Optimization method and system for manipulator control system

Abstract

The embodiment of the invention provides an optimization method and system for a manipulator control system, and the method comprises the steps: firstly obtaining an operation data set containing a control instruction sequence and corresponding motion state data in a historical operation period of a manipulator, carrying out the feature extraction of the operation data set, and obtaining an associated motion feature set; calling a pre-trained control strategy optimization model to analyze and generate strategy adjustment parameters, adaptively updating the control instruction sequence of the current operation task according to the strategy adjustment parameters to obtain a target control instruction sequence, and finally inputting the target control instruction sequence into a control system to execute operation. And real-time motion state data is collected for subsequent iterative optimization, so that the performance of the manipulator control system can be continuously improved, and the operation precision and efficiency are improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent control technology, and in particular to an optimization method and system for a manipulator control system. Background Art

[0002] In the field of industrial automation, robots, as important execution devices, are widely used in various production and manufacturing scenarios. Currently, most robot control systems use preset control strategies to drive the robot to complete tasks. However, in actual operations, due to factors such as the robot's own structural characteristics, changes in the working environment (such as temperature, humidity, and load variations), and the complexity of the task, the preset control strategy often fails to ensure that the robot can achieve ideal operating results under various working conditions. Existing robot control system optimization methods often rely on manual adjustment of control parameters based on human experience. This approach is not only inefficient but also makes it difficult to accurately grasp the complex relationship between control instructions and the robot's motion state. It is unable to adapt to dynamic changes during the operation process in a timely manner, resulting in low robot operation precision and efficiency, and even possible failure, affecting the stability of the entire production process and product quality. Summary of the Invention

[0003] In view of this, an object of the embodiments of the present invention is to provide a method and system for optimizing a manipulator control system.

[0004] According to one aspect of an embodiment of the present invention, a method for optimizing a manipulator control system is provided, the method comprising: Acquire an operation data set of the manipulator in a historical operation cycle, wherein the operation data set includes a continuously collected control instruction sequence with a time stamp and corresponding executed motion state data; Performing feature extraction processing on the operation data set to obtain a set of associated motion features between the control instruction sequence and the motion state data, wherein the associated motion feature set is used to describe motion matching characteristics during the execution of the control instruction; Calling a pre-trained control strategy optimization model to perform strategy optimization analysis on the associated motion feature set to generate strategy adjustment parameters for the current task, wherein the strategy adjustment parameters include instruction correction rules and motion coordination constraints; Adaptively updating the control instruction sequence of the current task according to the strategy adjustment parameters to obtain a revised target control instruction sequence; The target control instruction sequence is input into the manipulator control system to execute the operation, and the real-time motion state data during the execution process is collected as a new operation data set for the iterative execution of the subsequent optimization method.

[0005] According to another aspect of an embodiment of the present invention, a robot control system optimization system is provided, comprising: a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus; the memory is used to store computer programs; the processor is used to implement any one of the above steps of the robot control system optimization method when executing the computer program.

[0006] According to another aspect of an embodiment of the present invention, a readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned method for optimizing the manipulator control system can be executed.

[0007] Through any of the above aspects, the embodiment of the present invention obtains the operation data set within the historical operation cycle of the manipulator, comprehensively records the control instruction sequence and the corresponding motion state data, performs feature extraction processing on the operation data set, and obtains the associated motion feature set, which can deeply explore the action matching characteristics between the control instructions and the motion state data, call the pre-trained control strategy optimization model to analyze the associated motion feature set, generate strategy adjustment parameters, and realize the precise optimization of the control strategy. The control instruction sequence is adaptively updated according to the strategy adjustment parameters to obtain the corrected target control instruction sequence, so that the manipulator can better adapt to the current operation task and actual working conditions. Finally, the target control instruction sequence is input into the manipulator control system to perform the operation, and real-time motion state data is collected for the iterative execution of subsequent optimization methods, which can continuously improve the performance of the manipulator control system, significantly improve the operation accuracy and efficiency of the manipulator, and enhance the stability of the production process and product quality.

[0008] In order to make the above-mentioned objects, features and advantages of the embodiments of the present invention more obvious and easy to understand, the embodiments will be described in detail below in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0010] Figure 1 A schematic diagram showing components of an optimization system for a manipulator control system provided by an embodiment of the present invention is shown; Figure 2 A flow chart of an optimization method for a manipulator control system provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0011] To help those skilled in the art better understand the present invention, the following will provide a clear and complete description of the technical solutions in the embodiments of the present invention, with reference to the accompanying drawings. It is apparent that the described embodiments are only a portion of the embodiments of the present invention, not all of them. All other embodiments derived by those skilled in the art based on the embodiments of the present invention without creative effort are intended to fall within the scope of protection of the present invention.

