Multi-line laser based robot grinding and polishing processing adaptive compensation method

Abstract

The present application relates to a kind of robot grinding and polishing processing adaptive compensation method based on multi-line laser, it is related to robot positioning accuracy field.The technical scheme is realized workpiece online high-resolution measurement by multi-line laser scanning sensor, combined with multi-view calibration and data fusion dynamic update hand-eye parameter, utilize the hybrid error model of fusion MDH kinematics and machine learning to correct robot end pose in real time, guarantee grinding and polishing accuracy and consistency.MDH model is responsible for coarse positioning to ensure kinematic structure interpretability, machine learning model captures nonlinear dynamic error such as thermal expansion and wear, both realize adaptive real-time compensation of complex error sources.Error compensation accuracy is improved from millimeter level to sub-millimeter level, and the cost of line laser measurement system is only several ten thousand yuan, small in size, easy to install, significantly reduce the cost and installation difficulty compared with laser tracker.

Description

Technical Field

[0001] This invention relates to the field of robot positioning accuracy technology, and in particular to an adaptive compensation method for robot grinding and polishing based on multi-line laser. Background Technology

[0002] In the field of grinding and polishing, industrial robots are gradually replacing traditional grinding and polishing methods due to their significant advantages such as low cost, high flexibility, large workspace, and wide range of processing objects. However, current robotic grinding and polishing technology is mainly applied to scenarios with light loads and moderate precision requirements, making it difficult to achieve large-scale promotion in the field of precision machining. The root cause lies in the inherent defects of robots themselves, such as low absolute positioning accuracy and insufficient rigidity. During long-term continuous operation, the thermal expansion and zero-point drift of joint links caused by high loads, the transmission clearance caused by reducer wear, the flexible deflection of mechanical components caused by external forces, as well as sensor errors, environmental interference, and model deviations, all interact to cause the positioning error of the robot's end effector to accumulate continuously, seriously affecting the material removal accuracy, surface quality consistency, and trajectory planning rationality in the grinding and polishing of precision parts.

[0003] To address robot positioning errors, existing technologies typically employ laser trackers combined with end effector targets for pose error compensation. However, this method has significant limitations: firstly, the high cost of equipment procurement and maintenance significantly increases production costs; secondly, it demands stringent installation environmental requirements, necessitating a large, unobstructed working space and making equipment placement and relocation extremely inconvenient during production line layout adjustments or multi-station operations, failing to meet the flexible and ever-changing production demands of modern manufacturing. Therefore, there is an urgent need for an efficient, low-cost, and highly adaptable error compensation technology to achieve high-precision operation of industrial robots in the grinding and polishing of precision parts. Summary of the Invention

[0004] The purpose of this invention is to provide an adaptive compensation method for robotic grinding and polishing based on multi-line lasers, so as to solve the problems existing in the prior art.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: In a first aspect, the present invention provides an adaptive compensation method for robotic grinding and polishing based on multi-line laser, the method comprising: S1. Construct a multi-line laser measurement system to ensure that the robot end effector is within the coverage area of ​​the multi-line laser measurement system; S2. Scan the calibration block at the end of the robot using the multi-line laser measurement system to obtain the coordinates of feature points of the calibration block in different poses in the coordinate system of the multi-line laser measurement system. S3. Using the Horn or Umeyama absolute orientation algorithm, determine the rigid transformation matrix from the coordinate system of the multi-line laser measurement system to the coordinate system of the robot end effector; S4. Construct a robot kinematic model based on the MDH parameter method to describe the transformation relationship between the robot's base coordinate system and the robot's end effector coordinate system; S5. In response to different robot running time, joint load, joint angle, and ambient temperature conditions, obtain the robot's real end pose and theoretical pose, and construct an error dataset; S6. Construct a multilayer perceptron (MLP) residual neural network model, and define the input layer, output layer, and mean squared error loss function. S7. Use the error dataset to train the multilayer perceptron (MLP) residual neural network model to fit the nonlinear error of the robot end pose. S8. Based on the trained multilayer perceptron (MLP) residual neural network model, combined with the LM algorithm, determine the pose parameter increment, update the pose parameters based on the pose parameter increment, and generate the compensated robot end pose based on the updated pose parameters and the multilayer perceptron (MLP) residual neural network model. S9. Periodically acquire the robot's actual end-effector pose and theoretical pose. When the error between the two exceeds a preset threshold, repeat steps S5 to S7 to correct the multilayer perceptron (MLP) residual neural network model.

