Wafer carrying speed control method and control system

Abstract

The application discloses a wafer carrying speed control method and a control system, and the control method comprises the following steps: obtaining three-dimensional data of a carrying path and decomposing the three-dimensional data into straight lines and curved sections to obtain geometric characteristics; calculating the curvatures of the curved sections and smoothly connecting the curved sections to determine continuous curvature distribution; generating a preliminary speed curve according to dynamic constraints of the upper limits of the speeds of each point according to the curvatures; calculating the speeds of each axis through inverse kinematics and iteratively optimizing the speeds of each axis to coordinate the speeds of each axis to generate a multi-axis collaborative speed sequence; fusing and calculating posture deviation based on real-time sensor data; and dynamically adjusting acceleration parameters according to the posture deviation to optimize and reduce the posture deviation to obtain final control instructions. Through dynamic planning of curvatures, multi-axis collaborative optimization and real-time posture self-adaptive adjustment, the control method systematically realizes synchronous improvement of carrying speed and stability.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of wafer conveying control, in particular to a wafer conveying speed control method and a control system. BACKGROUND

[0002] In the field of precision manufacturing, especially in the link of semiconductor wafer conveying, the high-speed and high-precision operation of automatic equipment is the key to improving the efficiency of the production line. With the increase of wafer size and the shortening of process cycle, the conveying system not only needs faster movement speed to reduce non-productive time, but also must ensure extremely high stability during movement to prevent breakage or contamination of brittle wafers due to vibration, sliding or attitude instability. However, the improvement of movement speed often accompanies problems such as increased vibration and enlarged dynamic error, which makes it difficult to reconcile the inherent contradiction between "high speed" and "high stability", becoming the most prominent challenge currently faced by the industry.

[0003] The existing wafer conveying technology has significant limitations in multi-axis motion control. Many traditional methods focus on the optimization of movement in a single direction, ignoring the synergy of multi-axis movement in three-dimensional space. This approach often cannot effectively balance the speed changes in each direction when dealing with complex paths. For example, when the wafer is rapidly turning or vertically lifting, there is a lack of coordinated control of multi-axis speed, resulting in sudden changes in speed at some key nodes of the equipment. This sudden change not only increases the unevenness of the force on the wafer, but also may cause additional wear and tear of the mechanical parts of the equipment, reducing the long-term stability of the system.

[0004] How to achieve coordinated control of multi-axis speed in complex three-dimensional paths, while optimizing speed adjustment according to trajectory curvature in the bending section, has become a key problem that needs to be solved in wafer conveying technology. Existing solutions have not been able to systematically optimize the movement path, multi-axis dynamics and real-time state feedback from the system level, resulting in the performance of the equipment always being unable to break through the technical bottleneck between speed and stability. SUMMARY

[0005] To this end, the technical problem to be solved by the present application is to overcome the defect that the wafer conveying efficiency and reliability are difficult to be considered simultaneously due to the inherent contradiction between high speed and high stability in the prior art, and to provide a wafer conveying speed control method, which can systematically realize the simultaneous improvement of conveying speed and running stability through dynamic speed planning based on path curvature, multi-axis speed collaborative iterative optimization and adaptive adjustment based on real-time attitude feedback.

[0006] To solve the above technical problems, the present application provides a wafer conveying speed control method, comprising the following steps:

[0007] Three-dimensional coordinate data of the object carrying path is acquired, and the three-dimensional coordinate data is segmented into straight line segments and curved segments through a preset path decomposition algorithm to obtain geometric feature descriptions of the segments;

[0008] According to the obtained geometric feature descriptions, the trajectory curvature values are calculated for the curved segments, the straight line segments and the curved segments are smoothly connected, and a continuous path curvature distribution is determined;

[0009] Based on the path curvature distribution, the upper limit of the speed of the path points is dynamically constrained according to the curvature values, and a preliminary speed planning curve is generated;

[0010] For the obtained preliminary speed planning curve, joint parameters of multi-axis motion are acquired, velocity vectors of each axis are calculated through an inverse kinematics algorithm, and iteration optimization is performed on the velocity vectors to make the difference between the velocity vectors of each axis within a coordination threshold, so that a coordinated multi-axis speed sequence is generated;

[0011] From the determined coordinated multi-axis speed sequence, an object inclination angle is acquired according to real-time sensor feedback for the posture stability requirement, position and angle data are fused through Kalman filtering to obtain a posture deviation estimation value;

[0012] Based on the posture deviation estimation value, the acceleration parameters in the speed sequence are dynamically adjusted, and an optimization algorithm is used to reduce the posture deviation, so as to determine an optimized speed control instruction.

[0013] In an embodiment of the present application, three-dimensional coordinate data of the object carrying path is acquired, and the three-dimensional coordinate data is segmented into straight line segments and curved segments through a preset path decomposition algorithm to obtain geometric feature descriptions of the segments, including:

[0014] Three-dimensional coordinate data of the object carrying path is acquired, and spatial trajectory points are collected from sensors or positioning systems to obtain an original three-dimensional coordinate sequence;

[0015] The three-dimensional coordinate sequence is processed through a preset path segmentation algorithm to obtain a straight line segment coordinate set and a curve segment coordinate set;

[0016] For the straight line segment coordinate set, the length and direction vector of each segment are calculated to obtain the geometric feature description of the straight line segment;

[0017] According to the curve segment coordinate set, a spline interpolation algorithm is used to fit a curve trajectory, the curvature and tangent vector are calculated, and the geometric feature description of the curve segment is obtained;

[0018] The motion direction change is extracted from the geometric feature descriptions of the straight line segments and the curve segments, the angle and distance between the path points are analyzed, and a dynamic feature sequence of the path is obtained;

[0019] By means of the dynamic characteristic sequence, the path segments are classified by using a K-means clustering algorithm, the spatial trajectory characteristics of each segment are judged, and a classified path segment set is obtained;

[0020] According to the classified path segment set, the geometric characteristics of the straight line segments and the curved line segments are integrated, and a complete path geometric characteristic description is generated.

[0021] In an embodiment of the present application, according to the obtained geometric characteristic description, the trajectory curvature value is calculated for the curved segment, the straight line segment and the curved segment are smoothly connected, and a continuous path curvature distribution is determined, including:

[0022] Based on the geometric characteristic description of the curved segment, a curvature continuity constraint algorithm is used to optimize the arrangement of the control points, and a fitting curve with smooth curvature transition is generated;

[0023] Based on the geometric characteristic description of the straight line segment, the curvature change rate at the connecting point of the straight line to the curved line is calculated, based on the curvature change rate, a Bezier curve splicing algorithm is used to generate a smooth transition segment at the connecting point of the straight line segment and the curved segment;

[0024] According to the fitting curve with smooth curvature transition and the smooth transition segment, the curvature value of the continuous sampling points on the path is calculated, and a discrete curvature distribution sequence is generated;

[0025] The discrete curvature distribution sequence is smoothed by using a Gaussian filtering algorithm to eliminate local curvature fluctuations, and a continuous path curvature distribution is obtained.

