A robot contour error fairing compensation method and system based on polynomial splines

By using a polynomial spline smoothing compensation method, the problem of trajectory discontinuity in robot contour error compensation is solved, thereby improving the smoothness and stability of the trajectory and enhancing the execution performance and control robustness of the robot system.

CN121132625BActive Publication Date: 2026-11-10HUAZHONG UNIV OF SCI & TECH
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202511133445.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2025-08-13
Publication Date
2026-11-10
Estimated Expiration
2045-08-13

AI Technical Summary

Technical Problem

Existing robot contour error compensation methods are insufficient in handling trajectory smoothness, resulting in discontinuous compensation trajectories, causing sudden changes in robot speed and acceleration, and affecting trajectory accuracy and stability.

Method used

A smoothing compensation method based on polynomial splines is adopted. By establishing a robot end-effector contour error estimation model, the non-smooth kinematic intervals are identified, and the trajectory is smoothed using seventh-order polynomial splines. The polynomial parameters are optimized to achieve the continuity and smoothness of the trajectory.

Benefits of technology

It effectively suppresses abrupt changes in velocity and acceleration caused by non-smooth compensation, improves the continuity and smoothness of the compensation trajectory, and enhances the execution performance and control robustness of the robot system in complex trajectory tracking scenarios.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121132625B_ABST
    Figure CN121132625B_ABST
Patent Text Reader

Abstract

The present application belongs to the field of robot technology, and discloses a robot contour error fairing compensation method based on a polynomial spline, which comprises: establishing a robot end contour error estimation model to realize position and direction contour error estimation of the robot; calculating a robot direct compensation trajectory under a direct contour compensation method through position and direction contour error of the robot; detecting a kinematics unsmooth interval in the robot direct compensation trajectory by setting an end trajectory acceleration threshold; and smoothing the trajectory in the kinematics unsmooth interval in the robot direct compensation trajectory by designing a polynomial spline. The present application can accurately locate the kinematics fluctuation interval in the trajectory through an acceleration mutation identification mechanism, and use a high-order polynomial to fit and process the trajectory in the interval, thereby effectively suppressing the speed and acceleration mutation caused by non-smooth compensation, and improving the continuity and smoothness of the compensation trajectory.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to, but is not limited to, the field of robotics, and particularly relates to a method and system for smoothing robot contour errors based on polynomial splines. Background Technology

[0002] With the rapid development of industrial automation and intelligent manufacturing, industrial robots are widely used in high-precision scenarios such as welding, grinding, polishing, and assembly. In these tasks, the contour accuracy of the robot trajectory has become one of the key indicators for measuring operational quality. Contour error, namely the deviation between the actual trajectory and the expected trajectory of the robot's end effector, not only affects the dimensional accuracy and surface quality of the product, but may also cause mechanical impact and structural fatigue, thus becoming a research hotspot in the field of robot control.

[0003] To improve contour accuracy, contour error compensation technology has emerged. Its basic principle is to correct the original trajectory during trajectory planning or control based on predicted or measured error information, thereby bringing the robot actuator closer to the target path. Traditional contour error compensation methods mostly employ a direct compensation strategy, directly superimposing the predicted error value onto the original trajectory to correct the execution trajectory. This method is simple to implement and has a rapid response, hence its widespread adoption.

[0004] However, in practical applications, robot systems exhibit complex dynamic characteristics, especially under conditions such as low-speed reversal and load fluctuations, where tracking errors often manifest as discontinuous or non-smooth variations. If these non-smooth error values ​​are directly applied to the trajectory compensation without smoothing the compensation trajectory, it will cause drastic changes in the compensation trajectory at error abrupt change points, leading to sudden changes in robot speed and acceleration. This kinematic discontinuity not only causes joint jitter and actuator vibration but may also further amplify new contour errors, negating the original compensation effect.

[0005] Therefore, in order to improve the actual effect of contour error compensation, it is urgent to study a compensation method that combines trajectory smoothing processing. This method can effectively correct errors while avoiding kinematic degradation caused by non-smooth compensation values, thereby achieving more stable and higher precision contour control. Summary of the Invention

[0006] To address the aforementioned deficiencies or improvement needs of existing technologies, this invention provides a robot contour error smoothing compensation method based on polynomial splines. This method primarily addresses the problem of abrupt changes in the compensation trajectory caused by the discontinuous and non-smooth characteristics of robot tracking errors in direct compensation strategies. In particular, it addresses the practical application needs of adverse kinematic effects such as velocity fluctuations, acceleration abrupt changes, and robot joint jitter that are easily triggered at error abrupt changes, thereby further affecting the contour error compensation effect. A robot contour error smoothing compensation method based on polynomial splines has been studied.

