A time rounding method and system based on velocity superposition

Abstract

The present invention relates to the field of intelligent control technology of a robotic arm, and in particular to a time rounding method and system based on speed superposition. The steps include the following steps: obtaining remote sensing control instructions of the robotic arm, encoding the remote sensing control instructions, obtaining basic data of motion nodes, filtering the basic data of motion nodes, obtaining motion change data of the robotic arm, performing motion trajectory result analysis and processing according to the motion change data, generating motion planning result data, performing data dimension reduction processing according to the motion planning result data, generating standard two-dimensional planning result data, performing planning time extraction on the standard two-dimensional planning result data, generating total planning time data, performing robotic arm time rounding processing on the total planning time data, and generating rounded time data. The present invention realizes a time rounding method of speed superposition by identifying and processing the motion control instructions of the robotic arm.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent control of robotic arms, and in particular to a time rounding method and system based on speed superposition. Background Art

[0002] Initially, robotic arms were designed for repetitive tasks on industrial production lines, such as handling and assembly. With the rapid development and continuous advancement of the Internet and artificial intelligence technologies, intelligent robotic arms have gradually become more flexible and intelligent, and can complete more complex tasks, such as precision machining and collaborative operations. The advantages of intelligent robotic arms include high efficiency, high precision, high reliability, high safety, and saving human resources. However, when multiple robotic arms operate at the same time, problems such as poor coordination, collision, and misoperation may occur. If there is no good coordination and communication between the robotic arms, they may interfere with each other and cause operation failure. However, in response to the coordination and communication problem, traditional robotic arms are prone to jitter when receiving continuous instructions, and cannot solve the time synchronization between multiple robotic arms well, and improve the accuracy and efficiency of robotic arm operations. Therefore, a time rounding method and system based on velocity superposition came into being. Summary of the invention

[0003] Based on this, the present invention provides a time rounding method and system based on velocity superposition to solve at least one of the above technical problems.

[0004] To achieve the above object, a time rounding method based on velocity superposition is provided, the method comprising the following steps:

[0005] Step S1: Obtain the remote sensing control instructions of the robot arm, encode the remote sensing control instructions, and obtain basic data of the motion node;

[0006] Step S2: Filter the basic data of the motion node to obtain the motion change data of the robot arm;

[0007] Step S3: Analyze and process the motion trajectory results according to the motion change data to generate motion planning result data;

[0008] Step S4: performing data dimensionality reduction processing according to the motion planning result data to generate standard two-dimensional planning result data;

[0009] Step S5: extracting planning time from the standard two-dimensional planning result data to generate total planning time data;

[0010] Step S6: performing robot arm time rounding processing on the planned total time data to generate rounded time data;

[0011] Step S7: Perform simulation superimposed motion processing on the rounded time data to generate actual motion values.

[0012] The present invention obtains the rocker control instructions of the mechanical arm and encodes the remote sensing control instructions, so that the mechanical arm can process and execute instructions faster, improve the processing speed of the mechanical arm, reduce the instruction storage space, and improve the storage efficiency. The encoded motion node basic data can be transmitted faster, reducing the data transmission time, improving the data transmission efficiency, improving the security of the data, preventing the data from being illegally obtained and tampered with, facilitating the subsequent data processing and analysis, and improving the efficiency and accuracy of data analysis; by filtering the motion node basic data, invalid data and erroneous data are removed, and the accuracy and reliability of the data are improved. According to the filtering processing of the motion node basic data, the time and cost of data processing can be reduced, the efficiency of subsequent processing can be improved, the motion node basic data can be made more standardized, and the influence of uncertain factors can be reduced; by performing motion trajectory result analysis on the motion change data, the pre-operation trajectory of the mechanical arm is analyzed, and the accuracy and reliability of the motion of the mechanical arm are improved. By performing operation trajectory analysis and calculation on the motion change data of the mechanical arm, the result is obtained. The optimal trajectory path reduces the speed loss and data calculation amount during the movement of the robot arm; by performing data dimensionality reduction processing on the motion planning result data, the complexity and redundancy of the data are reduced, the efficiency of data processing and analysis is improved, the three-dimensional data is reduced to two-dimensional data, the unstable factors in the motion planning result data are eliminated, and the motion planning result data is clearer and more obvious; by performing planning time extraction on the standard two-dimensional planning result data, the robot arm movement time period can be accurately extracted, the accuracy and stability of the robot arm movement can be improved, and the cost and time of manual intervention can be reduced; the total planning time data is rounded off, and by rounding off the minimum time period, the robot arm movement interference and error are reduced, the accuracy and efficiency of the robot arm movement are improved, and the movement speed control of the robot arm is improved; by performing simulation superposition motion processing on the time rounding data, the motion data is superimposed and simulated, thereby improving the actual speed, while eliminating the robot arm jitter caused by continuous instructions of the robot arm, and improving the efficiency and stability of the robot arm movement. Therefore, the speed superposition time rounding method of the present invention can clearly identify and round the continuous control instructions of the robot arm, improve the corresponding speed of the actual control of the robot arm, and at the same time eliminate the jitter of the robot arm caused by continuous instructions, saving tedious manual solution steps.