[0026] In an embodiment of the present application, the smoothing of the discrete curvature distribution sequence by using the Gaussian filtering algorithm includes:

[0027] The Gaussian filtering window size and the standard deviation parameters are set, and the filtering strength is adaptively adjusted based on the total length of the path and the curvature change frequency;

[0028] The discrete curvature sequence is filtered by using convolution operation to eliminate local curvature fluctuations while retaining global curvature change characteristics;

[0029] The filtered curvature distribution is subjected to continuity test to ensure the continuity of the first and second derivatives of the curvature distribution curve.

[0030] In an embodiment of the present application, based on the path curvature distribution, the upper limit of the speed of the path point is dynamically constrained according to the curvature value, and a preliminary speed planning curve is generated, including:

[0031] The curvature value of each path point is obtained from the path curvature distribution, the curvature size is judged by using a preset threshold value, if the curvature value is greater than the preset threshold value, the upper limit set of the speed is determined according to the mapping relationship between the curvature value and the upper limit of the speed;

[0032] A speed upper limit curve is fitted by using a spline interpolation method based on the speed upper limit set and the coordinate sequence of the path point set, to obtain a preliminary speed planning curve.

[0033] In an embodiment of the present application, for the obtained preliminary speed planning curve, joint parameters of multi-axis motion are acquired, velocity vectors of each axis are calculated by inverse kinematics algorithm, and iterative optimization is performed thereon to make the difference between the velocity vectors of each axis within a coordination threshold, so as to generate a coordinated multi-axis speed sequence, including:

[0034] The preliminary speed planning curve is acquired, joint parameters of multi-axis motion are extracted therefrom, and initial motion states of each axis are determined;

[0035] By inverse kinematics algorithm, joint velocity vectors of each axis are calculated based on the extracted joint parameters, to obtain an initial velocity vector set;

[0036] Based on the initial velocity vector set, a particle swarm optimization algorithm is used to iteratively adjust the joint velocity vectors of each axis, with the aim of minimizing the difference between the velocity vectors of each axis, so as to obtain an optimized velocity vector set;

[0037] A coordinated multi-axis speed sequence is generated by the optimized velocity vector set, and the motion trajectory of each axis is determined.

[0038] In an embodiment of the present application, from the determined coordinated multi-axis speed sequence, an object inclination angle is acquired based on real-time sensor feedback for posture stability requirements, and a posture deviation estimation value is obtained by Kalman filtering fusion of position and angle data, including:

[0039] From the multi-axis speed sequence, a coordinated control instruction is acquired, each axis speed is synchronously processed by a preset motion planning algorithm, and a unified motion instruction set is obtained;

[0040] According to the unified motion instruction set, in combination with sensor feedback data, an accelerometer and a gyroscope are used to collect the object inclination angle in real time, to obtain original posture data;

[0041] By Kalman filtering fusion of position data and angle data in the original posture data, state estimation is iteratively updated, to obtain an optimized posture deviation estimation value.

[0042] In an embodiment of the present application, based on the posture deviation estimation value, the acceleration parameters in the speed sequence are dynamically adjusted, and an optimization algorithm is used to reduce the posture deviation, so as to determine an optimized speed control instruction, including:

[0043] Obtain an attitude deviation estimation value and a speed sequence from sensor data, classify the attitude deviation estimation value using a preset threshold, and if the attitude deviation estimation value is greater than the threshold, smooth the deviation data by using a Kalman filtering algorithm for smoothing processing.

[0044] According to the deviation data, extract an acceleration parameter in the speed sequence, determine whether the acceleration parameter exceeds a preset range using logical judgment, and if it does, adjust the acceleration parameter by using a gradient descent algorithm to obtain an optimized acceleration parameter.

[0045] Using the acceleration parameter, update the speed sequence by using a dynamic adjustment mechanism in combination with real-time feedback sensor data to generate an adjusted speed sequence.

[0046] According to the speed sequence, generate a control instruction using logical judgment to obtain an optimized speed control instruction.

[0047] In an embodiment of the present application, it further comprises: based on the target carrying time, globally optimizing the speed control instruction, adjusting the speed parameters of multiple path segments in parallel to meet the time constraint, thereby determining the final executed speed control instruction.

[0048] The above technical solution of the present application has the following advantages compared with the prior art:

[0049] The wafer carrying speed control method of the present application constructs a closed-loop optimization system from global path planning to local parameter adaptive adjustment, which can systematically realize the synchronous improvement of carrying speed and running smoothness through dynamic speed planning based on path curvature, multi-axis speed collaborative iterative optimization and adaptive adjustment based on real-time attitude feedback.

[0050] The method first plans from the motion source, generates a speed baseline strictly matched with the geometric shape by decomposing the path and calculating the continuous curvature distribution, ensures automatic speed reduction in high-curvature sections, and avoids the risk of instability in turning from the source. Furthermore, the method of the present application deeply understands the collaborative nature of multi-axis mechanical systems, converts the smooth trajectory at the end into highly matched speed instructions between each driving shaft through inverse kinematics calculation and iterative optimization, eliminates the internal torsional vibration caused by the incoordination between shafts, and provides a solid dynamic foundation for high-speed operation.

[0051] The method introduces an adaptive control layer based on real-time sensor feedback, fuses multi-source data through a Kalman filter, and the system can accurately estimate the dynamic attitude deviation that cannot be predicted by the model; based on this real-time estimate, instead of relying on fixed control parameters, the acceleration command is dynamically adjusted to actively offset the attitude error. This means that the control process has the ability to intelligently respond to uncertain disturbances, so it can maintain extreme stability under complex working conditions. The entire technical solution is progressive, from offline path and speed planning to online multi-axis coordination and adaptive adjustment, forming an organic whole that balances efficiency and stability.

[0052] Therefore, the beneficial effects brought by the method are as follows: the dynamic speed planning based on curvature lays the foundation for smooth motion, avoiding speed impact at sharp bends; the coordinated optimization of multi-axis speed eliminates the vibration inducement from the mechanical root; and the real-time attitude feedback and adaptive adjustment give the system strong robustness against disturbances, ensuring accurate control at high speed. These beneficial effects and technical features complement each other, enabling significant improvement in carrying speed and shortening of operation cycle while not only maintaining stability but also achieving higher dynamic precision and reliability, ultimately achieving the dual goals of improving production efficiency and ensuring product yield. BRIEF DESCRIPTION OF DRAWINGS

[0053] In order to make the content of the present application more easily understood, the following further describes the present application in detail according to specific embodiments of the present application and in conjunction with the accompanying drawings, in which:

[0054] Figure 1 is a step flowchart of the wafer carrying speed control method of the present application;

[0055] Figure 2 is a step flowchart of path acquisition and decomposition of the present application;

[0056] Figure 3 is a step flowchart of curvature calculation and smoothing of the present application;

[0057] Figure 4 is a step flowchart of speed preliminary planning of the present application;

[0058] Figure 5 is a step flowchart of multi-axis coordination optimization of the present application;

[0059] Figure 6 is a step flowchart of attitude monitoring feedback of the present application;

[0060] Figure 7 is a step flowchart of control instruction generation of the present application;

[0061] Figure 8is a structural framework diagram of a wafer conveying speed control system of the present application. DETAILED DESCRIPTION

[0062] The present application will be further described below with reference to the drawings and specific embodiments, so that those skilled in the art can better understand the present application and implement it, but the embodiments are not limiting to the present application.

