A robot trajectory accuracy optimization method based on measurement equipment guidance

Abstract

The present invention discloses a robot trajectory accuracy optimization method based on measurement equipment guidance, which specifically relates to the technical field of precision optimization. The method comprises controlling a robot to execute a processing trajectory through visual servoing, and inputting a specific control trajectory with deviation by setting a receiving frequency in a visual servo controller to realize automatic correction of the robot motion deviation and return to a theoretical trajectory. The actual trajectory, the control trajectory and the theoretical trajectory are used as input to generate a new theoretical trajectory. Based on the high repeatability of the robot, the new theoretical trajectory is input to the robot controller, and the robot is directly controlled to perform processing according to the new theoretical trajectory without using a visual system. The new theoretical trajectory is evaluated by collecting time delay information and deviation information when the robot applies the new theoretical trajectory, and an early warning signal is generated. The present invention helps to improve processing efficiency and the accuracy of the processing trajectory, and ensures the rationality of the processing trajectory.

Description

Technical Field

[0001] The present invention relates to the technical field of precision optimization, and more particularly to a method for optimizing robot trajectory precision based on guidance by a measuring device. Background Art

[0002] With the rise of manufacturing and the development of automation, industrial robots are playing an increasingly important role in the manufacturing sector. Especially in the automotive assembly, aerospace and other fields, industrial robots have been widely used in various production tasks.

[0003] However, due to the characteristics of industrial robots, which have low absolute positioning accuracy and high repeatability, directly inputting the theoretical trajectory for processing will produce large deviations, while the teaching adjustment method is inefficient and cannot handle complex curves. In some areas with low precision requirements and large-scale single needs, such as automobile assembly, industrial robots are used on a large scale. However, in some areas with high precision requirements and small-batch customization needs, such as aerospace, the trajectory planning methods of traditional industrial robots still find it difficult to complete various tasks. In particular, for the processing of some customized complex curve workpieces, traditional trajectory planning methods are difficult to calibrate high-precision trajectories, resulting in low final processing accuracy, which in turn affects production quality. This method analyzes and controls the trajectory of the robot during processing, and by reproducing the robot control trajectory, fully utilizing the high repeatability of the robot, the actual position of the robot is close to the theoretical position, thereby ensuring the accuracy of the robot trajectory.

[0004] To address these issues, publication number CN115781716A discloses a visual servo trajectory compensation method for industrial robots. This method uses a binocular vision sensor to track the end-point of the industrial robot in real time, comparing the offline-generated trajectory with the end-point's position and correcting it in real time. While this method enables high-precision machining for industrial robots, the visual system is significantly affected by environmental factors and is easily obstructed.

[0005] Publication No. CN119511734A discloses a robot trajectory optimization method that uses B-spline curve fitting. While this method ensures trajectory smoothness, it fails to account for the unique characteristics of the robot's trajectory, making it difficult to guarantee high precision and accuracy. Furthermore, using B-spline and other fitting methods to optimize the robot's trajectory can lead to further inward deviations due to the robot's own refitting. Summary of the Invention

[0006] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a robot trajectory accuracy optimization method based on guidance of a measurement device to solve the problems raised in the above-mentioned background technology.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A method for optimizing robot trajectory accuracy based on guidance by a measuring device, specifically comprising:

[0009] S1: Build the hardware and software of the robot visual servo compensation system, including an industrial robot 1, a visual sensor 2, a robot end effector 3, and supporting structural components. The visual sensor 2 can return the actual position of the robot in real time, calculate the difference between the theoretical trajectory and the actual trajectory, and adjust the robot's motion trajectory in real time to correct the robot's position deviation.

[0010] S2: Controlling the robot's machining trajectory through visual servoing. This involves extracting the robot's actual trajectory, control trajectory, and theoretical trajectory during machining. By setting the receiving frequency in the visual servo controller, the time step consistency of the point data between the actual trajectory and the control trajectory is ensured. Furthermore, by inputting a specific control trajectory with deviations, the robot's motion deviations are automatically corrected and the robot returns to the theoretical trajectory.

[0011] S3: Taking the actual trajectory, the controlled trajectory, and the theoretical trajectory as input, the trajectory is segmented into straight and curved segments according to the curvature of the motion trajectory of the industrial robot 1 and a small line segment model is established. The extracted trajectory is subjected to hybrid filtering according to the characteristics of the controller of the industrial robot 1 and the visual sensor 2. Based on the characteristics of the arc transition of the controller of the industrial robot 1, the extracted trajectory is refitted to generate a new theoretical trajectory.

[0012] S4: Based on the high repeatability of the robot, the new theoretical trajectory is input into the robot controller. Without using the visual system, the robot is directly controlled to process according to the new theoretical trajectory. By collecting the delay information and deviation information when the robot applies the new theoretical trajectory, the new theoretical trajectory is evaluated and a warning signal is generated.