[0007] This invention is implemented as follows: a robot contour error smoothing compensation method based on polynomial splines, the method comprising:

[0008] S1: Establish a robot end-effector contour error estimation model, and use the robot end-effector expected pose and predicted pose to estimate the robot's position and orientation contour errors;

[0009] S2: The robot's direct compensation trajectory under the direct contour compensation method is obtained by calculating the robot's position and orientation contour errors;

[0010] S3: Detect kinematically unsmooth regions in the robot's direct compensation trajectory by setting an end-effector trajectory acceleration threshold;

[0011] S4: Smooth the trajectory in the kinematically unsmooth interval of the robot's direct compensation trajectory by designing polynomial splines, and optimize the polynomial parameters with robot contour trajectory deviation and kinematic deviation as optimization objectives.

[0012] Furthermore, the robot's end-effector pose in S1 consists of a position vector and an attitude angle, expressed as follows: ,in Indicates the position of the robot's end effector. This indicates the attitude angles defined using ZYZ Euler angles;

[0013] To uniformly describe the position and orientation information of the end effector, the robot's end effector pose is represented as a homogeneous transformation matrix. Its form is as follows:

[0014] (1)

[0015] in The rotation matrix is ​​used to represent the attitude of the end effector; the rotation matrix Constructed based on ZYZ Euler angles, its specific expression is as follows:

[0016] (2)

[0017] Furthermore, the desired pose trajectory of the robot's end effector in S1 is... The desired pose of the robot's end effector is ;in, and These are the expected position vector of the end effector of the i-th robot and the expected attitude angle of the end effector of the i-th robot. and It is the predicted end-effector position vector of the i-th robot and the predicted end-effector attitude angle of the i-th robot; the end-effector contour error is defined as... ,in, It is a positional profile error. It is the orientation profile error;

[0018] The i-th robot position contour error The estimation method is as follows:

[0019] (3)

[0020] in, It is the position on the expected position trajectory of the robot's end effector that is closest to the predicted position vector of the i-th robot's end effector;

[0021] The i-th robot orientation profile error The estimation method is as follows:

[0022] (4)

[0023] in, It is the predicted direction vector of the z-axis of the i-th robot end effector. It is the distance on the trajectory of the expected direction vector of the i-th robot end effector along the z-axis. The nearest unit vector can be obtained using the following method;

[0024] (5)

[0025] in Represents a spherical linear interpolation function. The interpolation coefficients range from 0 to 1; the spherical linear interpolation function The definition is as follows:

[0026] (6)

[0027] in: and The unit vectors at the start and end points of the interpolation represent the directional information, such as the normal direction or the attitude axis, on a certain segment of the robot's end-effector trajectory. Let be the angle between two unit vectors, and it is calculated as follows:

[0028] (7).

[0029] Furthermore, the robot direct compensation trajectory obtained by calculating the robot's position and orientation contour errors in step S2 is as follows: ;in Let i be the robot end-effector position vector after direct compensation. Let be the robot end effector angle after the i-th direct compensation.

[0030] in, The calculation can be performed as follows:

[0031] (8)

[0032] Let be the i-th directly compensated robot end effector direction vector, which can be calculated as follows:

[0033] (9)

[0034] Let be the robot end effector angle after the i-th direct compensation, where , and They are respectively The first, second, and third components; the calculation methods for each component are as follows:

[0035] (10)

[0036] in, , and They are respectively The first, second, and third components for The third component.

[0037] Furthermore, the acceleration threshold mentioned in S3 is Filter out time points in the entire trajectory that meet the following conditions:

[0038] (11)

[0039] And their indices are grouped into a set, denoted as . For any index in the set Construct a structure with this point as the center and a length of Symmetrical window interval , where the window radius It is a fixed value;

[0040] This window can cover all fluctuation regions near the acceleration abrupt change point; if multiple windows overlap or are adjacent, they are merged into a continuous fluctuation interval, forming a set of non-overlapping fluctuation time periods, denoted as . in, and They represent the first The starting and ending indices of each fluctuation range;

[0041] During the trajectory compensation phase, the positions are respectively... and posture By performing the above fluctuation range extraction methods respectively, the corresponding set of position fluctuation ranges is finally obtained. and attitude fluctuation range set .