[0063] Referring to Figure 1 The present application proposes a wafer conveying speed control method, including the following steps:

[0064] Obtain the three-dimensional coordinate data of the object conveying path, segment it into straight line segments and curved segments through a preset path decomposition algorithm, and obtain the geometric feature description of each segment; according to the obtained geometric feature description, calculate the trajectory curvature value for the curved segment, smoothly connect the straight line segment and the curved segment, and determine the continuous path curvature distribution; based on the path curvature distribution, dynamically constrain the speed upper limit of the path point according to the curvature value, and generate a preliminary speed planning curve; for the obtained preliminary speed planning curve, obtain the joint parameters of multi-axis motion, calculate the speed vector of each axis through inverse kinematics algorithm, and iteratively optimize it to make the difference between the speed vectors of each axis within the coordination threshold, thereby generating a coordinated multi-axis speed sequence; from the determined coordinated multi-axis speed sequence, obtain the object inclination angle fed back by the real-time sensor according to the posture stability requirement, fuse the position and angle data through Kalman filtering to obtain the posture deviation estimation value; based on the posture deviation estimation value, dynamically adjust the acceleration parameter in the speed sequence, and use the optimization algorithm to reduce the posture deviation, thereby determining the optimized speed control instruction.

[0065] Among them, the path decomposition algorithm refers to the calculation method of segmenting the three-dimensional trajectory into straight line segments and curved segments, which can be specifically implemented by using a clustering analysis method based on curvature change, and is used to extract the path geometric features. The path curvature distribution refers to a continuous data sequence describing the degree of curvature of the trajectory, which can be specifically generated by using spline interpolation and Gaussian filtering method, and is used to constrain the speed upper limit. The inverse kinematics algorithm refers to a mathematical method of mapping the end effector speed to the joint space, which can be specifically implemented by using the inverse of Jacobian matrix, and is used to calculate the speed vector of each axis. The coordination threshold refers to the allowable range of multi-axis speed difference, which can be specifically set according to the dynamic characteristics of the mechanical system, and is used to ensure the motion coordination. The posture deviation estimation value refers to the optimization calculation result of the wafer inclination angle, which can be specifically obtained by fusing multi-sensor data through Kalman filtering, and is used for feedback control adjustment.

[0066] Specifically, the method first analyzes the carrying path geometrically, decomposes the complex trajectory into straight line segments and curved segments, and establishes a structured path feature model. The curvature continuity constraint algorithm is used to process the path connection to generate a smooth curvature distribution curve, providing dynamic constraint basis for speed planning. When generating the preliminary speed curve, the speed upper limit is dynamically adjusted according to the curvature value of each point to avoid wafer sliding caused by centrifugal force exceeding limit. Then the path speed is converted into multi-axis joint speed, and an iterative optimization algorithm is used to eliminate the difference between speed vectors and ensure the synchronization of each axis movement. In the motion execution stage, real-time posture data is collected and measurement noise is eliminated through filtering algorithm to accurately estimate the wafer tilt state. Finally, the acceleration parameters are dynamically adjusted according to the posture deviation to form a closed-loop adjustment mechanism of speed control and posture stability.

[0067] Compared with the prior art, the traditional method uses a fixed speed curve when processing complex paths, which cannot adapt to the difference in dynamic constraints caused by curvature changes. Existing multi-axis coordinated control is mainly based on open-loop calculation, lacking active optimization of speed vector differences. The present scheme realizes the dynamic matching of speed planning and path features through the combination of path decomposition and curvature analysis; uses a multi-axis optimization algorithm with a coordination threshold to effectively suppress joint speed mutations; and introduces a real-time posture feedback adjustment mechanism to overcome the limitations of traditional methods relying on preset parameters.

[0068] Through the above technical scheme, the present application realizes the coordinated control of multi-axis speed in three-dimensional path, automatically adjusts the motion parameters according to the curvature change in the curved segment, and dynamically corrects the acceleration command through real-time feedback. This method ensures the carrying efficiency while significantly reducing the risk of wafer stress mutation and posture instability, solving the contradiction between high-speed motion and stability.

[0069] Reference Figure 2As shown, the application further proposes a specific implementation method for obtaining three-dimensional coordinate data of an object carrying path, segmenting it into straight line segments and curved segments through a preset path decomposition algorithm, and obtaining geometric feature description of each segment, including: obtaining three-dimensional coordinate data of the object carrying path, collecting spatial trajectory points from sensors or positioning systems to obtain original three-dimensional coordinate sequences; processing the three-dimensional coordinate sequences through a preset path segmentation algorithm to obtain a straight line segment coordinate set and a curve segment coordinate set; for the straight line segment coordinate set, calculating the length and direction vector of each segment to obtain the geometric feature description of the straight line segment; according to the curve segment coordinate set, fitting a curve trajectory using a spline interpolation algorithm, calculating the curvature and tangent vector to obtain the geometric feature description of the curve segment; extracting the motion direction change from the geometric feature description of the straight line segment and the curve segment, analyzing the angle and distance between the path points to obtain a dynamic feature sequence of the path; through the dynamic feature sequence, using a K-means clustering algorithm to classify the path segments, judging the spatial trajectory characteristics of each segment to obtain a classified path segment set; and integrating the geometric features of the straight line segment and the curve segment according to the classified path segment set to generate a complete path geometric feature description.

[0070] Among them, the path segmentation algorithm refers to a calculation method for automatically dividing the three-dimensional coordinate sequence into straight line segments and curve segments, which can be specifically implemented by using a curvature threshold or angle change judgment rule to eliminate subjective errors of manual division. The spline interpolation algorithm refers to a mathematical method for generating a continuous curve by fitting discrete trajectory points with a polynomial function, which can be specifically implemented by using cubic spline interpolation or B-spline interpolation to accurately describe the geometric characteristics of the curved path. The K-means clustering algorithm refers to an unsupervised machine learning method for automatically classifying data features, which can be specifically implemented by using Euclidean distance to measure the dynamic feature difference of the trajectory segment to identify path segments with different spatial trajectory characteristics.

[0071] Specifically, the original three-dimensional coordinate sequence is collected by a high-precision sensor to provide basic data for path decomposition. The path segmentation algorithm automatically processes the coordinate sequence, classifies the point set that meets the straight line judgment condition as a straight line segment, and classifies the remaining as a curve segment. For the straight line segment set, the length and direction of each segment are calculated through vector operation to form quantitative parameters; for the curve segment set, a continuous trajectory is generated using spline interpolation, and the curvature and tangent vector are calculated. The dynamic feature sequence is constructed by analyzing the direction change of adjacent path points, reflecting the turning characteristics of the motion trajectory. The K-means clustering algorithm divides the path segments into different categories according to the dynamic features, automatically identifying the distribution rule of straight line and curve segments. Finally, the geometric feature parameters of each segment are integrated to form a complete path description model, providing structured input for subsequent speed control.

[0072] Compared with the prior art, the traditional method relies on artificial experience to divide the path segment, which is easy to introduce subjective error and cause sudden change of speed. The scheme realizes accurate identification of path segments through automatic path segmentation and feature extraction combined with machine learning classification algorithm. The existing technology mainly uses simple circular arc approximation for geometric description of curve segment, while the scheme uses spline interpolation for accurate fitting, which significantly improves the accuracy of curvature calculation of curved segment. In addition, the construction and clustering analysis of dynamic feature sequence solve the defect of ignoring the motion direction change characteristics in the traditional method.

