determination step. If the action tensor of the degrees of freedom satisfies the third criterion, then the adjustment decision of the degrees of freedom for the next time step is generated, all sub-aperture focal spot state vectors in the current global focal spot state vector are updated to the sub-aperture focal spot state vectors of the next time step, and the calculation ends. If the condition is not met, then the action tensor of the degrees of freedom is modified, the adjustment amount of the degrees of freedom for the next time step is generated, all sub-aperture focal spot state vectors in the current global focal spot state vector are updated to the sub-aperture focal spot state vectors of the next time step, and the calculation ends.

[0015] Furthermore, the first determination requirement is: at the current moment, the line connecting the position of the centroid of the focal spot corresponding to the i-th sub-aperture and the position of the center coordinates corresponding to the i-th sub-aperture is less than or equal to the first threshold; at the current moment, the absolute value of the correlation coefficient of the elliptic Gaussian model is less than or equal to the second threshold. The second determination requirement is: the inner product of the current global focal spot state increment and the preset ideal assembly direction is greater than 0; The third determination requirement is: ; in, For norm, This represents the increment of the global focal spot state at the current moment. For the action tensor of degrees of freedom, The adjustment amount for the degrees of freedom at the current moment. The preset threshold; The global focal spot state increment includes the focal spot state increment vectors corresponding to all sub-apertures. Furthermore, the specific calculation content of the aforementioned degree of freedom action tensor is as follows: All the physical quantities of the assembly mechanical mechanism module are parameterized into a single state vector of freedom. Apply degree-of-freedom adjustment amounts to all components of the degree-of-freedom state vector, and calculate the global focal spot state increment caused by each degree-of-freedom adjustment amount. Normalize the global focal spot state increment caused by the adjustment of each degree of freedom to obtain the action vector of each degree of freedom; The action vectors of all degrees of freedom are combined to obtain the action tensor of the degrees of freedom.

[0016] Furthermore, the specific content of generating the degree-of-freedom adjustment amount for the next moment is as follows: It is calculated using a progressive optimization strategy based on the global focal spot state vector and the degree-of-freedom action tensor at the current moment, and the calculation formula is: ; in, Let be the adjustment amount of the degrees of freedom at time t+1. Step size factor Tensor for Degrees of Freedom The false rebellion.

[0017] Secondly, this invention proposes an optical active assembly method based on the aforementioned system.

[0018] The beneficial effects of this invention are: (1) In view of the nonlinearity and coupled ill-conditioning of the mapping between optical state and mechanical degree of freedom in one of the problems of existing active optical assembly technology, this invention captures the small changes in the brightness of the focal spot in real time by an event camera, uses a Gaussian ellipse model for parameterization, and constructs a local linear mapping by combining the action tensor of the degree of freedom, thereby avoiding ill-conditioning wavefront reconstruction, obtaining stable and analyzable state variables, and significantly improving the stability and interpretability of the inverse solution process. (2) In view of the problem that predictive control is highly sensitive to noise and model error, and the problem that the "one-step" strategy is difficult to converge stably in a high-dimensional space of freedom, this invention utilizes the high time resolution characteristics of event sensors to drive a step-by-step approximation closed-loop control strategy with continuous small incremental observations, so that each step of adjustment has a clear direction and controlled amplitude, effectively suppressing noise amplification and improving assembly convergence. (3) In view of the limitations of traditional methods, such as lack of real-time feedback and inability to capture transient wavefront changes, this invention relies on the approximate continuous time feedback provided by the event camera to realize a highly dynamic assembly process that can be monitored, corrected, and continuously approximated, which significantly improves the assembly efficiency, robustness and final assembly quality of complex optical systems. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the optical active assembly system used in this invention; Figure 2 This is the initial focal spot state acquisition module used in this invention; Figure 3 This is the real-time focal spot status update module used in this invention; Figure 4 This is the freedom degree action feature calibration module used in this invention; Figure 5 This is the assembly path correction module based on real-time perception used in this invention; Figure 6 This is the experimental result of capturing the initial sub-aperture focal spot intensity map and parametrically describing it according to the present invention; Figure 7 This is the experimental result of the present invention for real-time focal spot status update of a sub-aperture; Figure 8 This invention describes the experimental process and results of calibrating the degree-of-freedom action characteristics. Detailed Implementation

[0020] The present invention will be further described and illustrated below with reference to specific embodiments. The embodiments described are merely examples of the content of this disclosure and do not limit the scope of the invention. The technical features of each embodiment in the present invention can be combined accordingly, provided that there is no mutual conflict.