[0006] In one possible implementation, in step S1, the multi-line laser measurement system includes: Two line laser sensors are mounted on the worktable at a 90° angle.

[0007] In one possible implementation, the rigid transformation matrix from the multi-line laser measurement system coordinate system to the robot end effector coordinate system in step S3 includes: And satisfy , in, is the homogeneous transformation matrix from the coordinate system of the multi-line laser measurement system to the coordinate system of the robot end effector; It is a 3×3 rotation matrix; It is a 3×1 translation vector; These are the coordinates of the feature points in the robot's end effector coordinate system. The coordinates of the feature point in the coordinate system of the multi-line laser measurement system are given.

[0008] In one possible implementation, the adjacent coordinate system transformation matrix involved in the robot kinematics model in step S4 includes: ; in, This is the coordinate transformation matrix between two adjacent links; Link i Joint angles; Link i Linkage offset; Link i The joint twist angle; Link i The length of the connecting rod; Link i Y-axis parallelism parameter; c Represents the cosine function cos, s This represents the sine function sin.

[0009] In one possible implementation, the transformation matrix of the robot end-effector coordinate system {6} relative to the robot base coordinate system {0} involved in the robot kinematic model in step S4 includes: ; in, This is the direction vector of the X-axis in the robot's end-effector coordinate system in the robot's base coordinate system; This is the direction vector of the Y-axis in the robot's end-effector coordinate system in the robot's base coordinate system; This is the direction vector of the Z-axis in the robot's end-effector coordinate system in the robot's base coordinate system; This is the position vector of the origin of the robot's end effector coordinate system in the robot's base coordinate system.

[0010] In one possible implementation, in step S5, the error of the error dataset is defined as: ; in, This represents the residual between the robot's actual end-effector pose and its theoretical pose. This represents the robot's actual end-effector pose. This represents the theoretical end-effector pose of the robot. Link i Joint angles; T Ambient temperature; t Runtime; F For joint load.

[0011] In one possible implementation, in step S6: The input layer of the multilayer perceptron (MLP) residual neural network model includes: Theoretical end-effector pose of the robot ,link i joint angle Ambient temperatureT Running time t and joint load F ; The output layer of the multilayer perceptron (MLP) residual neural network model includes: The residual between the robot's actual end-effector pose and its theoretical pose ; The mean squared error loss function of the multilayer perceptron (MLP) residual neural network model includes: ; in, Loss This represents the mean squared error loss value; This represents the residual between the robot's actual end-effector pose and its theoretical pose. The residual value is predicted by the multilayer perceptron (MLP) residual neural network model. N This represents the number of training samples.

[0012] In one possible implementation, step S8, which involves determining the pose parameter increment based on the trained multilayer perceptron (MLP) residual neural network model and combining it with the LM algorithm, and updating the pose parameters based on the pose parameter increment, includes: ; ; ; in, The residual value is predicted by the multilayer perceptron (MLP) residual neural network model. For link offset; For joint torsion angle; The length of the link; p This is the current pose parameter vector; This represents the increment of the pose parameters; λ It is the damping factor; I It is the identity matrix; J It is a Jacobian matrix; T This is the matrix transpose. r It is the residual function; These are the updated pose parameters.

[0013] In one possible implementation, step S9 involves periodically acquiring the robot's actual end-effector pose and theoretical pose. When the error between the two exceeds a preset threshold, steps S5 to S7 are re-executed to correct the multilayer perceptron (MLP) residual neural network model, including: ; in, This represents the residual between the robot's actual end-effector pose and its theoretical pose. This is the preset error threshold coefficient.