[0073] Through the above technical scheme, the application realizes fine decomposition and geometric feature extraction of three-dimensional path, effectively avoids the problem of sudden change of motion speed caused by inaccurate path segmentation. Through the automatic identification mechanism of straight line segment and curve segment, the objectivity and accuracy of path classification are improved. Accurate curvature and tangent vector calculation provides a reliable geometric parameter basis for subsequent speed optimization, and dynamic feature clustering analysis ensures the rationality of path segment classification, thereby supporting the coordinated optimization of multi-axis motion control system.

[0074] Referring to Figure 3 As shown in the figure, the application further proposes a method for realizing smooth connection of straight line segment and curved segment without sudden change after path decomposition, and eliminating local fluctuation of curvature distribution, including the following steps: based on the geometric feature description of curve segment, using curvature continuity constraint algorithm to optimize control point arrangement, generating curvature smooth transition fitting curve; based on the geometric feature description of straight line segment, calculating the curvature change rate at the connection point of straight line to curve, based on the curvature change rate, using Bezier curve splicing algorithm to generate smooth transition segment at the connection of straight line segment and curve segment; according to the curvature smooth transition fitting curve and the smooth transition segment, calculating the curvature value of the continuous sampling points on the path, generating discrete curvature distribution sequence; using Gaussian filtering algorithm to smooth the discrete curvature distribution sequence, eliminating local curvature fluctuation, and obtaining continuous path curvature distribution.

[0075] The curvature continuity constraint algorithm refers to ensuring the continuity of the curvature value and its derivative of adjacent curve segments at the connection point through mathematical constraint conditions, which can be specifically implemented by solving the control point coordinates by using cubic spline interpolation combined with the curvature differential equation to eliminate the curvature mutation of the fitted curve at the segmented connection. The Bezier curve splicing algorithm refers to using the end point tangent vector control characteristics of the Bezier curve to insert a transition curve between the straight line segment and the curve segment, which can be specifically implemented by adjusting the control point weight to match the curvature change rate of the transition segment with the two side path segments, and is used to realize the smooth transition of the geometric shape and the motion parameter. The discrete curvature distribution sequence refers to a set of curvature values obtained by sampling at fixed intervals along the path, which can be specifically calculated by using the numerical differentiation method to calculate the curvature of each sampling point, and is used to reflect the local geometric characteristics of the path. The Gaussian filtering algorithm refers to using a Gaussian function as the weight to perform weighted average on the discrete curvature sequence, which can be specifically implemented by adaptively adjusting the filtering window width and the standard deviation parameter, and is used to suppress high-frequency noise and retain the global trend characteristics of the curvature distribution.

[0076] Specifically, first, the control points of the curve segment are arranged by the curvature continuity constraint algorithm, so that the adjacent curve segments not only satisfy the position continuity at the connection point, but also satisfy the first derivative continuity of the curvature, thereby eliminating the common curvature jump phenomenon in the traditional interpolation method. Further, in the junction area of the straight line segment and the curve segment, the curvature change rate at the connection point is calculated, the position and weight of the Bezier curve control point are dynamically adjusted, a transition segment with continuous change of curvature derivative is generated, and the acceleration mutation caused by the discontinuity of the path geometry is avoided. Subsequently, the optimized complete path is sampled at a fixed step, the curvature value of each sampling point is calculated by using the numerical differentiation method, and a discrete sequence containing all local characteristics of the path is formed. Finally, the Gaussian filtering parameters are adaptively selected according to the total length of the path and the curvature change frequency, the discrete curvature sequence is convolved, the local fluctuations are eliminated while the macroscopic continuity of the curvature distribution is maintained, and the filtered curvature curve satisfies the dynamic constraint condition through the derivative continuity test.

[0077] Compared with the prior art, the traditional path smoothing method usually only focuses on the position continuity and ignores the curvature continuity. For example, when the linear interpolation is used to connect the straight line segment and the curve segment, although the path geometry is continuous, the curvature mutation at the connection point will cause the vibration of the handling equipment. The present scheme realizes the double continuity of the curvature and its derivative at the geometric connection through the curvature continuity constraint algorithm and the Bezier curve splicing algorithm, and fundamentally eliminates the mutation of the motion parameter. In addition, the existing curvature filtering method usually uses the mean filtering with fixed parameters, which is easy to cause the curvature distribution to be excessively smoothed, while the adaptive Gaussian filtering is used in the present scheme, the filtering strength is dynamically adjusted according to the path characteristics, the local noise is effectively suppressed, and the global characteristics of the curvature change are completely retained.

[0078] By the technical solution, the curvature mutation problem at the connection between the straight line segment and the curved segment after path decomposition is effectively solved, the smooth transition segment with continuously changing curvature is generated, and the speed mutation of the handling equipment at the path turning point is avoided. Meanwhile, the adaptive Gaussian filtering is used to process the discrete curvature sequence, the local fluctuation interference is eliminated, the continuous curvature distribution conforming to the dynamic constraint is formed, a stable and reliable curvature constraint basis is provided for the subsequent speed planning module, thereby the vibration risk in the wafer handling process is reduced, and the motion stability is improved.

[0079] In the embodiment, the Gaussian filtering algorithm is used to perform the smoothing processing on the discrete curvature distribution sequence, including setting the Gaussian filtering window size and the standard deviation parameter, adaptively adjusting the filtering strength based on the total path length and the curvature change frequency; the convolution operation is used to perform the filtering processing on the discrete curvature sequence, eliminating the local curvature fluctuation while retaining the global curvature change characteristics; the continuity of the filtered curvature distribution is verified, to ensure the first and second derivatives of the curvature distribution curve to be continuous.

[0080] The Gaussian filtering window size refers to the neighborhood range used for calculating the weighted average of each curvature point, which can be specifically implemented in a sliding window manner, and the window size is dynamically adjusted according to the total path length in proportion, for example, when the total path length increases, the window size is correspondingly expanded, to adapt to the curvature change characteristics under different scales. The standard deviation parameter is used to control the weight distribution form of the Gaussian kernel function, which can be specifically adaptively matched according to the curvature change frequency, and when the curvature fluctuates at a high frequency, the standard deviation is increased to enhance the smoothing effect, and when the curvature fluctuates at a low frequency, the standard deviation is reduced to retain the detail characteristics. The convolution operation refers to the mathematical convolution operation of the Gaussian kernel function and the discrete curvature sequence, which can be specifically accelerated by the discrete Fourier transform, to realize the balance between local noise suppression and overall trend retention. The continuity verification refers to the differential calculation of the filtered curvature distribution, which can be specifically implemented by using the numerical differential algorithm to verify the first derivative continuity and the second derivative non-mutation, to ensure that the curvature curve meets the kinematic constraint condition.