[0021] like Figure 1As shown, the present invention provides an optical active assembly system based on event wavefront sensing, comprising four main modules: an optical system to be assembled, an event wavefront sensor, an optical active assembly decision terminal, and an assembly mechanical mechanism.

[0022] I. Optical system module to be assembled The optical system module to be assembled is divided into a fixed lens group and an adjustable lens group. The fixed lens group remains stationary during the assembly process, while the adjustable lens group is adjusted in multiple degrees of freedom by the assembly mechanical mechanism module, such as translation and rotation, to change its relative position and attitude with the fixed lens group, thereby changing the outgoing wavefront.

[0023] II. Event Wavefront Sensor Module The event wavefront sensor module consists of a microlens array and an event camera. The microlens array divides the outgoing wavefront of the optical system to be assembled into several sub-apertures, which are focused onto different regions of the image plane of the event camera to form corresponding sub-aperture focal spots.

[0024] The event wavefront sensor module is divided into two stages: the modulation stage before assembly begins and the assembly process.

[0025] During the modulation phase before assembly begins, the illumination intensity of the incident light source is changed by the optical active assembly decision terminal module. During this process, each pixel on the image plane of the event camera... The received light intensity gradually increases from 0. When the pixel... The change in light intensity reached the artificially set event trigger threshold. When a pixel triggers its first positive event, it is considered to have done so. The event camera records the time when each pixel on the event camera's image plane triggers its first positive event and records this time as the time delay. .

[0026] Ultimately, the event wavefront sensor module will determine the time delay of the first positive event triggered by each pixel on the image plane. Transmitted to the optical active assembly decision terminal module.

[0027] During assembly, the incident light intensity remains constant. As the adjustment lens group in the optical system module to be assembled moves, the outgoing wavefront also changes. During this process, the event camera senses the change in logarithmic light intensity of each pixel on the image plane in real time. Logarithmic light intensity, also known as luminance, is defined in relation to light intensity as follows: in, The light intensity of that pixel. This represents the brightness of the pixel.

[0028] Within a given time window, when the brightness of a pixel increases, if the change in pixel brightness exceeds a threshold... If the pixel's brightness decreases, it is considered to have triggered a positive event; if the decrease exceeds a threshold, it is considered to have triggered a positive event. If a pixel triggers a negative event, it is considered to have triggered a negative event. The data transmission of the event camera is asynchronous. Within a given moment, the event camera generates multiple event streams in chronological order and transmits them one by one to the optical active assembly decision terminal module. Each event stream contains the number of all positive and negative events captured by each pixel at that moment.

[0029] III. Optical Active Assembly Decision Terminal Module The optical active assembly decision terminal module is the core component, used to sense the wavefront state in real time, correlate the wavefront with the assembly degrees of freedom, correct the assembly path in real time, and provide assembly instructions to the assembly machinery. It consists of four sub-modules: initial focal spot state acquisition module, real-time focal spot state update module, degree of freedom action feature calibration module, and assembly path correction module based on real-time sensing.

[0030] Submodule S1: Initial Focal Spot State Acquisition Module like Figure 2 As shown, the initial focal spot state acquisition module is only called before assembly begins. Its main purpose is to obtain the initial sub-aperture focal spot intensity map of the optical system to be assembled, and based on this, extract the parameterized focal spot state vector for each sub-aperture focal spot. This module includes the following three sub-units.

[0031] S1.1) Active modulation of incident light This unit is mainly used for monitoring the overall intensity of the incident light source. Continuous modulation is performed so that the light intensity of each pixel in the focal spot region increases monotonically with time. During assembly, the illumination intensity of the incident light source remains constant at its maximum intensity, i.e., the illumination intensity at time T. In one embodiment of the invention, the incident light intensity... The following cosine-type rising modulation method is adopted: in, The maximum light intensity set by humans. T represents the modulation phase time, and T represents the total duration of the illumination modulation phase.