[0013] In a preferred embodiment, a measurement device control system is built in S1 to identify the actual posture of the robot end, correct the robot trajectory error through system calculation, and be able to return the actual posture and control posture synchronously with the robot controller;

[0014] In said S2, the actual trajectory of the robot extracted by the measuring device, the control trajectory extracted by the robot controller and the theoretical trajectory are used for analysis;

[0015] In the above S3, the curvature characteristics of the actual trajectory and the theoretical trajectory are used as the basis for judging the straight and curved segmentation, the control trajectory extracted by the robot controller is used to perform denoising and exclude outliers, a small line segment model for straight and curved separation is established through synchronous mapping of the actual trajectory, and the small line segment model is fitted by least squares straight line fitting and inverse curve segment control points respectively;

[0016] In the aforementioned S4, the measurement device guidance is not used, and the measurement device guidance effect can be achieved by inputting a new theoretical trajectory.

[0017] In a preferred embodiment, the trajectory is segmented into straight and curved segments and a small line segment model is established, including:

[0018] First, convert the robot trajectory XYZ into a parametric equation about time T:

[0019] ;

[0020] Use cubic spline to smooth X, Y, and Z respectively to get the spline function of the actual trajectory: ;

[0021] Find the first and second derivatives of a spline function:

[0022] ;

[0023] Curvature function of the robot trajectory:

[0024] ;

[0025] The parametric equation of the known space curve is:

[0026] ;

[0027] Then its tangent vector and second-order derivative are:

[0028] ;

[0029] ;

[0030] Substituting in:

[0031] ;

[0032] ;

[0033] Curvature function of the robot trajectory The component form is:

[0034] ;

[0035] According to the curvature of the curve, the moment when the actual trajectory approaches a straight line or an arc is distinguished, and the straight line segment and the arc segment are separated;

[0036] The theoretical stationary point is extracted separately, and the corresponding theoretical trajectory stationary point is found on the actual trajectory according to the minimum distance and marked as a curve segment.

[0037] In a preferred embodiment, hybrid filtering is performed on the extracted trajectory, including:

[0038] A hybrid filtering method of low-pass filtering and median filtering is adopted for the control trajectory. When extracting the trajectory, the points in the actual trajectory and the control trajectory are one-to-one corresponding. The small line segment models segmented by the actual trajectory are mapped one-to-one to the control trajectory. The segmented straight line part is fitted by least squares interpolation to obtain the coordinates of the straight line segment endpoints.

[0039] In a preferred embodiment, the delay information of the new theoretical trajectory includes:

[0040] The time delay information when the robot applies the new theoretical trajectory is represented by the trajectory response delay coefficient;

[0041] The trajectory response delay coefficient is obtained by determining a set of visual servo frequencies, applying the new theoretical trajectory to different visual servo frequencies, and determining the time interval between the time when the robot controller issues a command and the time when the robot actually reaches the new theoretical trajectory point as the response time at different visual servo frequencies. The robot's response time is then determined several times at different visual servo frequencies, and the average response delay of the robot is calculated. The average response delay of the robot under multiple repetitions at different visual servo frequencies is marked as: ; Wherein, f is the number of different visual servo frequencies, n is the number of repetitions, f=1, 2, 3, ..., F, F is a positive integer, n=1, 2, 3, ..., N, N is a positive integer;

[0042] Based on the number of trajectory points that the robot can actually process and track and the average response time of the robot, a timeliness decreasing effect model is constructed. Based on the average response delay of robots with different frequencies under the visual servo frequency set, different average response delays are scored and used as trajectory delay scores. The trajectory delay scores are used to optimize the model parameters of the timeliness decreasing effect model based on the nonlinear least squares method to obtain a calculation model for the trajectory response delay coefficient. The objective function used to optimize the timeliness decreasing effect model is: ; Where m is the number of the trajectory delay score, m=1, 2, 3, ..., M, M is a positive integer, is the trajectory delay score, A is the maximum upper limit of the trajectory delay score, and YC is the robot's forward tracking ability for the new theoretical trajectory;

[0043] The calculation model of the trajectory response delay coefficient is: ;in, is the trajectory response delay coefficient.

[0044] In a preferred embodiment, the deviation information of the new theoretical trajectory includes:

[0045] The deviation information when the robot applies the new theoretical trajectory is represented by the trajectory deviation evaluation coefficient;

[0046] The logic for obtaining the trajectory deviation evaluation coefficient is as follows: based on the new theoretical trajectory, obtain the marker points existing in the new theoretical trajectory, the marker points including the stationary point, the acceleration mutation point and the attitude switching point, and score the different marker points based on the marker points existing in the new theoretical trajectory to obtain the importance score value of the marker point;

[0047] The position deviation between the actual trajectory of the robot and the new theoretical trajectory is obtained. The position deviation between the actual trajectory and the new theoretical trajectory at different marking points and the importance score of different marking points are used as input features, and the trajectory deviation evaluation coefficient is used as the output feature to construct a logistic regression model. The calculation formula of the trajectory deviation evaluation coefficient is:

[0048] ;

[0049] in, is the trajectory deviation evaluation coefficient, 、 、 、……、 is the position deviation between the actual trajectory and the new theoretical trajectory at different marking points, 、 、 、……、 is the importance score of different marked points, and e is the base.