[0081] Specifically, in the region with long total path length and gentle curvature change, the filtering window size is expanded and the standard deviation parameter is reduced, to eliminate the slight fluctuation while retaining the overall curvature trend; in the region with dense path curvature, the window size is reduced and the standard deviation parameter is increased, to effectively suppress the high-frequency noise interference. In the convolution operation process, the symmetry weight distribution of the Gaussian kernel function makes the mutation points in the curvature sequence be weighted and averaged by the surrounding data, and the global curvature change amplitude is retained. After the filtering is completed, the slope continuity of the first derivative of the curvature distribution is verified, and the second derivative is checked for whether there is a step point, if the discontinuous region is detected, the filtering parameters are re-adjusted for iterative optimization, until the curvature smoothness requirement of the motion planning is met.

[0082] Compared with the prior art, the traditional method usually adopts mean filtering or median filtering with fixed parameters to process the curvature data, which cannot dynamically adjust the filtering strength according to the path geometry characteristics, resulting in excessive smoothing to lose key curvature characteristics or residual local noise. The scheme drives the parameter adaptive mechanism through the total length of the path and the curvature change frequency, so that the filtering strength is accurately matched with the path morphology. Combined with the mathematical properties of convolution operation, the global curvature distribution is maintained while the local fluctuations are eliminated. Further, through the differential continuity test, the physical realizability of the curvature curve is verified from the kinematics level, solving the problem of acceleration mutation caused by ignoring the derivative continuity in the traditional method.

[0083] Through the above technical scheme, the local fluctuations in the path curvature distribution can be effectively eliminated, ensuring the smooth transition of curvature change in the spatial and temporal domains, avoiding the sharp fluctuations of the speed command in the wafer carrying process caused by curvature mutation, thereby improving the stability of multi-axis cooperative motion and reducing the risk of inertial impact of the wafer caused by acceleration mutation.

[0084] Referring to Figure 4 Based on the path curvature distribution, the speed upper limit of the path point is dynamically constrained according to the curvature value, and a preliminary speed planning curve is generated, including obtaining the curvature value of each path point from the path curvature distribution, judging the curvature size by using a preset threshold value, if the curvature value is greater than the preset threshold value, determining a speed upper limit set according to the mapping relationship between the curvature value and the speed upper limit, fitting the speed upper limit curve by using a spline interpolation method through the speed upper limit set combined with the coordinate sequence of the path point set, and obtaining the preliminary speed planning curve.

[0085] The path curvature distribution refers to a sequence of curvature values of each point on the path, which is obtained by smoothing the discrete curvature distribution sequence through a Gaussian filtering algorithm, and is used to reflect the change trend of the path bending degree. The mapping relationship between the curvature value and the speed upper limit refers to setting the corresponding speed upper limit value according to the size of the curvature value, the greater the curvature, the lower the corresponding speed upper limit, and the relationship can be established through experimental data or a dynamics model. The spline interpolation method refers to using a cubic spline function to fit the discrete speed upper limit value to generate a continuous and derivable speed planning curve.

[0086] Specifically, the path curvature distribution processed by the Gaussian filter can accurately reflect the bending degree of each point of the path. When the curvature value exceeds the preset threshold, it indicates that there is a large centrifugal force risk in this area. At this time, the speed upper limit is reduced according to the curvature-speed mapping relationship to avoid displacement of the wafer due to excessive lateral acceleration. For the path segment that does not exceed the threshold, a higher carrying speed is allowed. The threshold is based on the system dynamics performance to ensure the wafer safety constraints and motion stability requirements. The engineering parameters are determined by comprehensive theoretical calculation, simulation and experiment. The discrete speed upper limit is converted into a continuous curve through spline interpolation, ensuring smooth transition of speed change between adjacent path points and eliminating mechanical vibration caused by discrete point mutation in traditional segmented speed control.

[0087] Compared with the prior art, the traditional method uses a fixed speed threshold or a segmented linear adjustment method, which cannot dynamically constrain the speed according to the continuous change of the path curvature, resulting in insufficient speed reduction in high-curvature areas or speed waste in low-curvature areas. The scheme dynamically maps the curvature distribution and the speed upper limit, and combines the mathematical properties of spline interpolation to realize the second-order continuous derivability of the speed curve on the global path, overcoming the problem of speed mutation.

[0088] Through the above technical scheme, the application realizes the dynamic matching of the speed and the path curvature in the wafer carrying process, automatically reduces the speed in the curved area to suppress the influence of centrifugal force, maintains efficient transportation in the straight line area, and eliminates acceleration mutation through a continuous speed curve planning to ensure the stability of the wafer attitude.

[0089] Referring to Figure 5 Further, the application further obtains joint parameters of multi-axis motion based on the obtained preliminary speed planning curve, calculates speed vectors of each axis through an inverse kinematics algorithm, and iteratively optimizes the speed vectors to make the difference between the speed vectors of each axis within a coordination threshold, thereby generating a coordinated multi-axis speed sequence, including: obtaining a preliminary speed planning curve, extracting joint parameters of multi-axis motion from the preliminary speed planning curve, and determining initial motion states of each axis; calculating joint speed vectors of each axis based on the extracted joint parameters through an inverse kinematics algorithm to obtain an initial speed vector set; iteratively adjusting the joint speed vectors of each axis based on the initial speed vector set by using a particle swarm optimization algorithm to minimize the difference between the speed vectors of each axis, thereby obtaining an optimized speed vector set; and generating a coordinated multi-axis speed sequence based on the optimized speed vector set to determine motion trajectories of each axis.

[0090] The inverse kinematics algorithm refers to a mathematical method for converting motion parameters of an end effector into motion parameters of each joint axis, and can be specifically implemented by using inverse of a Jacobian matrix or a numerical iteration method, and is used to establish a kinematic correlation between path planning and multi-axis motion. The particle swarm optimization algorithm refers to a global optimization method based on swarm intelligence, and can be specifically implemented by using an inertia weight adjustment strategy or a neighborhood topology, and is used to find an optimal solution with minimum difference in a multi-axis velocity vector space. The coordination threshold refers to a maximum difference range allowed between multi-axis velocity vectors, and can be specifically determined by experiment calibration or a dynamic adjustment strategy, and is used to constrain a termination condition of the optimization process.

[0091] Specifically, first, multi-axis joint parameters are extracted from the preliminary velocity planning curve, such as displacement ranges and acceleration limits of each axis, to determine initial motion states of each axis. Then, the velocity planning of the path point is converted into velocity vectors of each joint axis by using the inverse kinematics algorithm, such as calculating velocity components of each axis by using a pseudo-inverse of the Jacobian matrix, to form an initial velocity vector set. Further, the initial velocity vectors are iteratively adjusted by using the particle swarm optimization algorithm, such as setting a fitness function as a standard deviation of the velocity vectors of each axis, and gradually reducing the velocity difference by using a group search mechanism. When the velocity vector difference of all axes reaches the coordination threshold, the optimization process is terminated, and the finally generated multi-axis velocity sequence can ensure synchronization of motion trajectories of each axis in time and space.

[0092] Compared with the prior art, the traditional method usually allocates multi-axis velocities by using a fixed ratio, and cannot dynamically adapt to motion coupling relationships under complex paths. The present scheme accurately establishes a mapping relationship between a path and joint motion by using inverse kinematics, and globally adjusts velocity parameters by using a group optimization algorithm, effectively solving the problem of uneven stress on a wafer caused by sudden changes in multi-axis velocities. For example, in the prior art, velocity mismatch between axes easily occurs when a straight path is switched to a curved segment, while in the present scheme, the velocity difference between axes is always within the coordination threshold by real-time optimization.