[0012] The terms "first," "second," "third," and the like (if any) in the description and claims of the present invention and the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, so that the embodiments of the invention described herein can, for example, be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having," and any variations thereof, are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or apparatus comprising a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to these processes, methods, products, or apparatus.

[0013] Figure 1 A schematic diagram of exemplary components of a system 100 for optimizing a manipulator control system is shown. The system 100 for optimizing a manipulator control system may include one or more processors 104, such as one or more central processing units (CPUs), each of which may implement one or more hardware threads. The system 100 for optimizing a manipulator control system may also include any storage medium 106 for storing any type of information, such as code, settings, data, and the like. For example, and without limitation, the storage medium 106 may include any one or more combinations of the following: any type of RAM, any type of ROM, a flash memory device, a hard disk, an optical disk, and the like. More generally, any storage medium may use any technology to store information. Furthermore, any storage medium may provide volatile or non-volatile retention of information. Furthermore, any storage medium may represent a fixed or removable component of the system 100 for optimizing a manipulator control system. In one embodiment, when the processor 104 executes instructions stored in any storage medium or combination of storage media with dependencies, the system 100 for optimizing a manipulator control system may perform any operation of the associated instructions. The optimization system 100 of the robot control system further includes one or more drive units 108 for interacting with any storage medium, such as a hard disk drive unit, an optical disk drive unit, and the like.

[0014] An embodiment of the present invention provides an optimization system for a manipulator control system, which includes control of a manipulator arm and a dexterous hand, and the integrated control of the two is considered as manipulator control. The optimization system 100 for the manipulator control system also includes an input / output 110 (I / O), which is used to receive various inputs (via an input unit 112) and to provide various outputs (via an output unit 114). A specific output mechanism may include a presentation device 116 and a graphical user interface (GUI) 118 with a dependent relationship. The optimization system 100 for the manipulator control system may also include one or more network interfaces 120, which are used to exchange data with other devices via one or more communication units 122. One or more communication buses 124 couple the components described above together.

[0015] The communication unit 122 can be implemented in any manner, for example, through a local area network, a wide area network (e.g., the Internet), a point-to-point connection, etc., or any combination thereof. The communication unit 122 can include any combination of hardwired links, wireless links, routers, gateway functions, the optimization system 100 of the robot control system, etc., governed by any protocol or combination of protocols.

[0016] Figure 2 The schematic diagram of the process of the optimization method and system of the manipulator control system provided by the embodiment of the present invention is shown. The optimization method and system of the manipulator control system can be Figure 1 The optimization system 100 of the manipulator control system shown in FIG is executed, and the detailed steps of the optimization method of the manipulator control system are introduced as follows.

[0017] Step S110: Acquire an operation data set of the robot in a historical operation cycle, wherein the operation data set includes a continuously collected control instruction sequence with a time stamp and corresponding executed motion state data.

[0018] In actual application scenarios, take the robot responsible for chip assembly in an electronic equipment production workshop as an example. First, the original control instruction records within the historical operation cycle are extracted from the log storage module of the robot control system. These original control instruction records are arranged in chronological order. The position control instruction is used to indicate the movement position of the robot in three-dimensional space, the speed control instruction specifies its movement speed, and the force control instruction determines the applied force. For example, the position control instruction may be accurate to the coordinate information of a certain position in space, the speed control instruction may involve speed settings in different directions, and the force control instruction will specify the magnitude of the force applied in each direction.

[0019] Next, motion state data is synchronously collected, aligned with the timestamp of the original control command recording. The robot arm's joint angle feedback reflects the real-time angle of each joint, the end effector's posture feedback reflects its position and posture in space, and the contact force feedback indicates the force applied when contacting an object during operation.

[0020] Afterwards, the original control command records and motion state data are timestamped. Through a series of time calibration mechanisms, control commands and motion state data with the same timestamp are ensured to accurately correspond, just like the two sets of data are precisely matched in chronological order.

[0021] Finally, the calibrated raw control command records and motion state data are formatted uniformly. This data is organized into a structured operational data set containing timestamps, control command types, command parameter values, and corresponding motion state feedback values, facilitating subsequent processing and analysis.

[0022] Step S120: performing feature extraction processing on the operation data set to obtain a set of associated motion features between the control instruction sequence and the motion state data, wherein the associated motion feature set is used to describe the motion matching characteristics during the execution of the control instruction.

[0023] Step S121: Divide the operation data set into multiple sub-cycle data units according to the operation task stage, each sub-cycle data unit contains a control instruction sequence of a complete operation action and corresponding motion state data.