[0042] Furthermore, the polynomial spline mentioned in S4 is a 7th-order polynomial spline, and its expression is as follows:

[0043] (12)

[0044] in, Indicates time The trajectory position below, arrive These are the polynomial coefficients;

[0045] In the polynomial spline fitting process, the optimization objective is to minimize the trajectory deviation and velocity fluctuation, and its objective function is defined as:

[0046] (13)

[0047] in, This indicates the position of the fitted trajectory. Indicates its speed, The location points of the original trajectory, , For speed fluctuation weighting coefficient, This is the coefficient vector of the fitted curve;

[0048] To ensure the continuity and smoothness of the fitted curve at both ends of the interval, the following position and velocity constraints are imposed at the start and end points of the interval:

[0049] (14)

[0050] Meanwhile, to reduce the contour error introduced during the fitting process, the error between all fitted points and their original points is evaluated, and a preset threshold is set for each point where the error exceeds the threshold. Among the points, select the point with the largest error. And it is introduced as a fitting constraint point, with the constraint condition being: By combining the above objective function and constraints, the coefficient parameters of the final fitted curve can be obtained; the solution process is as follows:

[0051] set up and The basis function vector is:

[0052] (15)

[0053] The positions and velocities at all sampling times are stacked into a position stacking matrix. Velocity stacking matrix With the target trajectory vector , means as follows:

[0054] (16)

[0055] Therefore, the original fitting optimization problem can be transformed into the following constrained optimization problem:

[0056] (17)

[0057] in, Constraint matrix for constraining objective function Constraint vectors and difference matrix The definition is as follows:

[0058] (18)

[0059] For ease of derivation, the objective function The rewritten standard quadratic form is as follows:

[0060] (19)

[0061] To find the optimal solution Construct the following Lagrange function:

[0062] (20)

[0063] in, These are Lagrange multiplier vectors;

[0064] To solve the optimization problem with equality constraints, the first-order necessary condition, namely the Karush-Kuhn-Tucker (KKT) condition, must be satisfied:

[0065] (twenty one)

[0066] The optimal solution for the polynomial coefficients can then be obtained. The optimal solution of the Lagrange multiplier vector :

[0067] (twenty two)

[0068] in, This indicates the Moore–Penrose pseudo-inverse.

[0069] Another objective of this invention is to provide a robot contour error smoothing compensation system based on polynomial splines, which is based on the aforementioned robot contour error smoothing compensation method based on polynomial splines. This system specifically includes:

[0070] The model building module is used to build a robot end-effector contour error estimation model, and to estimate the robot's position and orientation contour errors by using the robot end-effector's desired pose and predicted pose.

[0071] The trajectory calculation module is connected to the model building module. It calculates the robot's direct compensation trajectory under the direct contour compensation method by calculating the robot's position and orientation contour errors.

[0072] The interval setting module, connected to the trajectory calculation module, detects kinematically unsmooth intervals in the robot's direct compensation trajectory by setting an end-point trajectory acceleration threshold.

[0073] The parameter optimization module, connected to the interval setting module, smooths the trajectory in the kinematically unsmooth interval of the robot's direct compensation trajectory by designing polynomial splines, and optimizes the polynomial parameters with robot contour trajectory deviation and kinematic deviation as optimization targets.

[0074] Another object of the present invention is to provide a computer device, the computer device including a memory and a processor, the memory storing a computer program, the computer program being executed by the processor causing the processor to perform the steps of the robot contour error smoothing compensation method based on polynomial splines.

[0075] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the robot contour error smoothing compensation method based on polynomial splines.

[0076] Another objective of this invention is to provide an information data processing terminal, which is used to implement the robot contour error smoothing compensation system based on polynomial splines.

[0077] Based on the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solution to be protected by this invention are as follows:

[0078] 1. The robot contour error smoothing compensation method of the present invention can accurately locate the kinematic fluctuation range in the trajectory through the acceleration change identification mechanism, and perform fitting processing with a high-order polynomial in the range, thereby effectively suppressing the velocity and acceleration changes caused by non-smooth compensation and improving the continuity and smoothness of the compensation trajectory.

[0079] 2. The method described in this invention constructs a joint optimization objective function that includes trajectory deviation and velocity fluctuation, which ensures compensation accuracy while taking into account the dynamic stability of the compensation trajectory, thereby significantly enhancing the execution performance and control robustness of the robot system in complex trajectory tracking scenarios.