[0093] By using the above technical scheme, the present application realizes dynamic coordination control of multi-axis motion velocities, and eliminates uneven stress on a wafer caused by sudden changes in velocities. In the wafer carrying process, motion trajectories of each joint axis are kept synchronized by using the optimized velocity sequence, avoiding mechanical impact caused by excessively high single-axis velocity. At the same time, the application of the particle swarm optimization algorithm makes the multi-axis velocity adjustment process have a global optimal characteristic, effectively reducing dynamic errors of the equipment under complex paths.

[0094] Reference Figure 6As shown, the application further proposes a method for obtaining real-time sensor feedback object tilt angle for attitude stability requirements from the determined coordinated multi-axis speed sequence, fusing position and angle data through Kalman filtering to obtain attitude deviation estimation value, including: obtaining coordinated control instructions from the multi-axis speed sequence, synchronously processing the speed of each axis through a preset motion planning algorithm to obtain a unified motion instruction set; according to the unified motion instruction set, combining sensor feedback data, using an accelerometer and a gyroscope to collect the object tilt angle in real time to obtain the original attitude data; fusing the position data and angle data in the original attitude data through Kalman filtering, iteratively updating the state estimation to obtain the optimized attitude deviation estimation value.

[0095] Wherein, the coordinated control instruction refers to the speed parameter set of each actuator in the multi-axis motion system, which can be specifically implemented by using a timestamp-based instruction synchronization protocol to ensure the alignment of multi-axis motion instructions in the time dimension. The motion planning algorithm refers to an optimization method for eliminating multi-axis speed conflicts, which can specifically use a dynamic time warping algorithm to perform time sequence matching on the speed sequence of each axis to solve the motion asynchronization problem caused by the speed difference between axes. The accelerometer and the gyroscope refer to a sensor combination for detecting the spatial attitude of an object, which can be specifically implemented by using a MEMS inertial measurement unit to capture the dynamic tilt state of the wafer through joint measurement of three-axis acceleration and angular velocity. The Kalman filter refers to a state estimation algorithm for multi-source data fusion, which can specifically use a linear Kalman filter to perform weighted fusion of position coordinates and angle data, and eliminate sensor noise interference through the iterative process of prediction and correction.

[0096] Specifically, after the multi-axis speed sequence is generated, the motion parameters of each axis are extracted through the instruction synchronization protocol, and the time phase of the speed of each axis is adjusted based on the dynamic time warping algorithm to form a spatiotemporally consistent motion instruction set. During the execution of the instructions, the acceleration and angular velocity data of the wafer are collected in real time using the MEMS inertial measurement unit, and the target position information in the motion instructions is combined to construct an original attitude data set containing position error and angle deviation. Subsequently, the position coordinates and angle data are input into the Kalman filter, the attitude parameters at the next time are predicted through the state transition equation, and the predicted value is compared with the measured value of the sensor through the observation equation to minimize the covariance matrix as the target to update the state estimation value, and finally the optimized attitude deviation quantity is output. This process continuously corrects dynamic errors through a closed-loop feedback mechanism to ensure the real-time and accuracy of attitude estimation.

[0097] Compared with the prior art, the traditional method usually relies on a single sensor for attitude detection, such as indirectly calculating the wafer tilt angle through encoder feedback position data only, resulting in error accumulation and response lag. The multi-axis control strategy adopted in the prior art often ignores the timing alignment of the motion instructions, causing superposition of inter-axis speed fluctuations and exacerbating attitude instability. The present scheme estimates through the synchronization processing of motion instructions and the fusion of multi-source data, eliminates inter-axis motion conflicts, and effectively suppresses the influence of sensor noise on attitude detection by using the dynamic correction capability of Kalman filtering, achieving high-precision real-time deviation estimation.

[0098] Through the above technical solutions, the present application can solve the problem of attitude deviation caused by multi-axis speed asynchrony during wafer handling. Through motion instruction synchronization and multi-source data fusion, the vibration amplitude of the wafer during high-speed motion is significantly reduced, and the stability of the handling process is improved. At the same time, through the iterative optimization mechanism of Kalman filtering, the interference of sensor noise on attitude estimation is reduced, reliable deviation data is provided for subsequent dynamic adjustment, and the risk of misadjustment caused by detection error is avoided.

[0099] Referring to Figure 7 The present application further proposes a method for dynamically adjusting the acceleration parameter in the speed sequence based on the attitude deviation estimation value and reducing the attitude deviation by using an optimization algorithm, thereby determining the optimized speed control instruction, specifically including: obtaining the attitude deviation estimation value and the speed sequence from the sensor data, classifying the attitude deviation estimation value using a preset threshold, and if the attitude deviation estimation value is greater than the threshold, smoothing the deviation data by Kalman filtering algorithm; extracting the acceleration parameter in the speed sequence according to the deviation data, determining whether the acceleration parameter exceeds the preset range by logical judgment, and if it does, adjusting the acceleration parameter by gradient descent algorithm to obtain the optimized acceleration parameter; updating the speed sequence by using the dynamic adjustment mechanism to generate the adjusted speed sequence through the acceleration parameter combined with the real-time feedback of the sensor data; after generating the control instruction according to the speed sequence, performing global speed optimization on the speed control instruction based on the target handling time, and adjusting the speed parameters of multiple path segments in parallel to make them meet the time constraint.

[0100] The attitude deviation estimation value refers to a wafer tilt state deviation amount calculated by Kalman filtering fusion of position and angle data, and can be specifically implemented by a six-axis sensor data fusion algorithm, and is used to quantify the attitude instability degree in the wafer handling process. The gradient descent algorithm refers to a mathematical method of iterative optimization of the partial derivative of the acceleration parameter to the attitude deviation, and can be specifically implemented by an adaptive learning rate optimizer, and is used to find the optimal acceleration parameter under the premise of ensuring the convergence speed. The dynamic adjustment mechanism refers to a control architecture for updating the motion instruction based on real-time sensor feedback and optimized parameters, and can be specifically implemented by a feedforward-feedback composite control structure, and is used to realize fast response and stable execution of the control instruction. The global speed optimization refers to a technical means for balancing the speed parameters of each path segment by a multi-objective optimization algorithm, and can be specifically implemented by a Lagrange multiplier method under a parallel computing framework, and is used to maximize the motion efficiency under the condition of satisfying the total handling time constraint.

[0101] Specifically, when the sensor detects that the wafer tilt angle exceeds a safety threshold, which is a critical parameter determined based on control performance requirements, sensor accuracy and system stability, first, the original attitude data is processed by Kalman filtering, and a reliable deviation estimation value is obtained after eliminating measurement noise interference. According to the comparison result of the deviation data and the preset acceleration range, the gradient descent algorithm is triggered to iteratively optimize the current acceleration parameter, and the adjustment amount of each iteration is determined by the partial derivative of the deviation to the acceleration. The optimized acceleration parameter is input into the dynamic adjustment module, and the updated speed sequence is generated in combination with the real-time collected wafer position information. After completing the local parameter adjustment, the system starts the global optimization module to perform parallel calculation on the speed parameters of multiple path segments, and ensures that the total handling time does not exceed the process requirement through constraint solving, and finally generates a speed control instruction that takes into account stability and efficiency.