[0024] Taking chip assembly as an example, the entire process can be divided into different task phases, such as chip grabbing, chip movement, position alignment, and chip assembly. The operational data set is divided into multiple sub-cycle data units based on these phases. Each sub-cycle data unit fully covers the control instruction sequence corresponding to a specific operation (such as chip grabbing), including position control instructions, speed control instructions, force control instructions, etc., as well as the motion state data during the execution of the action, such as the robot arm joint angle feedback value, end effector posture feedback value, and contact force feedback value.

[0025] Step S122: performing a timing feature analysis on the control instruction sequence in each sub-cycle data unit to extract the change rate feature of the instruction parameters and the continuity feature of adjacent instructions.

[0026] For each sub-cycle data unit of the control instruction sequence, analyze the change rate characteristics of its instruction parameters. For example, in the sub-cycle data unit of the capture chip, observe the changes in the position control instruction parameters within the set time interval, calculate the change amount of its position parameters per unit time, and use this to determine the change rate characteristics of the position control instruction parameters. At the same time, analyze the continuity characteristics of adjacent instructions. For example, determine whether there is a large jump or smooth transition when the speed control instruction switches from one speed setting to the next, so as to measure the continuity of adjacent instructions.

[0027] Step S123: Perform spatial feature analysis on the motion state data in each sub-cycle data unit to extract joint angle change trajectory features, end position deviation features, and contact force fluctuation features.

[0028] For the motion state data of each sub-cycle data unit, the robot arm joint angle feedback value is plotted, and its angle change curve during the operation is plotted to obtain the joint angle change trajectory characteristics. For the end effector posture feedback value, the actual posture is compared with the ideal posture, and the deviation between the two is calculated to obtain the end posture deviation characteristics. For the contact force feedback value, its numerical fluctuation during the operation is observed, and the fluctuation amplitude and frequency are statistically analyzed to obtain the contact force fluctuation characteristics.

[0029] Step S124: Establish an association mapping relationship between the timing features and the spatial features, calculate the matching degree between the control instruction parameter change rate and the joint angle change trajectory, the correlation between the continuity of adjacent instructions and the end posture deviation, and the consistency between the contact force fluctuation and the force control instruction parameters.

[0030] Step S1241: For the control instruction parameter change rate characteristics and the joint angle change trajectory characteristics, statistics are made on the synchronization index of the parameter change rate and the angle change rate in the same time window. The synchronization index is measured by the consistency of the change trends of the two.

[0031] Within the same time window, for example, the time window from grasping to moving the chip, observe both the rate of change of control command parameters (such as the rate of change of position control command parameters) and the rate of change of angles in the joint angle change trajectory. Compare the change trends of the two. If both increase or decrease simultaneously and the magnitude of change is similar, the change trends are highly consistent and the synchronization index is correspondingly high. Conversely, if the change trends differ significantly, the synchronization index is low.

[0032] Step S1242: Analyze the correlation coefficient between the posture deviation value at the instruction switching moment and the instruction continuity based on the continuity characteristics of adjacent instructions and the terminal posture deviation characteristics. The correlation coefficient is calculated by the ratio of the average posture deviation in a continuous time period to the instruction switching frequency.

[0033] At the moment of command switching, such as when switching from the speed control command for chip grabbing to the speed control command for chip moving, the end effector's posture deviation is recorded. Over multiple consecutive command switching periods, the average posture deviation is calculated, and the command switching frequency is calculated. The average posture deviation is divided by the command switching frequency to obtain the correlation coefficient. A high correlation coefficient indicates a strong correlation between the continuity of adjacent commands and the end effector's posture deviation; conversely, a low correlation coefficient indicates a weak correlation.

[0034] Step S1243: With respect to the contact force fluctuation characteristics and the force control instruction parameter characteristics, the degree of consistency between the set value of the force control instruction parameter and the fluctuation frequency and amplitude of the contact force feedback value is compared. The degree of consistency is measured by the overlap ratio of the fluctuation periods of the two and the average absolute value of the amplitude difference.

[0035] Observe the fluctuations in the set values ​​of the force control command parameters and the contact force feedback values. Count the fluctuation cycles, calculate the overlap ratio, and calculate the mean absolute value of the amplitude differences between the two. A high overlap ratio and a low mean absolute value of the amplitude differences indicate a high degree of agreement. Conversely, a low overlap ratio and a high mean absolute value of the amplitude differences indicate a low degree of agreement.

[0036] Step S125: taking the matching degree, correlation and consistency as a set of associated motion features between the control instruction sequence and the motion state data.