[0032] S1.2) Delay-Intensity Mapping Unit This unit is mainly used to receive the time delay of each pixel triggering the first positive event during the modulation phase of the event wavefront sensor module's transmission on the image plane. And based on the time-delay-intensity mapping principle, Mapped to the initial relative intensity of each pixel .

[0033] In one embodiment of the present invention, the following analytical mapping relationship is adopted: By performing this mapping on all pixels, an initial focal spot intensity map of the outgoing wavefront can be obtained during a single incident light modulation process. .

[0034] S1.3) Sub-aperture focal spot state parameterization unit Microlens array contains Each effective microlens, after passing through the microlens array, divides the emitted wavefront into N sub-aperture wavefronts. Each sub-aperture wavefront forms a corresponding focal spot on the image plane of the event camera. Accordingly, the image plane can be divided into... The aperture region is denoted as ,in Indicates the first Each microlens corresponds to a focal spot area.

[0035] To efficiently describe the focal spot morphology during subsequent assembly, this invention uses an elliptical Gaussian model for the i-th sub-aperture region. The focal spot intensity within the area is parameterized. This can be written using an elliptic Gaussian model: in, Let be the coordinates of the centroid of the focal spot corresponding to the i-th sub-aperture. Let be the focal spot energy corresponding to the i-th sub-aperture; The covariance matrix is ​​2×2, which describes the shape of the focal spot; It is a Gaussian function.

[0036] Gaussian function It can be written as: covariance matrix It can be written as: in The focal spot corresponding to the i-th sub-aperture is in and Dimension of directional expansion The correlation coefficient is the value of the elliptic Gaussian model.

[0037] Therefore, the state of the focal spot can be naturally represented as a six-dimensional vector: in, Let be the focal spot state vector corresponding to the i-th sub-aperture under the elliptic Gaussian model, used to describe The state of the inner keratin spot.

[0038] In the initial stage of assembly, the above relationship is used to determine the initial focal spot intensity map. By parametrically describing each sub-aperture, the initial focal spot state vector of each sub-aperture at the start of assembly can be obtained, denoted as: .

[0039] Submodule S2: Real-time Focal Spot Status Update Module like Figure 3 As shown, the real-time focal spot state update module runs continuously throughout the assembly process, during which the assembly mechanical mechanism adjusts the optical degrees of freedom. This module utilizes the event stream output by the event camera, combined with the initial focal spot state vector obtained in sub-module S1, to update the focal spot state vector of each sub-aperture in real time without reconstructing the complete focal spot intensity map.

[0040] Module S2 contains the following four sub-units.

[0041] S2.1) Event Brightness Change Acquisition Unit This unit receives the event stream transmitted by the event wavefront sensor module and records the total number of positive and negative events contained in the event stream as follows: and Based on the event triggering model and the event stream transmitted by the event wavefront sensor module during the assembly phase, the brightness increment of each pixel can be calculated. : For each sub-aperture region The brightness increments of all pixels within the sub-aperture are denoted by their coordinates as follows: The change in brightness can be used as a direct observation for subsequent calculation of the focal spot state increment corresponding to each sub-aperture.

[0042] S2.2) Analytical linearization unit of the focal spot model During the assembly process, this unit determines the current state at time t based on the elliptic Gaussian model obtained from submodule S1. The data is linearized analytically to establish the relationship between brightness changes and state changes.

[0043] For the elliptic Gaussian model in Performing a first-order Taylor expansion in the vicinity yields a linear relationship between the brightness increment and the state increment: in, This is the focal spot state increment vector corresponding to the i-th sub-aperture at the current time, representing the difference between the current focal spot state and the initial focal spot state. Changes between them; Let the analytical Jacobian vector of the elliptic Gaussian model corresponding to the i-th sub-aperture be expressed as: According to the elliptic Gaussian model, we can obtain: S2.3) Incremental solution element for focal spot state This unit will use the observations obtained from the event brightness change acquisition unit in S2.1). Substituting the analytically linearized model constructed from the analytically linearized unit of the focal spot model in S2.2), the linear least squares problem is constructed as follows: in, is the state increment vector to be solved in the linear least squares problem, representing the incremental change of the focal spot state parameters.