[0102] Compared with the prior art, the traditional method usually uses a fixed acceleration threshold for speed adjustment, which cannot dynamically adjust the parameters according to the real-time attitude deviation, and lacks collaborative calculation of multiple path segments in global time optimization. The present scheme introduces a dynamic adjustment mechanism based on mathematical optimization to realize closed-loop control of the acceleration parameter and the attitude deviation, and uses a parallel computing architecture to process the global optimization problem of multiple path segments, effectively solving the technical problem that local adjustment and overall efficiency are difficult to balance under high-speed motion.

[0103] Through the above technical scheme, the present application can real-time correct the acceleration parameter to suppress the attitude deviation during high-speed wafer handling, avoid wafer sliding or damage caused by sudden vibration, and ensure overall handling efficiency through global optimization. The technical scheme not only ensures the motion stability of the key path segment, but also realizes the collaborative adjustment of the speed parameters of multiple path segments, and finally reaches a balanced state of high-speed motion and smooth handling.

[0104] In combination with actual needs, the application further proposes to perform global speed optimization on the speed control instruction based on the target carrying time, and to adjust the speed parameters of multiple path segments in parallel to make them meet the time constraint, so as to determine the final executed speed control instruction.

[0105] The target carrying time refers to a maximum time threshold allowed in the wafer carrying process, which can be implemented by using a process tact parameter or a device running cycle parameter as a constraint condition for global optimization. The global speed optimization refers to overall adjustment considering the correlation between speed parameters of multiple path segments, which can be implemented by using a multi-objective optimization algorithm or a constraint programming model to eliminate the cumulative time error caused by segmented optimization. The parallel adjustment refers to synchronous correction of the speed parameters of multiple path segments, which can be implemented by using a distributed computing framework or a parallel processing thread to maintain the coherence of the speed between different path segments. The time constraint refers to converting the target carrying time into a mathematical constraint condition, which can be implemented by using a Lagrange multiplier method or a penalty function method to force the speed parameters after optimization to meet the overall time requirement.

[0106] Specifically, after completing the multi-axis speed coordination control and the attitude deviation adjustment, the target carrying time is input as a rigid constraint condition into an optimization model. By constructing a multivariate equation group containing acceleration parameters of each path segment, a speed upper limit and a time distribution relationship, a nonlinear programming algorithm is used to globally solve the speed control instruction. In this process, the speed parameter adjustment operations of each path segment are executed synchronously, and the total time of each segment acceleration and deceleration process is converged to the target carrying time through iterative calculation. At the same time, by introducing the speed smoothness constraint between path segments, it is ensured that the adjusted speed curve maintains smooth motion under the premise of meeting the time limit.

[0107] Compared with the prior art, the traditional method usually adopts a sequential optimization strategy, that is, the speed parameters are adjusted segment by segment, and then the time is accumulated and verified. This way is easy to cause the end path segment to be forced to make a sharp speed adjustment to compensate for the time error in the early stage. The present scheme, by establishing a global optimization model, directly integrates the time constraint into the joint optimization process of multiple path segments, not only avoids the step-by-step transmission of time error, but also realizes the collaborative correction of the speed parameters of each path segment through a parallel computing mechanism.

[0108] Through the above technical scheme, the application can accurately control the overall carrying time under the premise of ensuring the attitude stability of the wafer carrying process, and effectively solve the time overrun problem caused by local speed optimization. By coordinating the speed parameters of multiple path segments from a global perspective, the application not only avoids the speed mutation phenomenon caused by segmented optimization in the traditional method, but also ensures the balance between the running efficiency and the time controllability of the device in the complex three-dimensional path.

[0109] Referring to Figure 8 To achieve the above control method, the application further discloses a wafer carrying speed control system, comprising:

[0110] a path processing module, configured to acquire three-dimensional coordinate data of a wafer carrying path, and split the three-dimensional coordinate data into straight line segments and curved segments through a path decomposition algorithm to obtain geometric feature descriptions of the path segments;

[0111] a curvature analysis module, connected with the path processing module, configured to calculate trajectory curvature values of the curved segments according to the geometric feature descriptions, smoothly connect the straight line segments and the curved segments, and determine a continuous path curvature distribution;

[0112] a speed planning module, connected with the curvature analysis module, configured to generate a preliminary speed planning curve based on the path curvature distribution and dynamically constrain the speed upper limit of a path point according to the curvature values;

[0113] a multi-axis coordination optimization module, connected with the speed planning module, configured to acquire joint parameters of multi-axis motion for the preliminary speed planning curve, calculate speed vectors of the axes through an inverse kinematics algorithm, and iteratively optimize the speed vectors to make the differences between the speed vectors of the axes within a coordination threshold, so as to generate a coordinated multi-axis speed sequence;

[0114] a posture monitoring and estimation module, connected with the multi-axis coordination optimization module, configured to acquire a wafer tilt angle fed back by a real-time sensor from the coordinated multi-axis speed sequence, and fuse position and angle data through a Kalman filtering algorithm to obtain a posture deviation estimation value;

[0115] an adaptive control module, connected with the posture monitoring and estimation module, configured to dynamically adjust acceleration parameters in the speed sequence based on the posture deviation estimation value, and reduce the posture deviation through an optimization algorithm, so as to output a final optimized speed control instruction to a carrying controller.

[0116] Obviously, the above embodiments are only examples for clearly illustrating, and are not intended to limit the embodiments. Based on the above description, other different forms of changes or variations can be made by those skilled in the art. Here, all the embodiments are not required to be exhausted, and the obvious changes or variations derived therefrom are still within the protection scope of the application.