[0037] The calculated matching between the control command parameter change rate and joint angle change trajectory, the correlation between the continuity of adjacent commands and the end-point posture deviation, and the consistency between contact force fluctuations and force control command parameters are integrated to form a set of associated motion features. This set of associated motion features comprehensively reflects the inherent connection between the control command sequence and the motion state data.

[0038] Step S130: calling a pre-trained control strategy optimization model to perform strategy optimization analysis on the associated motion feature set, and generating strategy adjustment parameters for the current task, wherein the strategy adjustment parameters include instruction correction rules and motion coordination constraints.

[0039] Step S131: inputting the associated motion feature set into the feature input layer of the control strategy optimization model, performing feature standardization processing and feature dimension alignment processing to obtain the associated motion feature set.

[0040] The control strategy optimization model has a defined structure, in which the feature input layer receives a set of interrelated motion features. Because each feature in the set may have different dimensions and scales, the model must perform feature normalization and dimension alignment to effectively process these features. For example, certain features may need to have their numerical range adjusted to a set interval using a defined normalization method. Simultaneously, the feature dimensions must be checked and adjusted to ensure they meet the requirements of subsequent model processing, resulting in a processed set of interrelated motion features.

[0041] Step S132: performing cross-feature association modeling processing on the associated motion feature set through the feature fusion layer of the control strategy optimization model to generate a fusion feature vector reflecting the intrinsic relationship between the control instruction and the motion state.

[0042] The feature fusion layer models cross-feature correlations within a standardized and dimensionally aligned set of associated motion features. This layer analyzes the relationships between the different features within the set, such as the degree of alignment between the rate of change of control command parameters and the trajectory of joint angle changes, and the correlation between the continuity of adjacent commands and end-point pose deviations. By fusing these features, a fused feature vector is generated that comprehensively reflects the intrinsic relationship between control commands and motion states.

[0043] Step S133: Utilize the strategy prediction layer of the control strategy optimization model to perform task adaptation analysis on the fused feature vector, and predict the type of control instruction that needs to be adjusted and the corresponding correction direction in combination with the type and accuracy requirements of the current task.

[0044] Step S1331: Obtain the type identification and precision requirement parameters of the current task, where the type identification is used to distinguish between a handling task, an assembly task, or a processing task, and the precision requirement parameters are used to indicate the allowable deviation range of the end effector posture.

[0045] Taking chip assembly as an example, we first identify the current task type as assembly and obtain the required accuracy parameters, which are the allowable deviation range of the end effector's position when assembling the chip. This tolerance is set based on the specific requirements of chip assembly. For example, position deviation in a specific direction cannot exceed a certain value, and attitude angle deviation also has corresponding limits.

[0046] Step S1332: Input the type identification, accuracy requirement parameters and fusion feature vector into the task association module of the strategy prediction layer to analyze the influence weight of each associated motion feature on the execution effect under different task types.

[0047] The task type identifier, precision requirement parameters, and fused feature vector are input into the task association module of the strategy prediction layer. This module analyzes the impact of each feature in the associated motion feature set on the assembly execution performance for assembly tasks. For example, for chip assembly tasks, the matching degree between the joint angle change trajectory characteristics and the rate of change of the control instruction parameters may have a significant impact on assembly accuracy, while the correlation between the continuity of adjacent instructions and the end-point posture deviation plays a significant role in assembly stability. This determines the weight of the impact of each associated motion feature on the execution performance.

[0048] Step S1333: performing weighted summation processing on each feature item in the fused feature vector according to the influence weight, and generating a key feature value reflecting the current task execution bottleneck.

[0049] Based on the weights of each associated motion feature's impact on the execution outcome, a weighted sum is applied to each feature item in the fused feature vector. For example, if the weighted match between the joint angle trajectory and the control command parameter change rate is high, the weight coefficient for this feature item will be relatively large in the weighted sum. This weighted summation calculation yields a key feature value that reflects the bottleneck in the current chip assembly task.

[0050] Step S1334: Based on the key characteristic value, determine the type of control instruction that causes the execution effect to fail to meet the standard, and the control instruction type includes a position control instruction, a speed control instruction, or a force control instruction.

[0051] Based on the calculated key eigenvalues, the control instruction type causing substandard chip assembly performance can be analyzed and determined. If the key eigenvalues ​​indicate deviations when moving the chip to a specific position, this may indicate a problem with the position control instruction. If speed variations during assembly cause instability, the speed control instruction may need adjustment. If anomalies in contact force are observed, the force control instruction may need correction.

[0052] Step S1335: determining a corresponding correction direction according to the deviation direction of the key characteristic value, wherein the correction direction includes increasing the instruction parameter value, decreasing the instruction parameter value, or adjusting the instruction change rate.