[0014] In one possible implementation, the calibration block in step S2 is a trapezoidal sheet metal calibration block, which serves as the measurement standard. The positions of its feature points are known and easily identified by the multi-line laser measurement system.

[0015] Secondly, the present invention provides a computer device, which includes a processor and a memory. The memory stores at least one instruction, at least one program, code set or instruction set. The processor can load and execute at least one instruction, at least one program, code set or instruction set to realize the adaptive compensation method for robotic grinding and polishing based on multi-line laser provided above.

[0016] Thirdly, the present invention provides a computer-readable storage medium storing at least one instruction, at least one program, code set, or instruction set, wherein a processor can load and execute at least one instruction, at least one program, code set, or instruction set to implement the adaptive compensation method for robotic grinding and polishing based on multi-line laser as provided above.

[0017] Fourthly, the present invention provides a computer program product or computer program including computer program instructions stored in a computer-readable storage medium. A processor reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the adaptive compensation method for robotic grinding and polishing based on multi-line lasers as described above.

[0018] The beneficial effects of the technical solution provided by this invention include at least the following: This technical solution achieves online high-resolution measurement of the workpiece using a multi-line laser scanning sensor. Combined with multi-view calibration and data fusion, it dynamically updates hand-eye parameters and utilizes a hybrid error model integrating MDH kinematics and machine learning to correct the robot's end-effector pose in real time, ensuring grinding and polishing accuracy and consistency. Specifically, the MDH model handles coarse positioning to ensure the interpretability of the kinematic structure, while the machine learning model captures nonlinear dynamic errors such as thermal expansion and wear. Together, they achieve adaptive real-time compensation for complex error sources. Error compensation accuracy is improved from millimeter-level to sub-millimeter-level, and the line laser measurement system costs only tens of thousands of yuan. It is small, easy to install, and significantly reduces cost and installation difficulty compared to laser trackers. Attached Figure Description

[0019] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0020] Figure 1 The diagram shows a flowchart of an adaptive compensation method for robotic grinding and polishing based on multi-line laser, provided by an exemplary embodiment of the present invention.

[0021] Figure 2 A schematic diagram of the linkage coordinate system of a six-axis robot provided by an exemplary embodiment of the present invention is shown.

[0022] Figure 3 The diagram illustrates a multilayer perceptron (MLP) residual neural network model provided by an exemplary embodiment of the present invention.

[0023] Figure 4 The diagram shows a schematic representation of a computer device for performing an adaptive compensation method for robotic grinding and polishing based on multi-line laser, according to an exemplary embodiment of the present invention. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0026] Figure 1 The diagram illustrates a flowchart of an adaptive compensation method for robotic grinding and polishing based on multi-line laser technology, provided by an exemplary embodiment of the present invention. This adaptive compensation method for robotic grinding and polishing based on multi-line laser technology includes: Step S1: Construct a multi-line laser measurement system to ensure that the robot end effector is within the coverage area of ​​the multi-line laser measurement system.

[0027] In detail, the multi-line laser measurement system includes two line laser sensors mounted at a 90° angle on the worktable.

[0028] In this embodiment, a multi-line laser measurement system is used to construct a three-dimensional spatial coordinate perception capability. Two line laser sensors are mounted on the worktable at a 90° angle. Based on the principle of laser triangulation, the spatial point coordinates are calculated by utilizing the offset of the light spot formed by the laser beam projected onto the object surface and combining it with the sensor imaging principle. This orthogonal mounting method can acquire three-dimensional data of the object from different perspectives through the binocular parallax effect, and form a dense point cloud through multi-line laser scanning, covering the robot's end effector motion space. This configuration can effectively eliminate the blind zone of single-sensor measurement and improve the spatial point positioning accuracy by utilizing geometric projection relationships.

[0029] Step S2: Scan the calibration block at the end of the robot using a multi-line laser measurement system to obtain the coordinates of feature points of the calibration block in different poses in the coordinate system of the multi-line laser measurement system.