[0050] In a preferred embodiment, evaluating the new theoretical trajectory includes:

[0051] The delay information and deviation information of the new theoretical trajectory in the application are comprehensively analyzed. The new theoretical trajectory evaluation model is constructed by calculating the weighted sum of the trajectory response delay coefficient and the trajectory deviation evaluation coefficient, and the new theoretical trajectory evaluation coefficient is generated. The calculation formula of the new theoretical trajectory evaluation coefficient is: ;in, is the new theoretical trajectory evaluation coefficient, 、 are the proportional coefficients of trajectory response delay coefficient and trajectory deviation evaluation coefficient respectively, 、 are both greater than 0;

[0052] A new theoretical trajectory evaluation coefficient threshold is set, and the new theoretical trajectory evaluation coefficient of the new theoretical trajectory is compared with the new theoretical trajectory evaluation coefficient threshold. If the new theoretical trajectory evaluation coefficient is less than the new theoretical trajectory evaluation coefficient threshold, a warning signal is generated; if the new theoretical trajectory evaluation coefficient is greater than the new theoretical trajectory evaluation coefficient threshold, no warning signal is generated.

[0053] The technical effects and advantages of the present invention are as follows:

[0054] The present invention is based on a measurement and compensation system. After processing the trajectory extracted by the visual system, a new theoretical trajectory is generated to directly control the robot. This advances the process of the measurement system in industrial production from the processing process to the process planning process, greatly reducing the cost of the measurement and compensation system processing, and is conducive to improving processing efficiency and the accuracy of the processing trajectory.

[0055] The present invention provides a method for fitting a robot motion trajectory that conforms to the robot's characteristics, effectively removing noise and interference points from the robot trajectory and improving the accuracy of fitting the robot trajectory;

[0056] The present invention provides a method for converting the absolute position of a robot into a repeatable position of the robot for robot trajectory processing. By utilizing the high repeatability of the robot, the control trajectory of the robot processing process is reproduced, thereby ensuring the accuracy of the robot processing.

[0057] The present invention provides an evaluation method for a robot's working trajectory. By fitting a new theoretical trajectory to the robot, the time delay information and deviation information when the robot applies the new theoretical trajectory are collected. Based on the influence of time delay and position deviation, the method helps to determine the accuracy of the robot's trajectory. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings;

[0059] Figure 1 A schematic diagram of the positions of various components in the simulation environment of the present invention;

[0060] Figure 2 Schematic diagram of each trajectory extracted by the present invention;

[0061] Figure 3 Schematic diagram of the curve segment fitting method of the present invention;

[0062] Figure 4 It is a schematic diagram of the process of the present invention.

[0063] In the figure: 1. Industrial robot; 2. Vision sensor; 3. Robot end effector. DETAILED DESCRIPTION

[0064] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0065] Example 1

[0066] Figure 4 The present invention provides a robot trajectory accuracy optimization method based on measurement equipment guidance, which specifically includes the following steps:

[0067] S1: Build the hardware and software of the robot visual servo compensation system, including an industrial robot 1, a visual sensor 2, a robot end effector 3, and supporting structural components. The visual sensor 2 can return the actual position of the robot in real time, calculate the difference between the theoretical trajectory and the actual trajectory, and adjust the robot's motion trajectory in real time to correct the robot's position deviation.

[0068] S2: Controlling the robot's machining trajectory through visual servoing. This involves extracting the robot's actual trajectory, control trajectory, and theoretical trajectory during machining. By setting the receiving frequency in the visual servo controller, the time step consistency of the point data between the actual trajectory and the control trajectory is ensured. Furthermore, by inputting a specific control trajectory with deviations, the robot's motion deviations are automatically corrected and the robot returns to the theoretical trajectory.

[0069] S3: Taking the actual trajectory, the controlled trajectory, and the theoretical trajectory as input, the trajectory is segmented into straight and curved segments according to the curvature of the motion trajectory of the industrial robot 1 and a small line segment model is established. The extracted trajectory is subjected to hybrid filtering according to the characteristics of the controller of the industrial robot 1 and the visual sensor 2. Based on the characteristics of the arc transition of the controller of the industrial robot 1, the extracted trajectory is refitted to generate a new theoretical trajectory.

[0070] S4: Based on the high repeatability of the robot, the new theoretical trajectory is input into the robot controller. Without using the vision system, the robot is directly controlled to process according to the new theoretical trajectory. The new theoretical trajectory is evaluated by collecting the delay information and deviation information when the robot applies the new theoretical trajectory.