[0053] Determine the direction of control command correction based on the deviations reflected by the key characteristic values. For example, if the key characteristic values ​​indicate that the chip's position deviation in a certain direction is too small, the position control command parameter value may need to be increased; if the speed is too fast, resulting in assembly instability, the speed control command parameter value may need to be reduced; if the control command parameter change rate is unreasonable, the command change rate may need to be adjusted.

[0054] Step S134: performing motion coordination constraint analysis on the correction direction through the constraint generation layer of the control strategy optimization model to extract motion coordination constraint conditions.

[0055] The constraint generation layer performs motion coordination constraint analysis based on the correction direction determined by the strategy prediction layer. For example, in chip assembly, when a position control instruction parameter value adjustment is determined, the robot arm analyzes whether the movement of each joint will conflict when executing the adjustment, and whether the position adjustment is coordinated with the end effector. This analysis extracts a series of motion coordination constraints, such as the range of joint angle changes and the sequence of joint movements. This ensures that the robot can smoothly and coordinately complete the task when executing the corrected control instructions.

[0056] Step S135: Outputting the control instruction type, correction direction, and action coordination constraint condition that need to be adjusted as strategy adjustment parameters.

[0057] The control instruction type that needs to be adjusted determined by the strategy prediction layer, the corresponding correction direction, and the action coordination constraints extracted by the constraint generation layer are integrated into strategy adjustment parameters for output.

[0058] Step S140: Adaptively updating the control instruction sequence of the current task according to the strategy adjustment parameters to obtain a revised target control instruction sequence.

[0059] Step S141: extracting the type of control instruction that needs to be adjusted and the correction direction from the strategy adjustment parameters, and determining the position of the control instruction that needs to be updated in the current control instruction sequence.

[0060] Taking the control instruction sequence for a chip assembly operation as an example, the strategy adjustment parameters are used to extract the type of control instruction to be adjusted, such as a position control instruction, and the corresponding correction direction, such as increasing the instruction parameter value. The specific location of the position control instruction related to chip position adjustment is then searched within the control instruction sequence for the current chip assembly operation, allowing for targeted updates.

[0061] Step S142: adjusting the control instruction parameter value of the corresponding position according to the correction direction, wherein the adjustment process includes parameter offset calculation based on historical data and parameter upper and lower limit constraints based on task accuracy requirements.

[0062] According to the determined correction direction, the position control instruction parameter value found is adjusted. First, the parameter offset is calculated based on historical data. Review the historical operation data and the adjustment amount of the control instruction parameters when similar deviations occur in similar assembly task scenarios. For example, analyze the relationship between the position deviation and the control instruction parameter adjustment amount in previous chip assembly operations, and use this as a basis to calculate the current parameter offset that should be adjusted. At the same time, considering the accuracy requirements of the current chip assembly task, set the upper and lower limit constraints of the parameters. Ensure that the adjusted parameter value can meet the requirements of correcting the deviation and is within the range allowed by the task accuracy.

[0063] Step S143: extract the action coordination constraint conditions in the strategy adjustment parameters, perform collaborative verification on the adjusted control instruction parameter values, perform time series smoothing on the verified control instruction parameter values, eliminate parameter mutations between adjacent instructions, and generate a modified control instruction sequence with continuous change characteristics.

[0064] Motion coordination constraints, such as the range of joint angle variations and the sequence of joint motions, are extracted from the strategy adjustment parameters. The adjusted control command parameter values ​​are then fed into a relevant verification mechanism, such as by establishing a kinematic and dynamic model of the manipulator. Assuming a kinematic model has been established to describe the mapping between joint angles and end-effector pose, the adjusted control command parameter values ​​for each joint are fed into the model. The corresponding predicted end-effector pose is then calculated and compared with the pose requirement for the current task to determine if any deviations exceed the allowable deviation range. Simultaneously, the adjusted rate of change parameters for each joint control command are extracted and fed into the manipulator's dynamic model to calculate the predicted drive torque for each joint. This is then compared with the maximum allowable drive torque of the manipulator joints to determine if there is a risk of motion conflict caused by excessive torque. If both the pose deviation requirements and the torque safety requirements are met, the adjusted control command parameter values ​​are deemed to have passed the coordination verification. Time series smoothing is performed on these verified control command parameter values. Observe the parameter changes between adjacent instructions, use smoothing algorithm to adjust the parameters, eliminate possible parameter mutations, and make the control instruction parameters show continuous change characteristics in the time series, thereby generating a corrected control instruction sequence.

[0065] Step S144: Integrate the modified control instruction sequence with the unadjusted control instruction sequence to form a target control instruction sequence covering the entire operation process.