[0080] 3. The fitting method described in this invention introduces boundary point position and velocity constraints and integrates an adaptive error point interpolation mechanism, which can correct points with large fitting errors, further improving the local accuracy and overall consistency of the smooth compensation curve.

[0081] 4. The optimization problem is constructed as a standard quadratic constrained minimization problem, which is solved using the Lagrangian function and KKT conditions. It has the characteristics of clear structure and efficient solution, making it easy to integrate into the online or offline compensation module of the robot control system. It has good engineering feasibility and scalability.

[0082] As supplementary evidence of the inventive step for the claims of this invention, it is also reflected in the following important aspects:

[0083] (1) The expected benefits and commercial value of the technical solution of this invention after transformation are as follows:

[0084] This invention can be widely applied to high-precision trajectory control of industrial robots, and is particularly suitable for processes with extremely high requirements for trajectory continuity and smoothness, such as CNC machining, automatic polishing, welding, and trajectory gluing. By dynamically compensating for end-effector contour errors and introducing a trajectory smoothing mechanism, it not only improves the quality consistency of robot-processed surfaces but also significantly reduces mechanical shock and energy consumption caused by discontinuous compensation values, extending the service life of critical components. This method has good system compatibility, is easy to integrate into existing control frameworks, and can bring significant economic benefits and commercial value to fields such as intelligent manufacturing and high-end equipment.

[0085] (2) The technical solution of this invention fills a technical gap in the industry both domestically and internationally:

[0086] Existing contour error compensation methods mostly focus on error prediction and correction, but lack systematic research on the smoothness control of the compensation trajectory, especially lacking modeling and suppression methods for new kinematic fluctuations introduced by discontinuous trajectory compensation values. This invention, based on the realization of robot end-effector error estimation and compensation, further proposes an acceleration abrupt change detection mechanism and combines polynomial spline curves to smooth the local trajectory of kinematic fluctuations generated after trajectory compensation. It pioneeringly achieves an integrated design of "error compensation + trajectory smoothing," filling the research gap in robot trajectory smoothing compensation.

[0087] (3) The technical solution of the present invention solves a technical problem that people have long wanted to solve but have never been able to solve successfully:

[0088] In industrial practice, the discontinuity of the trajectory during error compensation often leads to abrupt changes in joint velocity and acceleration, resulting in mechanical system jitter, vibration, and even damage. Traditional methods lack dynamic identification and active optimization mechanisms for trajectory smoothness, making it difficult to achieve a balance between high-precision compensation and kinematic stability. This invention constructs a trajectory acceleration threshold detection mechanism to automatically identify non-smooth regions in the compensation trajectory and uses high-order polynomial splines to achieve continuous processing of local trajectories, thereby effectively solving the long-standing industry problem of "motion shock caused by high-precision compensation."

[0089] (4) The technical solution of the present invention overcomes technical bias:

[0090] Existing technologies generally consider error compensation to focus on improving positional accuracy, paying insufficient attention to the kinematic continuity of the compensated trajectory, exhibiting a technical bias of "emphasizing accuracy while neglecting smoothness." This invention breaks this mindset, proposing a collaborative design approach for trajectory compensation and smoothing, and forming a complete technical closed loop from end-point error and compensated trajectory generation to the identification and reconstruction of non-smooth segments. By introducing polynomial splines, a balance between compensation accuracy and motion continuity is achieved, overcoming the limitation of traditional error compensation models that cannot simultaneously consider trajectory quality, demonstrating significant innovation and progress. Attached Figure Description

[0091] Figure 1 This is a flowchart of a robot contour error smoothing compensation method based on polynomial splines provided in an embodiment of the present invention;

[0092] Figure 2 This is a structural diagram of a robot contour error smoothing compensation system based on polynomial splines provided in an embodiment of the present invention;

[0093] In the diagram: 1. Model building module; 2. Trajectory calculation module; 3. Interval setting module; 4. Parameter optimization module.