[0071] In step 1, the robot end effector 3 is installed on the robot flange, the vision sensor 2 covers the robot processing area, and the robot end effector 3 moves in the processing area. The robot base and the robot end effector 3 are calibrated by the vision sensor 2 to achieve the unification of the robot coordinate system, the vision sensor coordinate system and the world coordinate system.

[0072] It should be noted that the actual trajectory is measured by the visual system of the visual sensor, the control trajectory is measured by the controller of the industrial robot 1 itself, and the theoretical trajectory is the input processing trajectory, that is, the trajectory obtained offline. The time step consistency of the point position data in the actual trajectory and the control trajectory can ensure that the point position data in the actual trajectory and the control trajectory can correspond one to one.

[0073] The further S2 step specifically comprises the following steps:

[0074] S21, in the offline programming software, define a spatial curve as the processing path according to the digital model of the object to be processed, and ensure that the robot will not have a singular or unreachable posture, and serve as the robot's theoretical trajectory.

[0075] It should be noted that offline programming software refers to a robot simulation offline programming platform. Using offline programming software, the robot can be planned and simulated on a computer. The robot's processing path is defined in the offline programming software to ensure that the length of the robot's arm and the range of joint rotation angles allow the end to reach all points on the processing path, and that no point on the processing path is near a singular position of the robot arm.

[0076] S22, real-time visual servo control is performed on the robot's motion process. Since there is data transmission between the robot controller and the visual sensor 2 during visual servoing, but the transmission frequencies of both parties are not necessarily the same, it is necessary to set the recording frequency and record according to the minimum value of the frequency between the robot controller and the visual sensor 2. This ensures that the recorded robot controller value and the visual sensor value are consistent in time, thereby ensuring that after the visual servo operation is completed, the actual trajectory extracted from the visual servo system and the points in the control trajectory correspond one to one.

[0077] It should be noted that real-time visual servo control means that the actual robot uses a camera (or other visual sensor) to continuously monitor the position and posture of the end during operation, and then compares the measured "actual posture" with the "trajectory command originally issued by the robot controller". Correction commands are issued in real time to allow the robot to dynamically align with the theoretical trajectory. During the real-time visual servo control process, the timing of the actual trajectory and the control trajectory is ensured to be aligned to obtain the correct deviation value.

[0078] S23, such as Figure 2As shown in the figure, given the influence of geometric errors and non-geometric error factors of the robot itself, when the robot controller controls the robot movement, there is a significant deviation between the actual position of the robot in space and the position actually measured by the visual system, while there is only a small deviation between the visual trajectory measured by the visual system and the pre-set theoretical trajectory. In addition, the control trajectory and actual trajectory data extracted from the robot controller and the visual system are often accompanied by a lot of noise and interference, and the number of data points is huge, which is not conducive to subsequent processing.

[0079] It should be noted that due to the presence of geometric errors (such as joint installation and processing errors) and non-geometric errors (such as stiffness deformation, joint friction, drive delay, etc.) in the robot body, there will be a large difference between the trajectory command (control trajectory) issued by the robot controller and the end position (actual trajectory) actually measured by the vision system; and the trajectory measured by the vision system usually has a relatively small error compared with the theoretical trajectory of offline planning.

[0080] Further S3 specific steps include the following steps:

[0081] S31, the robot's trajectory is divided into curved and straight segments. Since the derived control trajectory has obvious jitter and is far from the theoretical trajectory, it is difficult to accurately determine its curved-straight segments. The actual trajectory is close to the theoretical trajectory, and its noise and interference are relatively small. The theoretical trajectory is used as a corresponding reference, so the actual trajectory can be divided into its curved-straight segments and then mapped to the control trajectory.

[0082] First, convert the robot trajectory XYZ into a parametric equation about time T:

[0083] ;

[0084] Use cubic spline to smooth X, Y, and Z respectively to get the spline function of the actual trajectory: ;

[0085] Find the first and second derivatives of a spline function:

[0086] ;

[0087] Curvature function of the robot trajectory:

[0088] ;

[0089] The parametric equation of the known space curve is:

[0090] ;

[0091] Then its tangent vector (first-order derivative) and second-order derivative are:

[0092] ;

[0093] ;

[0094] Substituting in:

[0095] ;

[0096] ;

[0097] So we can get the curvature function of the robot trajectory The component form is:

[0098] ;

[0099] The moment when the actual trajectory approaches a straight line or an arc is distinguished according to the curvature of the curve, thereby separating the straight line segment and the arc segment.

[0100] S32, the theoretical stationary point needs to be extracted separately, and the corresponding theoretical trajectory stationary point is found on the actual trajectory according to the minimum distance, and is directly marked as a curve segment, without judging the curvature within a certain range on both sides of the stationary point.

[0101] It should be noted that the robot trajectory exhibits significant jitter at the theoretical stationary point, especially at large corners. The curvature near the corners varies significantly, and the robot's own position error is also large. Using curvature alone can easily lead to misjudgment of corners, or extracting multiple corner points simultaneously can result in a densely packed collection of points. Extracting the theoretical stationary points effectively ensures that the shape of the new theoretical trajectory remains consistent with the original.