Claims

1. A wafer handling speed control method, characterized by: The method comprises the following steps: Obtaining three-dimensional coordinate data of an object carrying path, segmenting the three-dimensional coordinate data into straight line segments and curved segments through a preset path decomposition algorithm, and obtaining geometric feature descriptions of each segment; According to the obtained geometric feature descriptions, calculating trajectory curvature values for the curved segments, smoothly connecting the straight line segments and the curved segments, and determining a continuous path curvature distribution; including: based on the geometric feature descriptions of the curved segments, using a curvature continuity constraint algorithm to optimize the arrangement of control points, and generating a fitting curve with smooth curvature transition; based on the geometric feature descriptions of the straight line segments, calculating the curvature change rate at the connection point of the straight line to the curve, and based on the curvature change rate, using a Bezier curve splicing algorithm to generate a smooth transition segment at the connection between the straight line segment and the curved segment; according to the fitting curve with smooth curvature transition and the smooth transition segment, calculating the curvature values of the continuous sampling points on the path, and generating a discrete curvature distribution sequence; using a Gaussian filtering algorithm to smooth the discrete curvature distribution sequence, eliminating local curvature fluctuations, and obtaining a continuous path curvature distribution; Based on the path curvature distribution, dynamically constraining the upper limit of the speed of the path points according to the curvature values, and generating a preliminary speed planning curve; For the obtained preliminary speed planning curve, obtaining joint parameters of multi-axis motion, calculating velocity vectors of each axis through inverse kinematics algorithm, and iteratively optimizing the velocity vectors to make the differences between the velocity vectors of each axis within a coordination threshold, thereby generating a coordinated multi-axis velocity sequence; including: obtaining the preliminary speed planning curve, extracting joint parameters of multi-axis motion from the preliminary speed planning curve, and determining initial motion states of each axis; through inverse kinematics algorithm, calculating joint velocity vectors of each axis for the extracted joint parameters, and obtaining an initial velocity vector set; based on the initial velocity vector set, using a particle swarm optimization algorithm to iteratively adjust the joint velocity vectors of each axis, taking the minimization of the differences between the velocity vectors of each axis as the target, thereby obtaining an optimized velocity vector set; through the optimized velocity vector set, generating a coordinated multi-axis velocity sequence, and determining the motion trajectories of each axis; From the determined coordinated multi-axis velocity sequence, obtaining an object inclination angle fed back by a real-time sensor according to the posture stability requirement, fusing position and angle data through Kalman filtering, and obtaining a posture deviation estimation value; Based on the attitude deviation estimation value, the acceleration parameters in the multi-axis speed sequence are dynamically adjusted, and an optimization algorithm is used to reduce the attitude deviation, so as to determine the optimized speed control instruction, including: obtaining the attitude deviation estimation value and the multi-axis speed sequence from the sensor data, using a preset threshold to classify the attitude deviation estimation value, if the attitude deviation estimation value is greater than the threshold, smoothing the deviation data by Kalman filtering algorithm; according to the deviation data, extract the acceleration parameters in the multi-axis speed sequence, use logical judgment to determine whether the acceleration parameters exceed the preset range, if exceed, adjust the acceleration parameters by gradient descent algorithm, get the optimized acceleration parameters; through the acceleration parameters, combined with the real-time feedback of the sensor data, using dynamic adjustment mechanism to update the multi-axis speed sequence, generate adjusted multi-axis speed sequence; according to the multi-axis speed sequence, using logical judgment to generate control instruction, get the optimized speed control instruction.

2. The wafer conveyance speed control method according to claim 1, characterized by: Three-dimensional coordinate data of the object carrying path is obtained, which is segmented into straight line segments and curved segments by a preset path decomposition algorithm, and the geometric feature description of each segment is obtained, including: Three-dimensional coordinate data of the object carrying path is obtained, which is segmented into straight line segments and curved segments by a preset path decomposition algorithm, and the geometric feature description of each segment is obtained, including: Three-dimensional coordinate data of the object carrying path is obtained, which is segmented into straight line segments and curved segments by a preset path decomposition algorithm, and the geometric feature description of each segment is obtained, including: Three-dimensional coordinate data of the object carrying path is obtained, which is segmented into straight line segments and curved segments by a preset path decomposition algorithm, and the geometric feature description of each segment is obtained, including: Three-dimensional coordinate data of the object carrying path is obtained, which is segmented into straight line segments and curved segments by a preset path decomposition algorithm, and the geometric feature description of each segment is obtained, including: Three-dimensional coordinate data of the object carrying path is obtained, which is segmented into straight line segments and curved segments by a preset path decomposition algorithm, and the geometric feature description of each segment is obtained, including: Three-dimensional coordinate data of the object carrying path is obtained, which is segmented into straight line segments and curved segments by a preset path decomposition algorithm, and the geometric feature description of each segment is obtained, including: Three-dimensional coordinate data of the object carrying path is obtained, which is segmented into straight line segments and curved segments by a preset path decomposition algorithm, and the geometric feature description of each segment is obtained, including:

3. The wafer handling speed control method according to claim 1, characterized by: The discrete curvature distribution sequence is smoothed by a Gaussian filtering algorithm, including: Set the Gaussian filtering window size and standard deviation parameters, and adaptively adjust the filtering strength based on the total length of the path and the curvature change frequency; Discrete curvature sequence is filtered by convolution operation, which can eliminate local curvature fluctuation while retaining global curvature change characteristics; The continuity of the filtered curvature distribution is tested to ensure the continuity of the first and second derivatives of the curvature distribution curve.

4. The wafer handling speed control method of claim 1, wherein: Based on the path curvature distribution, the speed upper limit of the path point is dynamically constrained according to the curvature value, and a preliminary speed planning curve is generated, including: The curvature value of each path point is obtained from the path curvature distribution, and a preset threshold is used to judge the curvature size, if the curvature value is greater than the preset threshold, the speed upper limit set is determined according to the mapping relationship between the curvature value and the speed upper limit; A speed upper limit curve is fitted by using a spline interpolation method based on the set of speed upper limits and the sequence of coordinate points of the path points, to obtain a preliminary speed planning curve.

5. The wafer handling speed control method of claim 1, wherein: From the determined coordinated multi-axis speed sequence, an object tilt angle is obtained based on real-time sensor feedback, a Kalman filter is used to fuse position and angle data, and an attitude deviation estimate is obtained, including: A coordinated control instruction is obtained from the multi-axis speed sequence, each axis speed is synchronously processed by a preset motion planning algorithm, and a unified motion instruction set is obtained; Based on the unified motion instruction set, sensor feedback data is combined, an accelerometer and a gyroscope are used to collect the object tilt angle in real time, and original attitude data is obtained; The position data and angle data in the original attitude data are fused by a Kalman filter, and the state estimate is iteratively updated to obtain an optimized attitude deviation estimate.

6. The wafer handling speed control method according to claim 1, characterized by: Also includes: Based on the target carrying time, the speed control instruction is globally optimized, the speed parameters of multiple path segments are adjusted in parallel to meet the time constraint, and the final executed speed control instruction is determined.

7. A wafer handling speed control system characterized by: For implementing the method of any one of claims 1-6, comprising: A path processing module is configured to obtain three-dimensional coordinate data of a wafer carrying path, and split the data into straight line segments and curved segments by a path decomposition algorithm to obtain geometric feature descriptions of each path segment; A curvature analysis module is connected to the path processing module and is configured to calculate trajectory curvature values of the curved segments based on the geometric feature descriptions, smoothly connect the straight line segments and the curved segments, and determine a continuous path curvature distribution; A speed planning module is connected to the curvature analysis module and is configured to generate a preliminary speed planning curve based on the path curvature distribution and dynamically constrain the speed upper limit of the path points according to the curvature values; A multi-axis coordination optimization module is connected to the speed planning module and is configured to obtain joint parameters of multi-axis motion for the preliminary speed planning curve, calculate velocity vectors of each axis by an inverse kinematics algorithm, and iteratively optimize the velocity vectors to make the differences between the velocity vectors within a coordination threshold, thereby generating a coordinated multi-axis speed sequence; An attitude monitoring and estimation module is connected to the multi-axis coordination optimization module and is configured to obtain a wafer tilt angle based on real-time sensor feedback from the coordinated multi-axis speed sequence, and fuse position and angle data by a Kalman filter algorithm to obtain an attitude deviation estimate; An adaptive control module is connected to the attitude monitoring and estimation module and is configured to dynamically adjust acceleration parameters in the speed sequence based on the attitude deviation estimate, and reduce the attitude deviation by an optimization algorithm, thereby outputting a final optimized speed control instruction to a carrying controller.

Images