[0030] In detail, the calibration block is a trapezoidal sheet metal calibration block, which serves as the standard for measurement. The positions of its feature points are known and easily identified by the multi-line laser measurement system.

[0031] In this embodiment, when a multi-line laser beam is projected onto the surface of a trapezoidal sheet metal calibration block, its regular geometric features (such as edges and corners) will form a specific light spot pattern on the sensor imaging plane. Utilizing the principle of laser triangulation, two orthogonally mounted line laser sensors synchronously capture the light spot offset. Combined with the known three-dimensional coordinates of the calibration block's feature points (preset physical dimensions converted to spatial coordinates), a mapping relationship between the measurement system coordinate system and the calibration block coordinate system can be established. By driving the robot to change its end-effector pose and collecting multiple sets of data, the least squares method in statistics is used to fit the spatial transformation parameters, thereby eliminating random errors from a single measurement.

[0032] Step S3: Using the Horn or Umeyama absolute orientation algorithm, determine the rigid transformation matrix from the coordinate system of the multi-line laser measurement system to the coordinate system of the robot end effector.

[0033] In detail, the rigid transformation matrix from the coordinate system of the multi-line laser measurement system to the coordinate system of the robot end effector includes: And satisfy , in, is the homogeneous transformation matrix from the coordinate system of the multi-line laser measurement system to the coordinate system of the robot end effector; It is a 3×3 rotation matrix; It is a 3×1 translation vector; These are the coordinates of the feature point in the robot's end effector coordinate system. The coordinates of the feature point in the coordinate system of the multi-line laser measurement system are given.

[0034] In this embodiment, the optimal rotation and translation parameters between two coordinate systems are solved using the Horn or Umeyama algorithm, with the goal of minimizing the Euclidean distance error between feature point pairs. First, the centroids of the two sets of coordinate points are calculated, and the coordinate system is transformed into the centroid coordinate system. Then, the rotation matrix is ​​solved by constructing the covariance matrix. Then use translation vectors The coordinate transformation is completed. The Horn algorithm, based on quaternion optimization, solves the rotation matrix and is suitable for rigid transformations without scaling; the Umeyama algorithm is extended to affine transformations with scaling factors. Both achieve closed-form solutions through singular value decomposition (SVD) or eigenvalue decomposition. This process maps the 3D point cloud coordinates acquired by the multi-line laser measurement system to the robot's end effector coordinate system, providing a unified coordinate reference for subsequent kinematic modeling and error compensation.

[0035] Step S4: Construct a robot kinematic model based on the MDH parameter method, describing the transformation relationship between the robot's base coordinate system and its end effector coordinate system. (See also...) Figure 2 (A schematic diagram of the linkage coordinate system of a six-axis robot is shown.) In detail, the adjacent coordinate system transformation matrices involved in the robot kinematics model include: ; in, This is the coordinate transformation matrix between two adjacent links; Link i Joint angles; Link i Linkage offset; Link i The joint twist angle; Link i The length of the connecting rod; Link i Y-axis parallelism parameter; c Represents the cosine function cos, s This represents the sine function sin.

[0036] Furthermore, the transformation matrix of the robot's end-effector coordinate system {6} relative to the robot's base coordinate system {0} involved in the robot's kinematic model includes: ; ; in, This is the direction vector of the X-axis in the robot's end-effector coordinate system in the robot's base coordinate system; This is the direction vector of the Y-axis in the robot's end-effector coordinate system in the robot's base coordinate system; This is the direction vector of the Z-axis in the robot's end-effector coordinate system in the robot's base coordinate system; This is the position vector of the origin of the robot's end effector coordinate system in the robot's base coordinate system.