[0094] Figure 3These are contour error compensation effect diagrams of the proposed method and the comparison method provided in the embodiments of the present invention;

[0095] Figure 4 The diagram shows the motion speed compensation effect of the proposed method and the comparison method provided in the embodiments of the present invention. Detailed Implementation

[0096] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0097] In high-precision manufacturing and assembly operations, trajectory tracking errors of industrial robots significantly affect the fitting accuracy of the end effector to the target contour. Especially under complex paths and high-speed operation conditions, due to the robot's dynamic response lag, control delay, and joint precision limitations, pose deviations often concentrate at non-uniform nodes such as path turning points and curvature changes. Traditional contour error compensation methods are mostly based on feedforward correction or single-step correction, ignoring the overall motion continuity and acceleration smoothness of the trajectory. This often introduces new dynamic irregularities while reducing errors, thereby triggering over-limit alarms, mechanical vibrations, or even execution failures.

[0098] To address the aforementioned technical challenges, this invention constructs a contour error estimation model for the end-effector pose from a systematic perspective. It considers the full-state differences between the desired trajectory and the predicted actual trajectory, and analyzes the position and orientation errors separately. This model enables the accurate acquisition of the spatial distribution and variation trend of contour errors without the need for external sensors, leveraging the robot's internal sensing and model prediction capabilities, thus providing high-fidelity error information support for subsequent compensation calculations.

[0099] Based on this, a direct contour compensation trajectory generation method is constructed, which back-maps the estimated error amount onto the desired trajectory to correct its geometric path. This step is not a simple point-to-point offset, but rather uses the robot's forward and inverse kinematics model to perform pose-level correction of the error source, ensuring that the compensated trajectory remains within the reachability and constraint torque range of the robotic arm. Although this method can effectively improve contour fitting accuracy, the abrupt change in peak acceleration caused by direct compensation becomes a key limiting factor affecting its industrial applicability.

[0100] To further improve the dynamic executability of the trajectory, this invention introduces an acceleration threshold screening mechanism. This mechanism performs point-by-point kinematic analysis on the directly compensated trajectory, marking segments where acceleration changes exceed a set threshold as "unsmooth intervals." Based on second-order derivative analysis, this screening mechanism can work in conjunction with the dynamic constraint model of the robot controller, thus balancing error correction effectiveness with execution smoothness. It is an engineering solution with practical deployment feasibility.

[0101] For the identified kinematic degradation regions, a polynomial spline interpolation structure is designed for trajectory smoothing. This spline not only provides a set of continuously differentiable trajectory alternatives to the original path, but also constructs an optimization objective by embedding contour deviation and kinematic cost function, jointly constraining trajectory deviation and control law changes, and employing a multi-objective optimization algorithm to adjust the spline parameters. This design ensures both high fitting accuracy of the terminal trajectory to the original path and guarantees the continuity and controllability of the trajectory velocity and acceleration curves throughout the entire trajectory.

[0102] Ultimately, the smooth compensation trajectory is executed via interpolation by the robot trajectory controller, allowing for seamless integration with existing servo systems and enabling dynamic performance reconstruction of traditional compensation paths. This method significantly improves the robot's execution stability and product consistency in tasks such as high-speed bonding welding, variable curvature spraying, and collaborative grasping without introducing external hardware, demonstrating significant industrial application value.

[0103] like Figure 1 As shown, this embodiment of the invention provides a robot contour error smoothing compensation method based on polynomial splines, the method comprising:

[0104] S1: Establish a robot end-effector contour error estimation model, and use the robot end-effector expected pose and predicted pose to estimate the robot's position and orientation contour errors;

[0105] S2: The robot's direct compensation trajectory under the direct contour compensation method is obtained by calculating the robot's position and orientation contour errors;

[0106] S3: Detect kinematically unsmooth regions in the robot's direct compensation trajectory by setting an end-effector trajectory acceleration threshold;

[0107] S4: Smooth the trajectory in the kinematically unsmooth interval of the robot's direct compensation trajectory by designing polynomial splines, and optimize the polynomial parameters with robot contour trajectory deviation and kinematic deviation as optimization objectives.