[0066] The adjusted and verified revised control instruction sequence is integrated with the unadjusted portions of the current control instruction sequence. For example, in the control instruction sequence for a chip assembly operation, while the position control instruction is adjusted, other instructions such as the speed control instruction and the force control instruction may not be adjusted. The revised position control instruction sequence is combined with the unadjusted speed control instruction and force control instruction sequences in the chronological order of the operation flow to form a target control instruction sequence that covers the entire chip assembly operation flow.

[0067] Step S150: inputting the target control instruction sequence into the manipulator control system to perform the operation, and collecting the real-time motion state data during the execution process as a new operation data set for the iterative execution of the subsequent optimization method.

[0068] Step S151: sending the target control instruction sequence to the controller of the robot in the order of timestamps, triggering the robot arm to perform the operation according to the modified control instructions.

[0069] Taking chip assembly as an example, a target control instruction sequence covering the entire chip assembly process is generated and sent to the robot controller one by one according to timestamp order. After receiving the instructions, the controller drives the robot arm to perform the operation according to the revised control instructions, such as first grasping the chip and then accurately assembling it to the specified position according to the adjusted position, speed, and force control instructions.

[0070] Step S152: During the execution process, the encoder installed at the joint of the robot arm collects the real-time feedback value of the joint angle, the posture sensor of the end effector collects the real-time feedback value of the posture, and the force sensor collects the real-time feedback value of the contact force.

[0071] As the robotic arm performs chip assembly operations, encoders installed at the arm's joints record the angle changes of each joint in real time, providing real-time feedback on the joint angles. Simultaneously, the posture sensor on the end effector continuously monitors its position and posture in space, providing real-time posture feedback. Force sensors detect the forces exerted by the contact with chips and other components during assembly, providing real-time feedback on the contact forces.

[0072] Step S153: Timestamp the joint angle real-time feedback value, the posture real-time feedback value, and the contact force real-time feedback value so that they correspond one-to-one with the timestamp of the target control instruction sequence.

[0073] The collected real-time feedback values ​​of joint angle, posture, and contact force are time-stamped. Through a precise time synchronization mechanism, the timestamps of these real-time feedback values ​​are ensured to correspond to the timestamps of each instruction in the target control instruction sequence.

[0074] Step S154: The real-time feedback value marked with the timestamp is stored in association with the target control instruction sequence to generate a new operation data set including the timestamp, the control instruction type, the corrected instruction parameter value and the corresponding real-time feedback value.

[0075] The time-stamped real-time feedback values ​​are associated with the target control instruction sequence. The timestamp, control instruction type (e.g., position control instruction, velocity control instruction, force control instruction, etc.), revised instruction parameter values, and corresponding real-time feedback values ​​(joint angle real-time feedback values, posture real-time feedback values, contact force real-time feedback values) are integrated and stored to generate a new operation data set. This new operation data set records detailed data on the robot's execution of the revised control instructions.

[0076] Step S155: storing the new operation data set in the log storage module of the manipulator control system for use as historical data supplement during subsequent iterative execution of the optimization method.

[0077] The generated new operation data set is stored in the log storage module of the manipulator control system for use in supplementing historical data during the subsequent iterative execution of the optimization method.

[0078] Or: the target control instruction sequence is input into the manipulator control system to perform the operation, and the real-time motion state data during the execution process is collected as a new operation data set, including: sending the target control instruction sequence to the manipulator controller in timestamp order, triggering the manipulator to perform the operation according to the revised control instruction; during the execution process, the joint angle real-time feedback value is collected by the encoder installed at the manipulator joint, the posture real-time feedback value is collected by the posture sensor of the end effector, the contact force real-time feedback value is collected by the force sensor, and the object contact pressure real-time feedback value is sensed by the tactile sensor; the joint angle real-time feedback value, posture real-time feedback value, contact force and contact pressure real-time feedback value are timestamped so that they correspond to the timestamp of the target control instruction sequence; the real-time feedback value marked with the timestamp is data-associated and stored with the target control instruction sequence to generate a new operation data set including the timestamp, control instruction type, revised instruction parameter value and corresponding real-time feedback value; the new operation data set is stored in the log storage module of the manipulator control system for historical data supplement during the iterative execution of the subsequent optimization method.

[0079] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0080] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations that come within the meaning and range of equivalents of the claims be embraced therein.