[0044] The focal spot state increments of all sub-apertures are stacked in sub-aperture index order, and the current global focal spot state increment vector is output. : S2.4) State update and steady-state filtering unit This unit utilizes the focal spot state increments obtained from the S2.3) focal spot state increment solution unit to solve for the focal spot state increments of each sub-aperture. Update the focal spot state; the updated state vector of the i-th sub-aperture focal spot is: : in, Let be the focal spot state vector of the i-th sub-aperture during the previous update.

[0045] For all event streams generated within time t, the i-th sub-aperture generated in the previous event stream will be... As the event flow of the i-th sub-aperture Repeat the above relational calculations until all event streams are traversed to obtain the final focal spot state vector of the i-th sub-aperture after time t.

[0046] To smooth out state fluctuations caused by event noise, mechanical vibration, etc., the final focal spot state vector of the i-th sub-aperture after time t is further subjected to exponential smoothing.

[0047] For the first For each aperture size, the smooth update formula is: in, This is a smoothing coefficient used to control the response speed of state updates. At the current time t, the smoothed i-th Individual aperture focal spot state vectors The smoothed-up time at time t-1 Individual aperture focal spot state vector.

[0048] Stack the smoothed state vectors of all sub-apertures in sub-aperture index order and output the current time step. Global focal spot state vector: Submodule S3: Degrees of Freedom Action Feature Calibration Module like Figure 4 As shown, the degree-of-freedom action characteristic calibration module is used to obtain the action characteristics of each degree of freedom on all sub-aperture focal spot state vectors by applying small perturbations to multiple mechanical degrees of freedom of the adjustment lens group before assembly begins. Submodule S3 includes the following four sub-units.

[0049] S3.1) Degrees of freedom perturbation drive unit This unit simulates the effects of different degrees of freedom on the system by applying small, controllable mechanical perturbations to the adjustment lens group of the optical system to be assembled. Assume the assembly mechanical mechanism has... There are 1 mechanical degrees of freedom, where each degree of freedom has a state. Each corresponds to a physical quantity of a degree of freedom, such as axial and lateral displacement, and meridional and sagittal tilt angles.

[0050] Stack all degrees of freedom states into a single degree-of-freedom vector. This unit covers each degree of freedom. The applied amplitudes are respectively Tiny perturbations: Disturbance amplitude The value is selected such that it simultaneously satisfies the following conditions: (1) the perturbation amplitude is small enough to approximate the state of the focal spot as linear with respect to the perturbation of the degrees of freedom; and (2) the perturbation amplitude is large enough to trigger a sufficient number of events for solving the state increment.

[0051] S3.2) Incremental acquisition unit for focal spot state caused by perturbation This unit utilizes submodule S2 to obtain the perturbation for each degree of freedom. The resulting global focal spot state increment vector ∆S is denoted as... .

[0052] S3.3) Solving unit for the action vector of degrees of freedom This unit perturbs each degree of freedom. Increased state of focal spot Normalizing to a perturbation of unit degrees of freedom yields the action vector for each degree of freedom. : Action vector The direction and magnitude of the effect of a unit degree of freedom change on the focal spot state of all sub-apertures are described.

[0053] S3.4) Action Tensor Construction and Storage Unit Combine the action vectors of all degrees of freedom into a degree of freedom action tensor : Each column corresponds to one degree of freedom, and each row corresponds to one dimension of the sub-aperture focal spot state. Degrees of freedom action tensor It is a first-order linear mapping used to describe the mechanical degrees of freedom to the state of the focal spot, i.e., the action characteristics of the degrees of freedom.

[0054] Submodule S4: Assembly path correction module based on real-time perception like Figure 5 As shown, the assembly path correction module based on real-time perception is the core module in the active optical assembly process. It is responsible for dynamically correcting the assembly path according to the real-time perceived focal spot state until the optical system to be assembled reaches the ideal assembly state. Submodule S4 includes the following four sub-units.

[0055] S4.1) Ideal State Discrimination Unit This unit updates the global focal spot state vector at the current moment based on the state update and steady-state filtering unit output by submodule S2(S2.4). Determine whether the current system is close to the ideal assembly state.

[0056] Generally speaking, for each sub-aperture Under ideal assembly conditions, the focal spot state vector is: in, for The center coordinates, .