[0108] The robot's end-effector pose in S1 consists of a position vector and an attitude angle, expressed as follows: ,in Indicates the position of the robot's end effector. This indicates the attitude angles defined using ZYZ Euler angles;

[0109] To uniformly describe the position and orientation information of the end effector, the robot's end effector pose is represented as a homogeneous transformation matrix. Its form is as follows:

[0110] (1)

[0111] in The rotation matrix is ​​used to represent the attitude of the end effector; the rotation matrix Constructed based on ZYZ Euler angles, its specific expression is as follows:

[0112] (2)

[0113] The desired pose trajectory of the robot end effector in S1 is: The desired pose of the robot's end effector is ;in, and These are the expected position vector of the end effector of the i-th robot and the expected attitude angle of the end effector of the i-th robot. and It is the predicted end-effector position vector of the i-th robot and the predicted end-effector attitude angle of the i-th robot; the end-effector contour error is defined as... ,in, It is a positional profile error. It is the orientation profile error;

[0114] The i-th robot position contour error The estimation method is as follows:

[0115] (3)

[0116] in, It is the position on the expected position trajectory of the robot's end effector that is closest to the predicted position vector of the i-th robot's end effector;

[0117] The i-th robot orientation profile error The estimation method is as follows:

[0118] (4)

[0119] in, It is the predicted direction vector of the z-axis of the i-th robot end effector. It is the distance on the trajectory of the expected direction vector of the i-th robot end effector along the z-axis. The nearest unit vector can be obtained using the following method;

[0120] (5)

[0121] in Represents a spherical linear interpolation function. The interpolation coefficients range from 0 to 1; the spherical linear interpolation function The definition is as follows:

[0122] (6)

[0123] in: and The unit vectors at the start and end points of the interpolation represent the directional information, such as the normal direction or the attitude axis, on a certain segment of the robot's end-effector trajectory. Let be the angle between two unit vectors, and it is calculated as follows:

[0124] (7).

[0125] The robot direct compensation trajectory obtained by calculating the robot's position and orientation contour errors in S2 is as follows: ;in Let i be the robot end-effector position vector after direct compensation. Let be the robot end effector angle after the i-th direct compensation.

[0126] in, The calculation can be performed as follows:

[0127] (8)

[0128] Let be the i-th directly compensated robot end effector direction vector, which can be calculated as follows:

[0129] (9)

[0130] Let be the robot end effector angle after the i-th direct compensation, where , and They are respectively The first, second, and third components; the calculation methods for each component are as follows:

[0131] (10)

[0132] in, , and They are respectively The first, second, and third components for The third component.

[0133] The acceleration threshold mentioned in S3 is Filter out time points in the entire trajectory that meet the following conditions:

[0134] (11)

[0135] And their indices are grouped into a set, denoted as . For any index in the set Construct a structure with this point as the center and a length of Symmetrical window interval , where the window radius It is a fixed value;

[0136] This window can cover all fluctuation regions near the acceleration abrupt change point; if multiple windows overlap or are adjacent, they are merged into a continuous fluctuation interval, forming a set of non-overlapping fluctuation time periods, denoted as . in, and They represent the first The starting and ending indices of each fluctuation range;

[0137] During the trajectory compensation phase, the positions are respectively... and posture By performing the above fluctuation range extraction methods respectively, the corresponding set of position fluctuation ranges is finally obtained. and attitude fluctuation range set .

[0138] The polynomial spline mentioned in S4 is a 7th-order polynomial spline, and its expression is as follows:

[0139] (12)

[0140] in, Indicates time The trajectory position below, arrive These are the polynomial coefficients;

[0141] In the polynomial spline fitting process, the optimization objective is to minimize the trajectory deviation and velocity fluctuation, and its objective function is defined as:

[0142] (13)

[0143] in, This indicates the position of the fitted trajectory. Indicates its speed, The location points of the original trajectory, , For speed fluctuation weighting coefficient, This is the coefficient vector of the fitted curve;

[0144] To ensure the continuity and smoothness of the fitted curve at both ends of the interval, the following position and velocity constraints are imposed at the start and end points of the interval:

[0145] (14)

[0146] Meanwhile, to reduce the contour error introduced during the fitting process, the error between all fitted points and their original points is evaluated, and a preset threshold is set for each point where the error exceeds the threshold. Among the points, select the point with the largest error. And it is introduced as a fitting constraint point, with the constraint condition being: By combining the above objective function and constraints, the coefficient parameters of the final fitted curve can be obtained; the solution process is as follows:

[0147] set up and The basis function vector is:

[0148] (15)

[0149] The positions and velocities at all sampling times are stacked into a position stacking matrix. Velocity stacking matrix With the target trajectory vector , means as follows:

[0150] (16)

[0151] Therefore, the original fitting optimization problem can be transformed into the following constrained optimization problem:

[0152] (17)

[0153] in, Constraint matrix for constraining objective function Constraint vectors and difference matrix The definition is as follows:

[0154] (18)

[0155] For ease of derivation, the objective function The rewritten standard quadratic form is as follows:

[0156] (19)

[0157] To find the optimal solution Construct the following Lagrange function:

[0158] (20)

[0159] in, These are Lagrange multiplier vectors;

[0160] To solve the optimization problem with equality constraints, the first-order necessary condition, namely the Karush-Kuhn-Tucker (KKT) condition, must be satisfied:

[0161] (twenty one)

[0162] The optimal solution for the polynomial coefficients can then be obtained. The optimal solution of the Lagrange multiplier vector :

[0163] (twenty two)

[0164] in, This indicates the Moore–Penrose pseudo-inverse.