[0037] In this embodiment, based on the Denavit-Hartenberg (DH) kinematic modeling theory, a robot kinematic model is constructed by defining link coordinate systems and joint parameters. Traditional DH models use four parameters—joint angles, link offsets, link lengths, and joint torsion angles—to describe the transformation between adjacent coordinate systems. The ModifiedDH (MDH) model of this application adds a Y-axis parallelism parameter, achieving homogeneous transformation between adjacent link coordinate systems through a matrix. cθ and sθ These are the cosine and sine values ​​of the joint angle, respectively. Furthermore, the transformation matrix of the robot's end-effector coordinate system {6} relative to the base coordinate system {0} is obtained by concatenating the transformation matrices of each link, and its direction vector... Describes the end effector's attitude and position vector. By determining the spatial coordinates of the end effector, the forward kinematics solution from joint angle to end effector pose can be achieved.

[0038] Step S5: In response to different robot running time, joint load, joint angle, and ambient temperature conditions, obtain the robot's actual end-effector pose and theoretical pose, and construct an error dataset.

[0039] In detail, the error of the error dataset is defined as follows: ; in, This represents the residual between the robot's actual end-effector pose and its theoretical pose. This represents the robot's actual end-effector pose. This represents the theoretical end-effector pose of the robot. Link i Joint angles; T The ambient temperature; t Runtime; F For joint load.

[0040] In this embodiment, an error mapping model is constructed by collecting robot pose data under dynamic working conditions. During robot operation, factors such as mechanical deformation caused by joint angle changes, transmission backlash due to load, thermal expansion caused by temperature fluctuations, and wear accumulated over time can cause the end-effector pose to deviate from the theoretical value. This step uses sensors to acquire the robot's actual end-effector pose in real time under different joint angles, load states, ambient temperatures, and operating times. Simultaneously, the theoretical pose is calculated based on a kinematic model, and the difference between the two constitutes an error dataset. The error is defined as the residual between the actual pose and the theoretical pose, and this dataset encompasses the error characteristics under multi-physics coupling.

[0041] Step S6: Construct a multilayer perceptron (MLP) residual neural network model, defining the input layer, output layer, and mean squared error loss function. (See also...) Figure 3 (A schematic diagram of a multilayer perceptron (MLP) residual neural network model is shown.) In detail, the input layer of the multilayer perceptron (MLP) residual neural network model includes: Theoretical end-effector pose of the robot ,link i joint angle Ambient temperature T Running time t and joint load F ; The output layer of the multilayer perceptron (MLP) residual neural network model includes: The residual between the robot's actual end-effector pose and its theoretical pose ; The mean squared error loss function of the multilayer perceptron (MLP) residual neural network model includes: ; in, Loss This represents the mean squared error loss value; This represents the residual between the robot's actual end-effector pose and its theoretical pose. The residual value is predicted by the multilayer perceptron (MLP) residual neural network model. N This represents the number of training samples.

[0042] In this embodiment, a dynamic prediction model for robot pose error is constructed using a multilayer perceptron based on the nonlinear mapping capability of neural networks. This model uses theoretical end-effector pose, joint angles, ambient temperature, runtime, and joint load as input layer variables. These parameters encompass error causes resulting from the coupling of theoretical robot kinematics with multi-physics fields. The output layer directly corresponds to the residual between the actual and theoretical poses, mapping complex nonlinear errors into calculable corrections through a residual learning mechanism. The mean squared error loss function minimizes the sum of squares between the predicted and actual residuals, and utilizes a gradient descent algorithm to optimize network weights, enabling the model to adaptively fit the nonlinear distribution characteristics of dynamic errors such as thermal expansion and wear gaps.

[0043] Step S7: Use the error dataset to train the multilayer perceptron (MLP) residual neural network model to fit the nonlinear error of the robot's end-effector pose.

[0044] In this embodiment, the MLP neural network is iteratively optimized using an error dataset. This process takes the error dataset constructed in step S5 as input, where theoretical pose, joint angles, and other parameters are input features, and pose residuals are labels. The backpropagation algorithm is used to adjust the network weights. The MLP extracts the implicit correlation between input parameters and errors layer by layer using multi-layer nonlinear activation functions (such as ReLU), capturing the nonlinear distribution patterns of dynamic errors such as thermal expansion and wear gaps. During training, the mean squared error loss function guides the model to continuously reduce the deviation between the predicted residual and the actual residual, enabling the network to map from multi-physics coupling parameters to end-point errors, ultimately forming a nonlinear error prediction model that can adapt to changes in operating conditions.