[0102] S33, after the segmentation of S32 and S31, a number of small line segments are obtained, and the small line segments are recorded in the form of point serial numbers.

[0103] S34, adopting a hybrid filtering method of low-pass filtering and median filtering to the control trajectory;

[0104] It should be noted that the hybrid filtering method of low-pass filtering plus median filtering for the control trajectory can effectively eliminate high-frequency errors and accidental errors, ensuring that the control trajectory is relatively smooth and convenient for subsequent fitting.

[0105] The low-pass filter uses a fourth-order Butterworth low-pass filter, which can be decomposed into a cascade of two second-order systems:

[0106] ;

[0107] where the transfer function of each second-order section is:

[0108] ;

[0109] Among them, the cut-off frequency , quality factor , ;

[0110] The median filter uses a window size of 3 to minimize significant changes to the data itself and thus distortion:

[0111]

[0112] S35 , based on the one-to-one correspondence between the points in the actual trajectory and the control trajectory when extracting the trajectory, the small line segment models segmented by the actual trajectory are mapped one-to-one to the control trajectory.

[0113] S36, performing linear fitting on the segmented straight line portion using least squares interpolation to obtain the coordinates of the straight line segment endpoints.

[0114] S37, by Figure 3 As shown, the robot's corner trajectory typically transitions into an inscribed arc. This characteristic of the robot allows the control vertices of curve segment P1P2 to be inversely determined. Extract the line segments L1 and L2 to the left and right of the curve segment. Project the two line segments onto the same plane, using the direction of the robot's minimum feed accuracy as the projection direction. Calculate the intersection point P0 of the two lines. Project P0 back onto lines L1 and L2, respectively, to obtain points P3 and P4. Take P5, the midpoint between P4 and P3, as the new control point for the curve segment. Merge line segments P6P5 and P2P7 with curve segment P1P2 to form two lines, P6P5 and P5P7.

[0115] S38: In the P1P2 segment, the robot's posture does not change much, so the posture of point P5 is the average of the postures of points P1 and P2.

[0116] S39, repeating steps S37 and S38 for all curve segments in sequence to complete the fitting of the original control trajectory and form a new control trajectory.

[0117] In summary, the present invention proposes a method for optimizing robot trajectory accuracy based on measurement device guidance, which can address the low absolute positioning accuracy of industrial robot 1. This method fully utilizes the robot's high repeatability while also avoiding the cost and environmental constraints of using visual servoing, significantly improving the efficiency and accuracy of robot processing. Furthermore, the proposed small line segment model and line segment fitting method for the robot trajectory significantly improve the accuracy of robot trajectory fitting.

[0118] In this specification, the "trajectory" of the robot refers to both the position information of the robot end and the posture information of the robot end.

[0119] Example 2

[0120] Based on the new theoretical trajectory in Example 1, in order to ensure that the new theoretical trajectory can be applied to the robot to work without using visual servoing, determine the reliability of the new theoretical trajectory under different conditions, and avoid the risk of "poor fitting" or "over-filtering" in the offline algorithm, a comprehensive evaluation is performed on the new theoretical trajectory to determine the rationality of the new theoretical trajectory. In this embodiment, the robot is not limited to the industrial robot 1.

[0121] The new theoretical trajectory solves the errors caused by fixed system errors, including those due to rigidity error, dynamic inertia, joint friction, visual delay, etc. In order to ensure that the new theoretical trajectory can meet the process requirements, the delay information and deviation information when the robot applies the new theoretical trajectory are collected. The delay information when the robot applies the new theoretical trajectory is represented by the trajectory response delay coefficient, and the deviation information when the robot applies the new theoretical trajectory is represented by the trajectory deviation evaluation coefficient.

[0122] The trajectory response delay coefficient is used to measure the impact of the delay from "the controller issuing the trajectory command" to "the robot end actually tracking the trajectory" on the trajectory fidelity (or tracking accuracy) in the visual servo closed loop. The specific advantages are:

[0123] The trajectory response delay coefficient can more comprehensively evaluate the feasibility and fidelity of each trajectory in a closed-loop environment, including the robot's continuous tracking capability and feedback delay. Through offline reliability testing of new theoretical trajectories at multiple visual frame rates, the optimized parameters of the robot's time-limited diminishing effect model are obtained. The trajectory response delay coefficient is determined using this model, effectively avoiding workpiece scrapping or rework caused by trajectory distortion.

[0124] By evaluating the new theoretical trajectory, we can also determine the performance of the time-diminishing effect model parameters under different camera, image algorithm, controller, or processor combinations under the new theoretical trajectory. This can be used to judge the pros and cons of hardware selection for the robot control system and ensure the accuracy of the robot trajectory.