[0165] like Figure 2 As shown, this embodiment of the invention provides a robot contour error smoothing compensation system based on polynomial splines, which is based on the aforementioned robot contour error smoothing compensation method based on polynomial splines. The system specifically includes:

[0166] Model building module 1 is used to build a robot end-effector contour error estimation model, and to estimate the position and orientation contour error of the robot by using the expected pose and predicted pose of the robot end-effector.

[0167] Trajectory calculation module 2 is connected to model building module 1. It calculates the robot's direct compensation trajectory under the direct contour compensation method by calculating the robot's position and orientation contour errors.

[0168] The interval setting module 3 is connected to the trajectory calculation module 2. It detects the kinematically unsmooth intervals in the robot's direct compensation trajectory by setting the end trajectory acceleration threshold.

[0169] The parameter optimization module 4 is connected to the interval setting module 3. It smooths the trajectory in the kinematically unsmooth interval of the robot's direct compensation trajectory by designing polynomial splines, and optimizes the polynomial parameters with the robot contour trajectory deviation and kinematic deviation as optimization targets.

[0170] This invention provides a computer device, which includes a memory and a processor. The memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the robot contour error smoothing compensation method based on polynomial splines.

[0171] This invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the robot contour error smoothing compensation method based on polynomial splines.

[0172] This invention provides an information data processing terminal, which is used to implement the robot contour error smoothing compensation system based on polynomial splines.

[0173] To verify the effectiveness of the proposed compensation method, two comparison methods were set up. Method 1 is the direct compensation method: first, based on the trained tracking error prediction model and robot kinematics model, the end-effector trajectory is predicted, and a robot compensation trajectory for direct compensation is generated. Method 2 is the proposed smoothing compensation method, which processes the compensation trajectory generated by Method 1 using the smoothing method proposed in this invention.

[0174] By using the planned robot trajectory as the test trajectory, the results of the end-effector position and attitude contour errors are shown in Table 1 and 2. Figure 4 As shown. Compared to the uncompensated case, Method 1 reduces the average position error / maximum error by approximately 73.5% and 48.7%, respectively, and the average attitude error / maximum error by approximately 69.1% and 5.6%, respectively; Method 2 reduces them by approximately 76.5%, 79.1%, 75.2%, and 61.4%, respectively. Compared to Method 1, Method 2 further reduces the average position error / maximum error by 11.1% / 59.4% and the average attitude error / maximum error by 20.0% / 59.1%.

[0175] Table 1

[0176]

[0177] Figure 3 The comparison of positional and orientation contour errors under different methods is presented. The blue curve represents the uncompensated state, green represents Method 1, and red represents Method 2. Experimental results show that both Method 1 and Method 2 significantly reduce contour errors, especially in complex motion ranges such as curve parameters of 0.2 to 0.4 and 0.6 to 0.8. Before compensation, there are obvious error peaks, which tend to stabilize after processing with Method 2, verifying the effectiveness of the proposed polynomial spline smoothing compensation strategy in error suppression.

[0178] Figure 3The magnified local area further verifies the superior continuity of Method 2 at the error boundary. In contrast, although Method 1 can reduce some errors, fluctuations still exist near the reversal and high curvature points; while Method 2 combines contour error estimation and spline trajectory reconstruction mechanisms, showing better error smoothing effects in multiple high dynamic ranges, especially in terms of robustness in reducing directional errors, and more stable control of error peaks.

[0179] Figure 4 The deviations between the actual velocities and reference velocities of the six joints were further compared, with green representing Method 1 and red representing Method 2. Overall, both methods fit the reference velocity curve well. However, Method 1 exhibits significant jitter in several key time intervals (e.g., 0-2s and 9-11s), indicating the introduction of non-smooth disturbances during dynamic compensation. In contrast, the velocity curve of Method 2 is closer to the ideal reference, demonstrating stronger trajectory dynamic executability after compensation.