[0045] Step S8: Based on the trained multilayer perceptron (MLP) residual neural network model and combined with the LM algorithm, determine the pose parameter increment, update the pose parameters based on the pose parameter increment, and generate the compensated robot end-effector pose based on the updated pose parameters and the multilayer perceptron (MLP) residual neural network model.

[0046] In detail, based on the trained multilayer perceptron (MLP) residual neural network model and combined with the LM algorithm, the pose parameter increments are determined, and the pose parameters are updated based on these increments, including: ; ; ; in, The residual value is predicted by the multilayer perceptron (MLP) residual neural network model. For link offset; For joint torsion angle; The length of the link; p This is the current pose parameter vector; This represents the increment of the pose parameters; λ It is the damping factor; I It is the identity matrix; J It is a Jacobian matrix; T This is the matrix transpose. r It is the residual function; These are the updated pose parameters.

[0047] In this embodiment, based on nonlinear optimization and iterative correction, the optimal correction amount of the pose parameters is solved using the LM algorithm. This algorithm uses the pose residual predicted by the MLP model as the optimization objective and constructs a residual function. r（p） And calculate its Jacobian matrix. J Through damping factor λThis involves balancing the local convergence of the Gauss-Newton method with the global search capability of the gradient descent method. During the iteration process, the formula... Solve for the pose parameter increments, where JᵀJ For the Hessian matrix approximation, λI To prevent matrix singularities; each iteration updates the pose parameters. The deviation between the predicted residual and the actual error is gradually reduced, and finally the compensated end pose is generated through the multilayer perceptron model.

[0048] Step S9: Periodically acquire the robot's actual end-effector pose and theoretical pose. When the error between the two exceeds a preset threshold, repeat steps S5 to S7 to correct the multilayer perceptron (MLP) residual neural network model.

[0049] In detail, the robot's actual end-effector pose and theoretical pose are periodically acquired. When the error between the two exceeds a preset threshold, steps S5 to S7 are re-executed to correct the multilayer perceptron (MLP) residual neural network model, including: ; in, This represents the residual between the robot's actual end-effector pose and its theoretical pose. This is the preset error threshold coefficient.

[0050] In this embodiment, based on adaptive learning and online optimization, a threshold-triggered mechanism is used to dynamically update the error model. During long-term operation, time-varying factors such as mechanical wear, lubricant decay, and environmental temperature changes can cause error characteristics to drift, gradually reducing the prediction accuracy of the originally trained MLP model. This step involves periodically collecting the residual between the actual pose and the theoretical pose and comparing it with a preset threshold coefficient. When the error exceeds the threshold, it indicates that the current model can no longer accurately describe the real-time error characteristics. At this point, the data acquisition and model training process is re-executed, and the latest error dataset is used to incrementally learn the MLP model, correcting the model deviation caused by time-varying factors and ensuring that the error compensation model always matches the robot's current actual error distribution characteristics.

[0051] Figure 4 This diagram illustrates a structural schematic of a computer device for implementing an adaptive compensation method for robotic grinding and polishing based on multi-line lasers, according to an exemplary embodiment of the present invention. The computer device includes: The processor 401 includes one or more processing cores. The processor 401 executes various functional applications and data processing by running software programs and modules.

[0052] The receiver 402 and transmitter 403 can be implemented as a communication component, which can be a communication chip. Optionally, this communication component can include signal transmission functionality. That is, the transmitter 403 can be used to transmit control signals to the image acquisition device and the scanning device, and the receiver 402 can be used to receive corresponding feedback commands.

[0053] The memory 404 is connected to the processor 401 via the bus 405.

[0054] The memory 404 can be used to store at least one instruction, and the processor 401 is used to execute the at least one instruction to implement the various steps in the above method embodiments.