[0057] The ideal state discrimination criterion is: ,and .in and This is a preset tolerance constant used to control the discrimination accuracy. If the discrimination conditions of the ideal state are not met, proceed to S4.2 for the next discrimination step; if the discrimination conditions of the ideal state are met, the assembly terminates.

[0058] S4.2) Assembly orientation correctness judgment unit This unit outputs the global focal spot state increment based on unit S2.3 of submodule S2. Determine whether the current adjustment of degrees of freedom brings the system closer to the ideal state.

[0059] Assembly orientation correctness judgment criteria: based on the preset ideal assembly state The ideal assembly direction is defined as If inner product If the inner product is less than 0, it means that the current system is moving towards the ideal state. The larger the inner product, the more correct the assembly direction. At this time, submodule S2 is called to update the real-time state and the judgment process of submodule S4 is restarted. If the inner product is less than 0, it means that the current system is moving in the wrong direction. At this time, S4.3 is entered for the next judgment.

[0060] S4.3) Functional Characteristic Consistency Detection Unit This unit is used to detect the action tensor of degrees of freedom. Does it still apply to the current assembly state?

[0061] The system at any time The degree of freedom adjustment was actually performed. And the corresponding actual focal spot state increment is sensed through submodule S2. The degree-of-freedom adjustment amount is determined by the different perturbation amplitudes applied to different mechanical degrees of freedom at the current moment. Composition. This unit calculates the degree of freedom adjustment decision. Increment of global focal spot state Relative deviation between: in, It is a norm.

[0062] Compare the calculated relative deviation result with the preset tolerance. Compare. If the deviation exceeds the tolerance, then determine the action tensor of the current degree of freedom. No longer accurate, calling submodule S3's S3.3), utilizing and right After correction, step 4.4) is called to generate the degree-of-freedom adjustment amount for the next moment. The assembly mechanism adjusts the lens group according to the degree-of-freedom adjustment amount, and then submodule S2 updates the real-time state of the current focal spot. Then restart the judgment process of submodule S4; if the deviation is within the tolerance range, proceed directly to S4.4.

[0063] S4.4) Degrees of freedom adjustment decision generation unit This unit provides the adjustment amount of the degree of freedom at the next moment based on the action tensor of the degree of freedom and the ideal assembly direction.

[0064] An incremental optimization strategy is used to indicate the degree of freedom adjustment for the next time step: in The step size factor is the adjustment value for the degree-of-freedom vector at the next time step. For a constant scalar, Tensor for Degrees of Freedom The false rebellion.

[0065] IV. Assembly Machinery The assembly mechanism adjusts its degrees of freedom for the next moment based on the data transmitted at each moment from the optical automatic assembly decision terminal (i.e., ...). ) Perform assembly operations to change and adjust multiple degrees of freedom of the lens assembly, such as axial displacement and lens assembly tilt.

[0066] Based on the above system components and core modules, this invention proposes an optical active assembly method based on event wavefront sensing, which is implemented according to the following steps: Step 1: Entering the modulation stage, the optical automatic assembly decision terminal module actively modulates the incident light. The incident light intensity increases monotonically with time, passing sequentially through the fixed lens group and the adjustment lens group in the optical system module to be assembled, resulting in a changing outgoing wavefront.

[0067] Step 2: The changing outgoing wavefront passes through the event wavefront sensor module, causing a change in the brightness of each pixel in the focal spot area on the event camera image plane. The time delay is recorded as the moment when each pixel triggers the first positive event.

[0068] Step 3: Based on the time delay of the first positive event triggered by each pixel in the focal spot, the initial focal spot intensity of each pixel is obtained through the time delay intensity mapping principle, thus forming an initial focal spot intensity map. Initial focal spot intensity map Through parametric analysis, the initial focal spot state of each sub-aperture was obtained. The initial global focal spot state is obtained by combining the initial focal spot states of all sub-apertures.

[0069] Step 4: At the end of the modulation phase, the optical automatic assembly decision terminal module maintains the incident light intensity at the maximum intensity achieved during the modulation phase and invokes submodule S3. Specifically, this includes: Apply small perturbations to each degree of freedom of the assembly mechanical mechanism module; Call submodule S2 to calculate the global focal spot state increment vector caused by the small perturbation corresponding to each degree of freedom; further calculate the action vector of each degree of freedom based on the global focal spot state increment vector caused by the small perturbation corresponding to each degree of freedom. The combination of the action vectors of all degrees of freedom yields the action tensor of the degrees of freedom; Based on the initial global focal spot state and the action tensor of the degrees of freedom, the initial degree of freedom adjustment amount is obtained by calling unit 4.4).