[0125] The trajectory response delay coefficient is obtained by determining a set of visual servo frequencies, applying the new theoretical trajectory to different visual servo frequencies, and determining the time interval between the time when the robot controller issues a command and the time when the robot actually reaches the new theoretical trajectory point as the response time at different visual servo frequencies. The robot's response time is then determined several times at different visual servo frequencies, and the average response delay of the robot is calculated. The average response delay of the robot under multiple repetitions at different visual servo frequencies is marked as: ; Wherein, f is the number of different visual servo frequencies, n is the number of repetitions, f=1, 2, 3, ..., F, F is a positive integer, n=1, 2, 3, ..., N, N is a positive integer;

[0126] It should be noted that the visual servo frequency represents the update rate at which the camera captures images, performs error calculations, and then outputs control instructions in the visual servo control loop, that is, the frequency determined in S22 in Example 1. The visual servo frequency is affected by the hardware equipment and directly affects the robot's response delay to trajectory deviation and control accuracy.

[0127] Based on the number of trajectory points that the robot can actually process and track and the average response time of the robot, a timeliness decreasing effect model is constructed. Based on the average response delay of robots with different frequencies under the visual servo frequency set, different average response delays are scored and used as trajectory delay scores. The trajectory delay scores are used to optimize the model parameters of the timeliness decreasing effect model based on the nonlinear least squares method to obtain a calculation model for the trajectory response delay coefficient. The objective function used to optimize the timeliness decreasing effect model is: ; Where m is the number of the trajectory delay score, m=1, 2, 3, ..., M, M is a positive integer, is the trajectory delay score, A is the maximum upper limit of the trajectory delay score, and YC is the robot's forward tracking ability for the new theoretical trajectory;

[0128] It should be noted that the robot's forward tracking capability for the new theoretical trajectory refers to the duration for which the robot can continuously and effectively maintain trajectory tracking (the error remains within an acceptable threshold) under a given visual servo frequency f.

[0129] The calculation model of the trajectory response delay coefficient is: ;in, is the trajectory response delay coefficient.

[0130] As can be seen from the formula, the larger the trajectory response delay coefficient, the smaller the average delay from the robot issuing the trajectory point command to the end point actually tracking the target, and the faster it can complete a closed-loop correction. In other words, the robot accumulates less error during the entire closed-loop process, and the actual running trajectory is closer to the designed theoretical trajectory.

[0131] The trajectory deviation evaluation coefficient is used to quantify the impact of key node errors on overall tracking. Its specific advantages are:

[0132] In a visual servo closed loop, the robot needs to correct lag and jitter in real time from visual capture to actual movement. Using a logistic regression model, the trajectory deviation evaluation coefficient can intuitively reflect the current visual servo tracking accuracy at key locations and whether the closed loop can maintain high fidelity.

[0133] Traditional methods often require multiple rounds of long-path test runs to determine whether a certain trajectory performs well in a closed-loop environment. The trajectory deviation evaluation coefficient can determine the key points of the new theoretical trajectory, reducing a large amount of ineffective test runs and parameter adjustment work.

[0134] The logic for obtaining the trajectory deviation evaluation coefficient is as follows: based on the new theoretical trajectory, obtain the marker points existing in the new theoretical trajectory, the marker points including the stationary point, the acceleration mutation point and the attitude switching point, and score the different marker points based on the marker points existing in the new theoretical trajectory to obtain the importance score value of the marker point;

[0135] It should be noted that the stationary point refers to the point where the robot completes the switching of acceleration and deceleration, which is usually the connection point of a curve and a straight line. The acceleration mutation point refers to the point where the robot needs to instantly switch to uniform speed or uniform deceleration after a period of uniform acceleration. The acceleration profile generated by interpolation will jump at the switching point. The posture switching point is usually specified by the trajectory planning software (or offline programming) at this point. The scoring principle for different marking points is based on expert experience or marking point sensitivity analysis, and the scoring values ​​are normalized.

[0136] The position deviation between the actual trajectory of the robot and the new theoretical trajectory is obtained. The position deviation between the actual trajectory and the new theoretical trajectory at different marking points and the importance score of different marking points are used as input features, and the trajectory deviation evaluation coefficient is used as the output feature to construct a logistic regression model. The calculation formula of the trajectory deviation evaluation coefficient is:

[0137] ;

[0138] in, is the trajectory deviation evaluation coefficient, 、 、 、……、 is the position deviation between the actual trajectory and the new theoretical trajectory at different marking points, 、 、 、……、 is the importance score of different marked points, and e is the base.

[0139] It can be seen from the formula that the larger the trajectory deviation evaluation coefficient, the easier it is for the robot to accurately execute according to the theoretical trajectory at the key position, the smaller the error, and the higher the overall tracking accuracy. In other words, the new theoretical trajectory has strong executableness and fidelity under the current visual servo closed loop.