[0013] Preferably, step S1 comprises the following steps:

[0014] Step S11: Obtaining robot arm rocker instructions;

[0015] Step S12: performing binary transcoding processing on the robot arm remote sensing instruction to generate a robot arm digital signal;

[0016] Step S13: using supervised learning to perform a robot arm behavior logic prediction on the robot arm digital signal, thereby generating robot arm judgment data;

[0017] Step S14: extracting data orientation features from the robot arm judgment data, thereby obtaining robot arm orientation data;

[0018] Step S15: Screen the motion node data of the robot arm orientation data to obtain motion node basic data.

[0019] The present invention can reduce the error and interference caused by the control system by performing binary transcoding processing on the robot arm rocker instructions, thereby improving the accuracy and stability of the control; use supervised learning to perform behavioral logic prediction on the robot arm digital signal, so as to judge the operating state of the robot arm, which is convenient for subsequent data processing; perform data-oriented feature extraction on the robot arm judgment data, and extract the robot arm movement-related data, which is convenient for subsequent processing and identification; perform data screening on the robot arm orientation data, so as to remove some repeated data and invalid data that interfere with the subsequent processing, thereby improving the accuracy of the subsequent processing.

[0020] Preferably, step S2 comprises the following steps:

[0021] Step S21: collecting historical data of the motion node basic data to generate historical motion node basic data;

[0022] Step S22: extracting motion change feature data from historical motion node basic data to generate motion feature data;

[0023] Step S23: using deep learning to predict the motion change of the robot arm based on the motion feature data to generate motion dynamic change data;

[0024] Step S24: performing dynamic change data encoding processing on the motion dynamic change data to obtain a motion change data encoding segment;

[0025] Step S25: performing time series extraction processing on the motion change data encoding segment to generate a motion change time series;

[0026] Step S26: performing motion change data segment splicing processing on the motion change data encoding segments according to the motion change time sequence to generate initial robot arm motion change data;

[0027] Step S27: performing effective data extraction processing on the initial robot arm motion change data according to a preset effective time period to generate effective motion change data.

[0028] The present invention can help identify various data types and styles of historical motion nodes by collecting historical data of motion node basic data, improve data accuracy, and thus obtain historical motion node basic data, which can be used for subsequent processing of the robotic arm; extract motion change feature data of the historical motion node basic data, and extract features such as speed and acceleration of the robotic arm's historical motion change data, so as to facilitate subsequent motion prediction of the robotic arm; use deep learning in neural networks to perform motion change prediction on motion feature data, and more accurately and quickly obtain the dynamic motion change data of the robotic arm, saving computing power and time and reducing the load pressure of the system; perform motion change data encoding processing on the motion change data, and obtain motion change data encoding. The code segment can reduce the storage space and transmission bandwidth of data, and improve the transmission efficiency and speed of data; the time series extraction of the motion change data code segment generates the motion change time series, which can effectively understand the changing law of the robot arm motion, discover the key points and characteristics of the robot arm motion, and facilitate subsequent processing, while reducing the complexity of the data and improving the efficiency of data processing and analysis; splicing the motion change time series and the motion change data code segment to obtain more accurate initial robot arm motion change data, which is convenient for subsequent processing and identification, and more reliable for automated processing; the effective data extraction of the initial robot arm motion change data through the preset effective time period can eliminate invalid data segments, thereby obtaining effective motion change data and improving data accuracy.