[0070] Step 5: Assemble and adjust the mechanical mechanism module according to the initial degree of freedom adjustment amount, and then enter the assembly stage.

[0071] Step Six: Call submodule S2 to calculate the current global focal spot state based on the event stream transmitted by the event wavefront sensor module. .

[0072] Step 7: Call submodule S4 to retrieve the current global focal spot state output by submodule S2. The process involves making a judgment. Specifically, this includes: Unit 4.1) Determine whether the current global focal spot state has reached the ideal state. If the condition for determining the ideal state is met, the assembly terminates; otherwise, proceed to the next step. Calling unit 4.2) determines whether the current assembly adjustment direction is correct. If the assembly direction is correct, call submodule S2 to obtain the global focal spot state at the next moment and restart the determination in step six. If the assembly direction is incorrect, proceed to the next step. Unit 4.3) determines whether the action tensor of the degree of freedom conforms to the actual observation. If it conforms to the actual observation, unit 4.4) is called to obtain the degree of freedom adjustment amount at the next moment. If it does not conform to the actual observation, unit S3.3) is called to correct the action tensor of the degree of freedom, and unit 4.4) is called again to obtain the degree of freedom adjustment decision at the next moment.

[0073] Step 8: Repeat step 6 until the assembly termination condition is met, that is, the focal spot state converges to the ideal state.

[0074] The present invention further verified the expected effect of the method by designing specific experiments, and the experimental results are as follows: Figure 6 The results of submodule S1 acquiring the initial sub-aperture focal spot intensity map and completing the parameterization description are shown. Figure (a) shows the initial focal spot intensity map output by subunit S1.2. Within this aperture range The local distribution within the focal spot. Thanks to the high dynamic range and high quantization accuracy of the time-delay-intensity mapping, this figure can present the detailed intensity structure of the focal spot. Figure (b) shows the local distribution of the focal spot in subunit S1.3 based on an elliptic Gaussian model and utilizing the initial focal spot state vector. The fitted elliptical Gaussian focal spot is shown in the table on the right. The specific fitting parameter values ​​and error analysis were presented. Experimental results show that the relative error of the fitting intensity is only 0.73%, indicating that the focal spot parameterization method of the present invention has extremely low descriptive error and good stability.

[0075] Figure 7 The results of real-time focal spot status updates performed by submodule S2 are shown. Assuming the assembly start time is 0, Figure (a) shows... Time s aperture number The true value of the focal spot intensity map; Figure (b) shows the time interval. The event frames are composed of event streams collected within s, where red pixels represent positive events and blue pixels represent negative events; Figure (c) is based on the initial state. The current state is obtained by combining continuous event input with real-time updates. The fitted focal spot image is then plotted based on this. In the right figure, the red cross indicates the centroid position of the focal spot, and the red ellipse corresponds to the shape of the fitted ellipse. A quantitative comparison of Figure (a) and Figure (c) shows that the relative intensity error is only 0.84%, verifying that module S2 can achieve high-precision continuous updating of the focal spot state based solely on the initial state and real-time event stream.

[0076] Figure 8 The calibration process for the degrees of freedom of submodule S3 is shown. Figure (a) shows the initial intensity diagram. Figure (b) shows the degrees of freedom. The events generated when the perturbation is applied are shown, with red representing positive events and blue representing negative events. Figure (c) visualizes the centroid coordinates of each sub-aperture under this perturbation using blue arrows. The changes. The state corresponding to the arrow's starting point. The arrow indicating the centroid position is 100 times longer than the actual centroid position change, to clearly show the perturbation effect. Experimental results show that the submodule S3 of this invention can directly obtain the effect characteristics of each degree of freedom on each sub-aperture focal spot at the initial position by utilizing real-time focal spot state updates and degree-of-freedom perturbations, thereby providing a clear adjustment basis for the subsequent active assembly process and effectively avoiding the time-consuming traversal search in traditional methods.

[0077] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the invention. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the invention. Therefore, all technical solutions obtained through equivalent substitution or transformation fall within the protection scope of the present invention.