[0180] Figure 4 The embedded magnified view clearly reveals the local velocity fluctuations that may be caused by non-smoothing compensation. Especially in higher-order joints (such as joints 4 to 6), Method 1 exhibits abrupt changes or non-physical transitions in the velocity trajectory, while Method 2 effectively suppresses such disturbances through spline smoothing, demonstrating its ability to optimize the compensation path under dynamic constraints. Therefore, the smoothing compensation mechanism proposed in this invention not only improves contour accuracy but also significantly enhances joint-level velocity stability, meeting the practical requirements of industrial robots for high-frequency motion continuity.

[0181] It should be noted that embodiments of the present invention can be implemented in hardware, software, or a combination of both. The hardware portion can be implemented using dedicated logic; the software portion can be stored in memory and executed by a suitable instruction execution system, such as a microprocessor or dedicated-design hardware. Those skilled in the art will understand that the above-described devices and methods can be implemented using computer-executable instructions and / or included in processor control code, for example, such code provided on a carrier medium such as a disk, CD, or DVD-ROM, a programmable memory such as read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The devices and modules of the present invention can be implemented by hardware circuitry such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field-programmable gate arrays, programmable logic devices, etc., or by software executed by various types of processors, or by a combination of the above-described hardware circuitry and software, such as firmware.

[0182] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications, equivalent substitutions, and improvements made by those skilled in the art within the scope of the technology disclosed in the present invention, and within the spirit and principles of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A robot contour error smoothing compensation method based on polynomial splines, comprising the following steps: Step S1: Establishing a robot end-effector contour error estimation model, estimating position contour error and orientation contour error through the robot end-effector's desired pose and predicted pose; Step S2: Calculating a direct compensation trajectory based on the position contour error and orientation contour error; Step S3: Detecting kinematically unsmooth intervals in the direct compensation trajectory based on a preset acceleration threshold; Step S4: Smoothing each kinematically unsmooth interval using polynomial splines, optimizing the polynomial parameters with contour trajectory deviation and velocity fluctuation as optimization objectives; The position contour error is the spatial distance between the predicted position of the robot end effector and the nearest point on the desired trajectory of the end effector. The orientation contour error is calculated by spherical linear interpolation between the predicted orientation vector and the desired orientation vector. The polynomial spline is a seventh-order polynomial. The polynomial coefficients are obtained by minimizing the objective function of the weighted sum of the position deviation and velocity fluctuation of the fitted trajectory, combined with the continuity constraints of the interval endpoint positions and velocity, and the fitting constraint of the maximum deviation point.

2. The method according to claim 1, wherein the robot end-effector pose in step S1 consists of a position vector and an attitude angle defined based on ZYZ Euler angles, wherein the position vector and attitude angle are represented by a homogeneous transformation matrix, wherein the rotation matrix is ​​constructed from ZYZ Euler angles.

3. The method according to claim 1, wherein the direct compensation trajectory in step S2 includes: after compensation The position vector is obtained by adding the position profile error to the end-point desired position vector; the compensated attitude is obtained by performing spherical linear interpolation on the predicted direction vector and the desired direction vector to obtain the compensated direction vector and converting it into attitude angles.

4. The method according to claim 1, wherein step S3 comprises: An index set is created by identifying time points where the acceleration amplitude exceeds a preset threshold. For each index, construct a symmetrical window centered at that time point; Merging overlapping or adjacent windows forms the final set of kinematically non-smooth intervals.

5. A robot contour error smoothing compensation system based on polynomial splines for implementing the method of any one of claims 1 to 4, comprising: The model building module is used to execute step S1; The trajectory calculation module is used to execute step S2; The interval detection module is used to execute step S3; The parameter optimization module is used to execute step S4.

6. A robot controller for implementing the robot contour error smoothing compensation method based on polynomial splines as described in any one of claims 1 to 4, the controller comprising a processor and a memory, the memory storing program instructions for the method, and the processor executing the program instructions.

7. A non-volatile computer-readable storage medium storing program instructions for implementing the method of any one of claims 1 to 4 when executed on a robot controller.

8. A robot, comprising a robot body and the robot contour error smoothing compensation system based on polynomial splines as described in claim 5.

Citation Information

Patent Citations

  • Robot gluing track error compensation method

    CN119760285A

  • Line width control and trajectory planning for robot guided inkjet deposition

    US20200406615A1