[0029] Preferably, step S3 comprises the following steps:

[0030] Step S31: Processing the motion change data into planning results, thereby generating planning result feature data;

[0031] Step S32: performing feature vector extraction processing on the planning result feature data, thereby generating a planning result feature vector;

[0032] Step S33: using the robot arm trajectory path coordinate analysis formula to perform motion trajectory coordinate analysis on the planning result feature vector to generate a motion trajectory path coordinate vector;

[0033] Step S34: performing vector superposition processing on the motion trajectory path coordinate vector to generate a motion planning result vector;

[0034] Step S35: Perform feature vector compression encoding processing on the planning result vector to generate motion planning result data.

[0035] The present invention processes the motion change data through planning results, so that the motion change data can be planned according to the result characteristics, generate planning result characteristic data, and reduce data redundancy; extract the data with vector characteristics in the data through feature vector extraction of the planning result characteristic data, generate the planning result characteristic vector, reduce the influence of noise data on the subsequent, prevent overfitting, and improve the reliability of the data; use the robot arm trajectory path coordinates to perform motion trajectory analysis on the planning result characteristic vector, obtain the specific path coordinate vector of the robot arm, reduce the collision risk that may occur during the movement of the robot arm, and improve the data accuracy; perform vector superposition on the motion trajectory path coordinate vector, improve the accuracy of the motion trajectory path coordinate vector, and better represent the motion trajectory path coordinate vector characteristics; perform feature vector compression encoding on the planning result vector, improve the efficiency of vector processing and analysis, improve the privacy and security of the vector, and reduce the storage space of the vector.

[0036] Preferably, the robot arm trajectory path coordinate analysis formula in step S33 is as follows:

[0037]

[0038] In the formula, q represents the path coordinate of the robot arm, A i Indicates the length of the motion timestamp of the i-th robot terminal state, a i represents the axis motion angle of the i-th robot arm in the final state, B i Indicates the length of the initial motion timestamp of the i-th robot arm, b i represents the initial state axis motion angle of the i-th robot arm, e represents the power loss constant of the robot arm motion, s represents the robot arm motion space coordinate system, w represents the robot arm space coordinate projection value, m represents the upper limit of the robot arm motion acceleration, t represents the time from the initial state to the terminal state of the robot arm, θ represents the robot arm direction offset, and ε represents the abnormal adjustment value of the robot arm's path coordinate.

[0039] The present invention provides a robot arm trajectory path coordinate analysis formula, which fully considers the path coordinate q of the robot arm, the length A of the motion timestamp of the i-th robot arm terminal state, and the i , the final axis motion angle a of the i-th robot arm i , length of the initial motion timestamp of the i-th robot arm B i , the initial axis motion angle b of the i-th robot arm i , the power loss constant e of the robot movement, the space coordinate system s of the robot movement, the projection value w of the space coordinate of the robot movement, the upper limit m of the acceleration of the robot movement, the time t of the robot from the initial state to the terminal state, the direction offset θ of the robot, the abnormal adjustment value ε of the path coordinate of the robot, and the interaction between the motion timestamp length and the abnormal adjustment value ε of the path coordinate of the robot, a functional relationship has been formed Through the interaction between the length of the motion timestamp and the projection value of the robot's spatial coordinates, the spatial feature extraction is performed while ensuring the accuracy of the data, and the robot's motion spatial coordinate system is extracted. The robot's orientation offset is generated by using the robot's terminal state axis motion angle and the length of the robot's terminal state motion timestamp. The data redundancy is reduced while ensuring the accuracy of the data, the computing power is saved, and the calculation is achieved to converge quickly. The robot's orientation offset is corrected by the abnormal adjustment value ε of the robot's path coordinates, so that the robot's trajectory path coordinates are generated more accurately by calculating the acceleration upper limit of the motion power loss constant, thereby improving the accuracy and reliability of the robot's motion trajectory. At the same time, the parameters such as the robot's orientation offset and the robot's motion power loss constant in the formula can be adjusted according to the actual situation, so as to adapt to different robot motion scenarios and improve the applicability and flexibility of the algorithm.