Claims

1. An optical active assembly system based on event wavefront sensing, characterized in that, The system includes: The optical system module to be assembled consists of a set of fixed mirrors and a set of adjustable mirrors, used to receive modulated incident light and obtain the outgoing wavefront. During the modulation stage, the positions of both the fixed lens group and the adjusting lens group remain unchanged. During the assembly stage, the position of the fixed lens group remains fixed, while the adjusting lens group is adjusted by the assembly mechanical mechanism module. The event wavefront sensor module consists of a microlens array and an event camera. It is used to receive the outgoing wavefront of the optical system module to be assembled. After the outgoing wavefront is decomposed by the microlens array, multiple focal spots corresponding to the sub-apertures are formed on the image plane of the event camera. During the modulation phase, the event wavefront sensor module outputs the moment when each pixel on the image plane triggers the first positive event. During the assembly phase, the event wavefront sensor module outputs the total number of positive events and the total number of negative events triggered by each pixel on the image plane. The optical active assembly decision terminal module is used to modulate the incident light during the modulation stage to obtain the modulated incident light, and to receive the moment when each pixel triggers the first positive event transmitted by the event wavefront sensor module, and finally obtain the initial state vector of the focal spot corresponding to each sub-aperture. During the assembly phase, the optical active assembly decision terminal module is used to receive the initial state vector of the focal spot corresponding to each sub-aperture, all the physical quantities of the degrees of freedom of the assembly mechanical mechanism module, and the total number of positive events and the total number of negative events triggered by each pixel transmitted by the event wavefront sensor module. Finally, it generates the degree of freedom action tensor, the current sub-aperture focal spot state vector and the degree of freedom adjustment amount at the next moment. It determines whether the current global focal spot state vector satisfies the ideal assembly state. If it does, the assembly terminates; otherwise, the assembly continues. The global focal spot state vector is composed of the focal spot state vectors of all sub-apertures. The assembly mechanical mechanism module is used to receive the next-moment degree of freedom adjustment amount transmitted by the optical active assembly decision terminal module, and to adjust the degree of freedom of the adjustment lens group according to the degree of freedom adjustment decision.

2. The optical active assembly system based on event wavefront sensing according to claim 1, characterized in that, During the modulation phase, the incident light intensity is monotonically increased by the optical active assembly decision terminal module; during the assembly phase, the incident light intensity remains unchanged at the maximum intensity during the modulation phase.

3. The optical active assembly system based on event wavefront sensing according to claim 1, characterized in that, The initial state vector of the focal spot corresponding to each sub-aperture is calculated using the following method: Based on the time-delay-intensity mapping principle, the moment when each pixel on the focal spot triggers its first positive event is converted into the pixel's initial relative intensity: ； in, For pixels The moment when the first positive event is triggered, where T is the total duration of the modulation phase. For pixels The initial relative intensity; The initial relative intensity of all focal spot pixels is calculated by iterating through them to obtain the initial focal spot intensity map of the outgoing wavefront; The focal spot state of each sub-aperture corresponding to the focal spot is parameterized by using an elliptical Gaussian model to obtain the focal spot state vector; Based on the focal spot state of each sub-aperture corresponding to the initial focal spot intensity map, substitute it into the focal spot state vector to obtain the initial state vector of the focal spot corresponding to each sub-aperture.

4. The optical active assembly system based on event wavefront sensing according to claim 3, characterized in that, The specific expression for the focal spot state vector is as follows: ； in, Let i be the focal spot state vector corresponding to the i-th sub-aperture. Let i be the coordinates of the centroid of the focal spot corresponding to the i-th sub-aperture. Let be the focal spot energy corresponding to the i-th sub-aperture. The focal spot corresponding to the i-th sub-aperture is in and Dimension of directional expansion The correlation coefficient of the elliptic Gaussian model is shown in the upper right corner. This indicates transpose.