[0055] This invention also provides a computer-readable storage medium storing at least one instruction, at least one program, code set, or instruction set, which can be loaded and executed by a processor to implement the above-described adaptive compensation method for robotic grinding and polishing based on multi-line laser.

[0056] The present invention also provides a computer program product or computer program comprising computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the adaptive compensation method for robotic grinding and polishing based on multi-line laser as described in any of the above embodiments.

[0057] Optionally, the computer-readable storage medium may include: read-only memory (ROM), random access memory (RAM), solid-state drive (SSD), or optical disk, etc. The random access memory may include resistive random access memory (ReRAM) and dynamic random access memory (DRAM). The sequence numbers of the above embodiments are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0058] It is understood that the specific examples in this document are only intended to help those skilled in the art better understand this disclosure, and are not intended to limit the scope of the invention.

[0059] It is understood that in the various embodiments of this specification, the sequence number of each process does not imply 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 this disclosure.

[0060] It is understood that the various implementation methods described in this specification can be implemented individually or in combination, and this disclosure does not limit them.

[0061] Unless otherwise stated, all technical and scientific terms used in this disclosure have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of this specification. The term "and / or" as used in this specification includes any and all combinations of one or more of the associated listed items. The singular forms "a," "the," and "the" as used in this disclosure and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise.

[0062] It is understood that the processor disclosed herein can be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method implementation can be completed by integrated logic circuits in the processor's hardware or by instructions in software form. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in this disclosure. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed herein can be directly implemented by a hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory; the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method.

[0063] It is understood that the memory in this disclosure can be volatile memory or non-volatile memory, or may include both. Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can be random access memory (RAM). It should be noted that the memory in the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0064] Those skilled in the art will recognize that the units and algorithm steps of the various examples 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 implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art 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 specification.

[0065] The above description is merely a specific embodiment of this specification, but the scope of protection of this invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this specification should be included within the scope of protection of this specification. Therefore, the scope of protection of this invention should be determined by the scope of the claims.

Claims

1. An adaptive compensation method for robotic grinding and polishing based on multi-line laser, characterized in that, The method includes: S1. Construct a multi-line laser measurement system to ensure that the robot end effector is within the coverage area of ​​the multi-line laser measurement system; the multi-line laser measurement system includes two line laser sensors mounted on the worktable at a 90° angle; S2. Scan the calibration block at the end of the robot using the multi-line laser measurement system to obtain the coordinates of feature points of the calibration block in different poses in the coordinate system of the multi-line laser measurement system. S3. Using the Horn or Umeyama absolute orientation algorithm, determine the rigid transformation matrix from the coordinate system of the multi-line laser measurement system to the coordinate system of the robot end effector; S4. Construct a robot kinematic model based on the MDH parameter method to describe the transformation relationship between the robot's base coordinate system and the robot's end effector coordinate system; S5. In response to different robot operating times, joint loads, joint angles, and ambient temperatures, acquire the robot's actual end-effector pose and theoretical pose, and construct an error dataset; the error of the error dataset is defined as: ； in, This represents the residual between the robot's actual end-effector pose and its theoretical pose. This represents the robot's actual end-effector pose. This represents the theoretical end-effector pose of the robot. Let be the joint angle of link i; T be the ambient temperature; t be the running time; and F be the joint load. S6. Construct a multilayer perceptron (MLP) residual neural network model, defining the input layer, output layer, and mean squared error loss function; the input layer of the MLP residual neural network model includes: Theoretical end-effector pose of the robot Joint angle of link i Ambient temperature T, running time t, and joint load F; The output layer of the multilayer perceptron (MLP) residual neural network model includes: The residual between the robot's actual end-effector pose and its theoretical pose ; The mean squared error loss function of the multilayer perceptron (MLP) residual neural network model includes: ； Where Loss is the mean squared error loss value; This represents the residual between the robot's actual end-effector pose and its theoretical pose. The residual value is predicted by the multilayer perceptron (MLP) residual neural network model; N is the number of training samples. S7. The multilayer perceptron (MLP) residual neural network model is trained using the error dataset to fit the nonlinear error of the robot end pose. The MLP extracts the implicit correlation between the input parameters and the error layer by layer through multilayer nonlinear activation functions to capture the nonlinear distribution law of dynamic error. The dynamic error includes thermal expansion and wear gap. S8. Based on the trained multilayer perceptron (MLP) residual neural network model, combined with the LM algorithm, determine the pose parameter increment, update the pose parameters based on the pose parameter increment, and generate the compensated robot end pose based on the updated pose parameters and the multilayer perceptron (MLP) residual neural network model. S9. Periodically acquire the robot's actual end-effector pose and theoretical pose. When the error between the two exceeds a preset threshold, repeat steps S5 to S7 to correct the multilayer perceptron (MLP) residual neural network model.