[0140] The delay information and deviation information of the new theoretical trajectory in the application are comprehensively analyzed. The new theoretical trajectory evaluation model is constructed by calculating the weighted sum of the trajectory response delay coefficient and the trajectory deviation evaluation coefficient, and the new theoretical trajectory evaluation coefficient is generated. The calculation formula of the new theoretical trajectory evaluation coefficient is: ;in, is the new theoretical trajectory evaluation coefficient, 、 are the proportional coefficients of trajectory response delay coefficient and trajectory deviation evaluation coefficient respectively, 、 Both are greater than 0.

[0141] It can be seen from the formula that the smaller the trajectory response delay coefficient and the trajectory deviation evaluation coefficient, the smaller the new theoretical trajectory evaluation coefficient. This means that when the new theoretical trajectory is applied to the robotic process, there are problems such as slow trajectory response and position deviation. This means that the new theoretical trajectory may only improve the error caused by the fixed system error, but in actual applications it may lead to insufficient overall trajectory tracking effect, making it difficult to meet the requirements of high-precision or high-consistency processes.

[0142] A new theoretical trajectory evaluation coefficient threshold is set, and the new theoretical trajectory evaluation coefficient of the new theoretical trajectory is compared with the new theoretical trajectory evaluation coefficient threshold. If the new theoretical trajectory evaluation coefficient is less than the new theoretical trajectory evaluation coefficient threshold, a warning signal is generated to indicate that the execution fidelity and real-time performance of the current new theoretical trajectory under the visual servo control system do not meet the process requirements, and the new theoretical trajectory, visual servo parameters or control strategy need to be re-optimized or adjusted. If the new theoretical trajectory evaluation coefficient is greater than the new theoretical trajectory evaluation coefficient threshold, no warning signal is generated.

[0143] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0144] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center via wired or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0145] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0146] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0147] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0148] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0149] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A robot trajectory accuracy optimization method based on measurement equipment guidance, characterized in that: Specifically include: S1: Build the hardware and software of the robot visual servo compensation system, including an industrial robot (1), a visual sensor (2), a robot end effector (3) and supporting structural parts, wherein the visual sensor (2) can return the actual position of the robot in real time, calculate the difference between the theoretical trajectory and the actual trajectory, and adjust the robot motion trajectory in real time to correct the robot position deviation; S2: Controlling the robot's machining trajectory through visual servoing. This involves extracting the robot's actual trajectory, control trajectory, and theoretical trajectory during machining. By setting the receiving frequency in the visual servo controller, the time step consistency of the point data between the actual trajectory and the control trajectory is ensured. Furthermore, by inputting a specific control trajectory with deviations, the robot's motion deviations are automatically corrected and the robot returns to the theoretical trajectory. S3: Taking the actual trajectory, the control trajectory and the theoretical trajectory as input, the trajectory is segmented into straight and curved segments according to the curvature of the motion trajectory of the industrial robot (1) and a small line segment model is established. The extracted trajectory is subjected to hybrid filtering according to the characteristics of the controller of the industrial robot (1) and the visual sensor (2). Based on the characteristics of the arc transition of the controller of the industrial robot (1), the extracted trajectory is refitted to generate a new theoretical trajectory. S4: Based on the high repeatability of the robot, the new theoretical trajectory is input into the robot controller. Without using the visual system, the robot is directly controlled to process according to the new theoretical trajectory. By collecting the delay information and deviation information when the robot applies the new theoretical trajectory, the new theoretical trajectory is evaluated and a warning signal is generated.

2. The method for optimizing robot trajectory accuracy based on measurement equipment guidance according to claim 1, characterized in that: include: In the aforementioned S1, a measurement device control system is built to identify the actual posture of the robot end, correct the robot trajectory error through system calculation, and be able to synchronize with the robot controller to return the actual posture and control posture respectively; In said S2, the actual trajectory of the robot extracted by the measuring device, the control trajectory extracted by the robot controller and the theoretical trajectory are used for analysis; In the above S3, the curvature characteristics of the actual trajectory and the theoretical trajectory are used as the basis for judging the straight and curved segmentation, the control trajectory extracted by the robot controller is used to perform denoising and exclude outliers, a small line segment model for straight and curved separation is established through synchronous mapping of the actual trajectory, and the small line segment model is fitted by least squares straight line fitting and inverse curve segment control points respectively; In the aforementioned S4, the measurement device guidance is not used, and the measurement device guidance effect can be achieved by inputting a new theoretical trajectory.

3. The method for optimizing robot trajectory accuracy based on measurement equipment guidance according to claim 2, characterized in that: Segment the trajectory into straight and curved segments and build a small line segment model, including: First, convert the robot trajectory XYZ into a parametric equation about time T: ； Use cubic spline to smooth X, Y, and Z respectively to get the spline function of the actual trajectory: ； Find the first and second derivatives of a spline function: ； Curvature function of the robot trajectory: ； The parametric equation of the known space curve is: ； Then its tangent vector and second-order derivative are: ； ； Substituting in: ； ； Curvature function of the robot trajectory The component form is: ； According to the curvature of the curve, the moment when the actual trajectory approaches a straight line or an arc is distinguished, and the straight line segment and the arc segment are separated; The theoretical stationary point is extracted separately, and the corresponding theoretical trajectory stationary point is found on the actual trajectory according to the minimum distance and marked as a curve segment.