5. The optical active assembly system based on event wavefront sensing according to claim 1, characterized in that, The current sub-aperture focal spot state vector is calculated using the following method: The brightness increment value of each pixel in the current event stream is calculated based on the total number of positive events and the total number of negative events triggered by each pixel in the current event stream transmitted by the event wavefront sensor module. A first-order Taylor expansion of the elliptical Gaussian model yields a linear relationship between the brightness increment of each pixel and the focal spot state increment corresponding to the sub-aperture. Substitute the brightness increment value of each pixel in the current event stream into the linear relationship to obtain the focal spot state increment corresponding to each sub-aperture in the current event stream; The current event flow sub-aperture focal spot state increment is added to the sub-aperture focal spot state vector of the previous event flow to calculate the current event flow sub-aperture focal spot state vector; this process continues until all event flows at the current time are traversed to obtain the final event flow sub-aperture focal spot state vector. Exponential smoothing is introduced to obtain the current focal spot state vector of the sub-aperture based on the previous moment's focal spot state vector and the final event flow's focal spot state vector. The calculation formula is as follows: ； in, For smoothing coefficients, The smoothed version of the current time t. Individual aperture focal spot state vectors Let be the smoothed state vector of the i-th sub-aperture focal spot at time t-1. Let be the state vector of the i-th sub-aperture focal spot under the final event flow.

6. The optical active assembly system based on event wavefront sensing according to claim 5, characterized in that, The time frame includes multiple event streams, each of which is transmitted sequentially to the optical active assembly decision terminal module according to its trigger time.

7. The optical active assembly system based on event wavefront sensing according to claim 1, characterized in that, The specific content of determining whether the global focal spot state vector at the current moment satisfies the ideal assembly state is as follows: Determine whether the current global focal spot state vector meets the first determination requirement. If it meets the first determination requirement, the assembly terminates. If it does not meet the first determination requirement, proceed to the next step of judgment. Determine whether the current global focal spot state increment meets the second determination requirement. If it does, update all sub-aperture focal spot state vectors in the current global focal spot state vector to the sub-aperture focal spot state vectors of the next moment and end the calculation. If it does not meet the requirement, proceed to the next determination step. If the action tensor of the degrees of freedom satisfies the third criterion, then the adjustment decision of the degrees of freedom for the next time step is generated, all sub-aperture focal spot state vectors in the current global focal spot state vector are updated to the sub-aperture focal spot state vectors of the next time step, and the calculation ends. If the condition is not met, then the action tensor of the degrees of freedom is modified, the adjustment amount of the degrees of freedom for the next time step is generated, all sub-aperture focal spot state vectors in the current global focal spot state vector are updated to the sub-aperture focal spot state vectors of the next time step, and the calculation ends.

8. The optical active assembly system based on event wavefront sensing according to claim 7, characterized in that, The first determination requirement is: at the current moment, the line connecting the position of the centroid of the focal spot corresponding to the i-th sub-aperture and the position of the center coordinates corresponding to the i-th sub-aperture is less than or equal to the first threshold; at the current moment, the absolute value of the correlation coefficient of the elliptic Gaussian model is less than or equal to the second threshold. The second determination requirement is: the inner product of the current global focal spot state increment and the preset ideal assembly direction is greater than 0; The third determination requirement is: ； in, For norm, This represents the current increment of the global focal spot state. For the action tensor of degrees of freedom, The adjustment amount for the degrees of freedom at the current moment. The preset threshold; The global focal spot state increment includes the focal spot state increment vectors corresponding to all sub-apertures.

9. The optical active assembly system based on event wavefront sensing according to claim 7, characterized in that, The specific calculation content of the action tensor of the degrees of freedom is as follows: All the physical quantities of the assembly mechanical mechanism module are parameterized into a single state vector of freedom. Apply degree-of-freedom adjustment amounts to all components of the degree-of-freedom state vector, and calculate the global focal spot state increment caused by each degree-of-freedom adjustment amount. Normalize the global focal spot state increment caused by the adjustment of each degree of freedom to obtain the action vector of each degree of freedom; The action vectors of all degrees of freedom are combined to obtain the action tensor of the degrees of freedom.

10. The optical active assembly system based on event wavefront sensing according to claim 1, characterized in that, The adjustment amount for generating the degrees of freedom at the next moment is specifically calculated as follows: using a progressive optimization strategy, it is obtained based on the global focal spot state vector and the action tensor of the degrees of freedom at the current moment. The calculation formula is as follows: ； in, Let be the adjustment amount of the degrees of freedom at time t+1. Step size factor Tensor for Degrees of Freedom The false rebellion.

Images