[0040] Preferably, step S4 comprises the following steps:

[0041] Step S41: obtaining motion planning result data;

[0042] Step S42: performing three-dimensional mapping of the motion planning result data to obtain a three-dimensional data block of the planning result;

[0043] Step S43: performing linear decision analysis and dimensionality reduction processing on the three-dimensional data block of the planning result to obtain two-dimensional planning result data;

[0044] Step S44: The two-dimensional planning result data is judged and processed with the preset two-dimensional data standard, and the two-dimensional planning result data that does not meet the two-dimensional data standard is eliminated, so as to generate standard two-dimensional planning result data.

[0045] The present invention acquires motion planning result data, performs three-dimensional mapping of the motion planning result data, and establishes clearer and more accurate three-dimensional data blocks of planning results, so that the changing trend of the motion planning result data can be better understood, thereby better discovering the law of the motion planning result and improving the efficiency and accuracy of data processing; the three-dimensional data blocks of planning results are subjected to linear decision analysis by a linear decision analysis algorithm, so that the motion planning result data blocks are mapped to a low-dimensional space, which helps to identify abnormal values ​​and outliers in the data blocks, facilitates subsequent processing, and improves the accuracy of data processing; the two-dimensional planning results are judged and processed with a preset two-dimensional data standard, the two-dimensional planning result data that does not meet the standard is eliminated, and standard two-dimensional planning result data is generated, so that the planning result data that meets the standard can be screened out, thereby improving the accuracy and reliability of the planning results, and the two-dimensional planning result data that does not meet the standard is eliminated, which can reduce the workload and time cost of subsequent processing.

[0046] Preferably, step S5 comprises the following steps:

[0047] Step S51: Dynamically time warping the standard two-dimensional planning result data to obtain a planning result timestamp;

[0048] Step S52: Perform time series sampling according to the planning result timestamp to generate a planning result time node;

[0049] Step S53: Perform time period synchronization processing on the planning result time node to generate the planning result time period;

[0050] Step S54: Use the preset time threshold and the planning result time period to perform proofreading. When the planning result time period is greater than the time threshold, generate a valid planning result time period. When the planning result time period is not greater than the time threshold, generate an invalid planning result time period, and remove the invalid planning result time period.

[0051] Step S55: Perform affine transformation on the effective planning result time period to obtain the total planning time data.

[0052] The present invention collects standard two-dimensional planning result data for dynamic time regularization, arranges the standard planning result data in chronological order, and can improve the accuracy and efficiency of subsequent analysis and processing; performs time series sampling on the planning result timestamp, samples the planning result data at a certain time interval, thereby obtaining the planning result time node, which is convenient for subsequent data analysis and processing; through time period synchronization processing, the planning result time node is synchronized according to a certain time period to obtain the planning result time period, which can better perform periodic analysis and processing on the planning result, realize subsequent periodicity-related functions, and reduce the time and cost of manual intervention; by using a preset time threshold and the planning result time period for proofreading processing, the effective planning result time period can be screened out, thereby improving the accuracy and reliability of the planning result, and at the same time, eliminating invalid planning result time periods can also reduce the workload and time cost of subsequent processing; the effective planning result time period is subjected to affine transformation processing to obtain the total planning time data, which can help understand the total planning time and improve the efficiency and quality of subsequent data processing.

[0053] Preferably, step S6 comprises the following steps:

[0054] Step S61: Obtaining planning total time data;

[0055] Step S62: using a timer to measure the planned total time data to obtain actual control time data;

[0056] Step S63: performing time axis segmentation processing on the actual control time data according to the preset time comparison priority to obtain a time control cycle;

[0057] Step S64: rounding the time control period to generate rounded time data.

[0058] The present invention obtains planned total time data, measures the planned total time data using a timer, and obtains actual control time data, thereby improving the accuracy and reliability of data and reducing data processing costs and risks; performs time axis segmentation processing on the actual control time data through a preset time comparison priority, cuts the continuous time axis into time discrete points, eliminates useless time control data, obtains the required time control period, reduces the coupling of data, and reduces the data storage space of the robot arm; performs integer processing on the time control period to generate rounded time data, reduces the risk of robot arm movement jitter during continuous trajectory movement, optimizes time management efficiency, and reduces data resource usage.