Claims

1. A method for optimizing a manipulator control system, characterized in that: The method comprises: Acquire an operation data set of the manipulator in a historical operation cycle, wherein the operation data set includes a continuously collected control instruction sequence with a time stamp and corresponding executed motion state data; Performing feature extraction processing on the operation data set to obtain a set of associated motion features between the control instruction sequence and the motion state data, wherein the associated motion feature set is used to describe motion matching characteristics during the execution of the control instruction; Calling a pre-trained control strategy optimization model to perform strategy optimization analysis on the associated motion feature set to generate strategy adjustment parameters for the current task, wherein the strategy adjustment parameters include instruction correction rules and motion coordination constraints; Adaptively updating the control instruction sequence of the current task according to the strategy adjustment parameters to obtain a revised target control instruction sequence; The target control instruction sequence is input into the manipulator control system to execute the operation, and the real-time motion state data during the execution process is collected as a new operation data set for the iterative execution of the subsequent optimization method.

2. The optimization method of the manipulator control system according to claim 1, characterized in that: The acquisition of the operation data set of the manipulator in the historical operation cycle includes a continuously collected control instruction sequence with a time stamp and the corresponding executed motion state data, including: Extracting original control instruction records within a historical operation cycle from a log storage module of the manipulator control system, wherein the original control instruction records include position control instructions, speed control instructions, and force control instructions arranged in chronological order; Synchronously collecting motion state data aligned with the original control instruction recording timestamp, the motion state data including the robot arm joint angle feedback value, the end effector posture feedback value and the contact force feedback value; Performing time stamp calibration on the original control instruction record and motion state data so that control instructions and motion state data corresponding to the same time stamp have a one-to-one correspondence; The calibrated original control instruction records and motion state data are formatted in a unified manner to generate a structured operation data set containing timestamps, control instruction types, instruction parameter values, and corresponding motion state feedback values.

3. The optimization method of the manipulator control system according to claim 1, characterized in that: The performing feature extraction processing on the operation data set to obtain a set of associated motion features between the control instruction sequence and the motion state data includes: Dividing the operation data set into a plurality of sub-cycle data units according to the operation task stage, each sub-cycle data unit contains a control instruction sequence of a complete operation action and corresponding motion state data; Performing timing feature analysis on the control instruction sequence in each sub-cycle data unit to extract the change rate features of instruction parameters and the continuity features of adjacent instructions; Perform spatial feature analysis on the motion state data in each sub-cycle data unit to extract the joint angle change trajectory features, end position deviation features, and contact force fluctuation features; Establishing a correlation mapping relationship between the temporal features and the spatial features, calculating the matching degree between the control instruction parameter change rate and the joint angle change trajectory, the correlation between the continuity of adjacent instructions and the end posture deviation, and the consistency between the contact force fluctuation and the force control instruction parameters; The matching degree, correlation and consistency are used as a set of associated motion features between the control instruction sequence and the motion state data.

4. The optimization method of the manipulator control system according to claim 3, characterized in that: The establishment of the correlation mapping relationship between the temporal features and the spatial features, the calculation of the matching degree between the control instruction parameter change rate and the joint angle change trajectory, the correlation between the continuity of adjacent instructions and the end posture deviation, and the consistency between the contact force fluctuation and the force control instruction parameter include: Based on the control instruction parameter change rate characteristics and joint angle change trajectory characteristics, the synchronization index of the parameter change rate and the angle change rate in the same time window is calculated. The synchronization index is measured by the consistency of the change trends of the two. Based on the continuity characteristics of adjacent instructions and the terminal posture deviation characteristics, the correlation coefficient between the posture deviation value at the instruction switching moment and the instruction continuity is analyzed. The correlation coefficient is calculated by the ratio of the average posture deviation in a continuous time period to the instruction switching frequency; Based on the contact force fluctuation characteristics and force control command parameter characteristics, the degree of consistency between the set value of the force control command parameter and the contact force feedback value in terms of the fluctuation frequency and amplitude is compared. The degree of consistency is measured by the overlap ratio of the two fluctuation periods and the average absolute value of the amplitude difference. The synchronization index, correlation coefficient and degree of coincidence are used as specific feature items of the associated motion feature set.

5. The optimization method of the manipulator control system according to claim 1, characterized in that: The calling of the pre-trained control strategy optimization model to perform strategy optimization analysis on the associated motion feature set to generate strategy adjustment parameters for the current task includes: Inputting the associated motion feature set into the feature input layer of the control strategy optimization model, performing feature standardization processing and feature dimension alignment processing to obtain the associated motion feature set; Performing cross-feature correlation modeling on the associated motion feature set through the feature fusion layer of the control strategy optimization model to generate a fusion feature vector reflecting the intrinsic relationship between the control command and the motion state; The strategy prediction layer of the control strategy optimization model is used to perform task adaptation analysis on the fused feature vector, and the type of control instruction that needs to be adjusted and the corresponding correction direction are predicted based on the type and accuracy requirements of the current task; Performing motion coordination constraint analysis on the correction direction through the constraint generation layer of the control strategy optimization model to extract motion coordination constraint conditions; The control instruction type, correction direction and action coordination constraint condition that need to be adjusted are output as strategy adjustment parameters.