4. The method for optimizing robot trajectory accuracy based on measurement equipment guidance according to claim 3, characterized in that: Perform hybrid filtering on the extracted trajectories, including: A hybrid filtering method of low-pass filtering and median filtering is adopted for the control trajectory. When extracting the trajectory, the points in the actual trajectory and the control trajectory are one-to-one corresponding. The small line segment models segmented by the actual trajectory are mapped one-to-one to the control trajectory. The segmented straight line part is fitted by least squares interpolation to obtain the coordinates of the straight line segment endpoints.

5. The method for optimizing robot trajectory accuracy based on measurement equipment guidance according to claim 4 is characterized in that ,The time delay information of the new theoretical trajectory includes: The time delay information when the robot applies the new theoretical trajectory is represented by the trajectory response delay coefficient; The trajectory response delay coefficient is obtained by determining a set of visual servo frequencies, applying the new theoretical trajectory to different visual servo frequencies, and determining the time interval between the time when the robot controller issues a command and the time when the robot actually reaches the new theoretical trajectory point as the response time at different visual servo frequencies. The robot's response time is then determined several times at different visual servo frequencies, and the average response delay of the robot is calculated. The average response delay of the robot under multiple repetitions at different visual servo frequencies is marked as: ; Wherein, f is the number of different visual servo frequencies, n is the number of repetitions, f=1, 2, 3, ..., F, F is a positive integer, n=1, 2, 3, ..., N, N is a positive integer; Based on the number of trajectory points that the robot can actually process and track and the average response time of the robot, a timeliness decreasing effect model is constructed. Based on the average response delay of robots with different frequencies under the visual servo frequency set, different average response delays are scored and used as trajectory delay scores. The trajectory delay scores are used to optimize the model parameters of the timeliness decreasing effect model based on the nonlinear least squares method to obtain a calculation model for the trajectory response delay coefficient. The objective function used to optimize the timeliness decreasing effect model is: ; Where m is the number of the trajectory delay score, m=1, 2, 3, ..., M, M is a positive integer, is the trajectory delay score, A is the maximum upper limit of the trajectory delay score, and YC is the robot's forward tracking ability for the new theoretical trajectory; The calculation model of the trajectory response delay coefficient is: ;in, is the trajectory response delay coefficient.

6. The method for optimizing robot trajectory accuracy based on measurement equipment guidance according to claim 5, characterized in that: Deviation information from the new theoretical trajectory, including: The deviation information when the robot applies the new theoretical trajectory is represented by the trajectory deviation evaluation coefficient; The logic for obtaining the trajectory deviation evaluation coefficient is as follows: based on the new theoretical trajectory, obtain the marker points existing in the new theoretical trajectory, the marker points including the stationary point, the acceleration mutation point and the attitude switching point, and score the different marker points based on the marker points existing in the new theoretical trajectory to obtain the importance score value of the marker point; The position deviation between the actual trajectory of the robot and the new theoretical trajectory is obtained. The position deviation between the actual trajectory and the new theoretical trajectory at different marking points and the importance score of different marking points are used as input features, and the trajectory deviation evaluation coefficient is used as the output feature to construct a logistic regression model. The calculation formula of the trajectory deviation evaluation coefficient is: ； in, is the trajectory deviation evaluation coefficient, 、 、 、……、 is the position deviation between the actual trajectory and the new theoretical trajectory at different marking points, 、 、 、……、 is the importance score of different marked points, and e is the base.

7. The method for optimizing robot trajectory accuracy based on measurement equipment guidance according to claim 6, characterized in that: Evaluate new theoretical trajectories, including: The delay information and deviation information of the new theoretical trajectory in the application are comprehensively analyzed. The new theoretical trajectory evaluation model is constructed by calculating the weighted sum of the trajectory response delay coefficient and the trajectory deviation evaluation coefficient, and the new theoretical trajectory evaluation coefficient is generated. The calculation formula of the new theoretical trajectory evaluation coefficient is: ;in, is the new theoretical trajectory evaluation coefficient, 、 are the proportional coefficients of trajectory response delay coefficient and trajectory deviation evaluation coefficient respectively, 、 are both greater than 0; A new theoretical trajectory evaluation coefficient threshold is set, and the new theoretical trajectory evaluation coefficient of the new theoretical trajectory is compared with the new theoretical trajectory evaluation coefficient threshold. If the new theoretical trajectory evaluation coefficient is less than the new theoretical trajectory evaluation coefficient threshold, a warning signal is generated; if the new theoretical trajectory evaluation coefficient is greater than the new theoretical trajectory evaluation coefficient threshold, no warning signal is generated.

Images