[0059] Preferably, step S7 comprises the following steps:

[0060] Step S71: using a difference processing formula to perform data difference processing on the rounded time data to obtain difference motion data;

[0061] Step S72: performing phase calculation on the difference motion data to obtain phase motion data;

[0062] Step S73: performing adversarial motion data generation processing on the phase motion data according to the adversarial neural network model to generate stable phase motion data;

[0063] Step S74: performing data cleaning processing on the steady phase motion data to obtain steady phase velocity data;

[0064] Step S75: performing Cartesian coordinate conversion on the difference motion data and the steady phase velocity data to generate a difference motion velocity vector and a steady phase velocity vector respectively;

[0065] Step S76: Perform vector superposition processing on the difference motion velocity vector and the steady phase velocity vector through a synthesis method to obtain an actual velocity value.

[0066] The present invention obtains rounded time data and performs data segmentation processing on the rounded time data using a difference processing formula, so as to more accurately grasp the utilization of time, improve the efficiency of time data processing, and reduce data redundancy; the difference motion data is subjected to phase solution calculation to obtain more accurate phase motion data, improve the accuracy and reliability of data analysis, and facilitate subsequent processing and identification; the phase motion data is subjected to adversarial motion data generation processing by an adversarial neural network model, so as to generate stable phase motion data, improve the quality and accuracy of the data, reduce the noise and interference of the data, make the data smoother and more stable, and improve the interpretability and visualization of the data; the fluctuation and error of the data are reduced by data cleaning processing, so that the data is smoother and more stable, obtain more accurate and reliable stable phase velocity data, improve the accuracy and credibility of data analysis and research, and achieve more accurate and stable motion control effect; the difference motion data and the stable phase velocity data are converted into a Cartesian coordinate system, and through the conversion, the difference motion velocity vector and the stable phase velocity vector can be generated respectively, so as to better describe the motion characteristics and motion laws, improve the data logic and accuracy, and achieve a stable motion control effect.

[0067] In this specification, a time rounding system based on velocity superposition is provided, comprising:

[0068] at least one processor; and,

[0069] a memory communicatively connected to the at least one processor; wherein,

[0070] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the time rounding method based on velocity superposition as described in any one of the above items.

[0071] The beneficial effect of the present invention is that the basic data of motion nodes are obtained by encoding the rocker commands of the robot arm. Since the basic data of motion nodes obtained are too large, direct identification and use will cause processing pressure on the robot arm system and slow down the response speed of the robot arm. Therefore, we need to clean the basic data of motion nodes to reduce the impact of useless, redundant and abnormal data, and predict and analyze the motion trajectory of the robot arm through the motion trajectory algorithm, so as to find out the optimal path for the robot arm to perform corresponding motion operations after receiving the command, and perform dimensionality reduction operation on the generated motion planning result data, converting it from three-dimensional to two-dimensional, and extracting the required motion-related features after the conversion. Data, useless data is eliminated, computing power and time are saved, and the load pressure of the system is reduced. The time slice time series sampling of the generated standard two-dimensional planning result data can accurately extract the time period of the robot movement, improve the accuracy and stability of the robot movement, and perform time rounding processing on the total planning time data. By rounding the minimum time period, the interference and error of the robot movement are reduced, and the accuracy and movement efficiency of the robot movement are improved. Then, the rounded time data after rounding is calculated by difference to obtain the robot movement value, and the robot movement value is vector decomposed, simulated and superimposed, and the motion data of each vector are superimposed to obtain the actual movement speed of the robot. Therefore, the time rounding method based on speed superposition of the present invention performs motion data analysis and processing on the obtained robot rocker command, the robot movement time is rounded, and then the processed robot movement data is vector superimposed to eliminate the robot arm jitter caused by the continuous trajectory movement of the robot, and at the same time improve the control speed and accuracy of the robot arm. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 A schematic diagram of the steps of a cloud platform-based smart grid resource scheduling and allocation method of the present invention;

[0073] Figure 2 for Figure 1 Detailed implementation steps of step S1 in FIG.

[0074] Figure 3 for Figure 1 Detailed implementation steps of step S2 in the flowchart;

[0075] Figure 4 for Figure 1 Detailed implementation steps of step S3 in FIG.

[0076] Figure 5 for Figure 1 Detailed implementation steps of step S7 in FIG.