6. The optimization method of the manipulator control system according to claim 5, characterized in that: The strategy prediction layer of the control strategy optimization model performs task adaptation analysis on the fused feature vector, and predicts the type of control instruction that needs to be adjusted and the corresponding correction direction based on the type and accuracy requirements of the current task, including: Obtain the type identification and precision requirement parameters of the current task, where the type identification is used to distinguish between a handling task, an assembly task, or a processing task, and the precision requirement parameters are used to indicate the allowable deviation range of the end effector's posture; Input the type identification, accuracy requirement parameters and fusion feature vector into the task association module of the strategy prediction layer to analyze the influence weight of each associated motion feature on the execution effect under different task types; Performing weighted summation processing on each feature item in the fused feature vector according to the influence weight to generate a key feature value reflecting the current task execution bottleneck; Determining the type of control instruction that causes the execution effect to fail to meet the standard based on the key characteristic value, the control instruction type including a position control instruction, a speed control instruction, or a force control instruction; A corresponding correction direction is determined according to the deviation direction of the key characteristic value, and the correction direction includes increasing the instruction parameter value, decreasing the instruction parameter value, or adjusting the instruction change rate.

7. The optimization method of the manipulator control system according to claim 1, characterized in that: Adaptively updating the control instruction sequence of the current task according to the strategy adjustment parameters to obtain a revised target control instruction sequence includes: Extracting the type of control instruction that needs to be adjusted and the correction direction from the strategy adjustment parameters, and determining the position of the control instruction that needs to be updated in the current control instruction sequence; Adjusting the control instruction parameter value of the corresponding position according to the correction direction, the adjustment process includes parameter offset calculation based on historical data and parameter upper and lower limit constraints based on task accuracy requirements; Extracting the action coordination constraint conditions in the strategy adjustment parameters, performing coordination verification on the adjusted control instruction parameter values, performing time series smoothing processing on the verified control instruction parameter values, eliminating parameter mutations between adjacent instructions, and generating a modified control instruction sequence with continuous change characteristics; The modified control instruction sequence is integrated with the unadjusted control instruction sequence to form a target control instruction sequence covering the complete operation process.

8. The optimization method of the manipulator control system according to claim 7, characterized in that: The collaborative verification of the adjusted control instruction parameter value includes: Establishing a kinematic model of the robotic arm, wherein the kinematic model is used to describe the mapping relationship between each joint angle and the position of the end effector; Inputting the adjusted control instruction parameter values ​​of each joint into the kinematic model to calculate the corresponding end effector posture prediction value; Comparing the predicted value of the end effector posture with the required value of the posture of the current task to determine whether there is a situation that exceeds the allowable deviation range; Extract the change rate parameters of each joint control command after adjustment, input them into the dynamic model of the robot arm to calculate the predicted driving torque value of each joint; Comparing the predicted driving torque value with the maximum allowable driving torque of the robotic arm joint to determine whether there is a risk of motion conflict caused by excessive torque; If both the posture deviation requirement and the torque safety requirement are met at the same time, the adjusted control instruction parameter value is determined to have passed the collaborative verification.

9. The optimization method of the manipulator control system according to claim 1, characterized in that: The step of inputting the target control instruction sequence into the manipulator control system to execute the operation and collecting the real-time motion state data during the execution as a new operation data set includes: Sending the target control instruction sequence to the controller of the robot in the order of timestamps, triggering the robot arm to perform the operation according to the revised control instructions; During the execution process, the encoder installed at the joint of the robot arm collects the real-time feedback value of the joint angle, the posture sensor of the end effector collects the real-time feedback value of the posture, and the force sensor collects the real-time feedback value of the contact force; Timestamping the joint angle real-time feedback value, the posture real-time feedback value, and the contact force real-time feedback value so as to correspond one-to-one with the timestamp of the target control instruction sequence; The real-time feedback value marked with the timestamp is stored in association with the target control instruction sequence to generate a new operation data set including the timestamp, the control instruction type, the corrected instruction parameter value and the corresponding real-time feedback value; The new operation data set is stored in the log storage module of the manipulator control system for use in supplementing historical data during the subsequent iterative execution of the optimization method.

10. An optimization system for a manipulator control system, characterized in that: include: A processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus; The memory is used to store a computer program; the processor is used to implement the steps of the manipulator control system optimization method described in any one of claims 1 to 9 when executing the computer program.

Images