2. The adaptive compensation method for robotic grinding and polishing based on multi-line laser as described in claim 1, characterized in that, In step S3, the rigid transformation matrix from the multi-line laser measurement system coordinate system to the robot end effector coordinate system includes: And satisfy ; in, is the homogeneous transformation matrix from the coordinate system of the multi-line laser measurement system to the coordinate system of the robot end effector; It is a 3×3 rotation matrix; It is a 3×1 translation vector; These are the coordinates of the feature points in the robot's end effector coordinate system. The coordinates of the feature point in the coordinate system of the multi-line laser measurement system are given.

3. The adaptive compensation method for robotic grinding and polishing based on multi-line laser as described in claim 1, characterized in that, The adjacent coordinate system transformation matrices involved in the robot kinematics model in step S4 include: ； in, This is the coordinate transformation matrix between two adjacent links; Let be the joint angle of link i; For link i, the link offset; Let be the joint torsion angle of link i; Let i be the length of the link. Let be the Y-axis parallelism parameter of link i; c represents the cosine function cos, and s represents the sine function sin.

4. The adaptive compensation method for robotic grinding and polishing based on multi-line laser as described in claim 1, characterized in that, The transformation matrix of the robot end-effector coordinate system {6} relative to the robot base coordinate system {0} involved in the robot kinematic model in step S4. include: ； in, This is the direction vector of the X-axis in the robot's end-effector coordinate system in the robot's base coordinate system; This is the direction vector of the Y-axis in the robot's end-effector coordinate system in the robot's base coordinate system; This is the direction vector of the Z-axis in the robot's end-effector coordinate system in the robot's base coordinate system; This is the position vector of the origin of the robot's end effector coordinate system in the robot's base coordinate system.

5. The adaptive compensation method for robotic grinding and polishing based on multi-line laser as described in claim 1, characterized in that, In step S8, determining the pose parameter increment based on the trained multilayer perceptron (MLP) residual neural network model, combined with the LM algorithm, and updating the pose parameters based on the pose parameter increment includes: ； ； ； in, The residual value is predicted by the multilayer perceptron (MLP) residual neural network model. For link offset; For joint torsion angle; p is the link length; p is the current pose parameter vector; λ is the pose parameter increment; I is the damping factor; J is the identity matrix; T is the matrix transpose; r is the residual function. These are the updated pose parameters.

6. The adaptive compensation method for robotic grinding and polishing based on multi-line laser as described in claim 1, characterized in that, In step S9, the periodic acquisition of the robot's actual end-effector pose and theoretical pose, and when the error between the two exceeds a preset threshold, re-execute steps S5 to S7 to correct the multilayer perceptron (MLP) residual neural network model, including: ； in, This represents the residual between the robot's actual end-effector pose and its theoretical pose. This is the preset error threshold coefficient.

7. The adaptive compensation method for robotic grinding and polishing based on multi-line laser as described in claim 1, characterized in that, The calibration block in step S2 is a trapezoidal sheet metal calibration block, which serves as the measurement standard. The positions of its feature points are known and easily identified by the multi-line laser measurement system.

Images