[0077] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0078] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0079] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0080] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0081] The embodiment of the present application provides a time rounding method and system based on speed superposition, wherein the basic data of the motion node includes but is not limited to at least one of: robot arm motion change data and the like.

[0082] To achieve this, please refer to Figures 1 to 5 , a time rounding method based on velocity superposition, the method comprising the following steps:

[0083] Step S1: Obtain the remote sensing control instructions of the robot arm, encode the remote sensing control instructions, and obtain basic data of the motion node;

[0084] Step S2: Filter the basic data of the motion node to obtain the motion change data of the robot arm;

[0085] Step S3: Analyze and process the motion trajectory results according to the motion change data to generate motion planning result data;

[0086] Step S4: performing data dimensionality reduction processing according to the motion planning result data to generate standard two-dimensional planning result data;

[0087] Step S5: extracting planning time from the standard two-dimensional planning result data to generate total planning time data;

[0088] Step S6: performing robot arm time rounding processing on the planned total time data to generate rounded time data;

[0089] Step S7: Perform simulation superimposed motion processing on the rounded time data to generate actual motion values.

[0090] The present invention obtains the rocker control instructions of the mechanical arm and encodes the remote sensing control instructions, so that the mechanical arm can process and execute instructions faster, improve the processing speed of the mechanical arm, reduce the instruction storage space, and improve the storage efficiency. The encoded motion node basic data can be transmitted faster, reducing the data transmission time, improving the data transmission efficiency, improving the security of the data, preventing the data from being illegally obtained and tampered with, facilitating the subsequent data processing and analysis, and improving the efficiency and accuracy of data analysis; by filtering the motion node basic data, invalid data and erroneous data are removed, and the accuracy and reliability of the data are improved. According to the filtering processing of the motion node basic data, the time and cost of data processing can be reduced, the efficiency of subsequent processing can be improved, the motion node basic data can be made more standardized, and the influence of uncertain factors can be reduced; by performing motion trajectory result analysis on the motion change data, the pre-operation trajectory of the mechanical arm is analyzed, and the accuracy and reliability of the motion of the mechanical arm are improved. By performing operation trajectory analysis and calculation on the motion change data of the mechanical arm, the result is obtained. The optimal trajectory path reduces the speed loss and data calculation amount during the movement of the robot arm; by performing data dimensionality reduction processing on the motion planning result data, the complexity and redundancy of the data are reduced, the efficiency of data processing and analysis is improved, the three-dimensional data is reduced to two-dimensional data, the unstable factors in the motion planning result data are eliminated, and the motion planning result data is clearer and more obvious; by performing planning time extraction on the standard two-dimensional planning result data, the robot arm movement time period can be accurately extracted, the accuracy and stability of the robot arm movement can be improved, and the cost and time of manual intervention can be reduced; the total planning time data is rounded off, and by rounding off the minimum time period, the robot arm movement interference and error are reduced, the accuracy and efficiency of the robot arm movement are improved, and the movement speed control of the robot arm is improved; by performing simulation superposition motion processing on the time rounding data, the motion data is superimposed and simulated, thereby improving the actual speed, while eliminating the robot arm jitter caused by continuous instructions of the robot arm, and improving the efficiency and stability of the robot arm movement. Therefore, the speed superposition time rounding method of the present invention can clearly identify and round the continuous control instructions of the robot arm, improve the corresponding speed of the actual control of the robot arm, and at the same time eliminate the jitter of the robot arm caused by continuous instructions, saving tedious manual solution steps.

[0091] In the embodiment of the present invention, reference Figure 1 The above is a schematic flow chart of the steps of a time rounding method based on velocity superposition of the present invention. In this example, the steps of the time rounding method based on velocity superposition include:

[0092] Step S1: Obtain the remote sensing control instructions of the robot arm, encode the remote sensing control instructions, and obtain basic data of the motion node;

[0093] In an embodiment of the present invention, remote sensing control instructions of a medical robot arm are obtained, the remote sensing control instructions include up, down, left, and right, and the remote sensing control instructions are encoded in a common encoding method, and the encoding method includes ASCII encoding.

[0094] Step S2: Filter the basic data of the motion node to obtain the motion change data of the robot arm;

[0095] In an embodiment of the present invention, data filtering is performed on the basic data of the motion node, and the data filtering operation includes: through a customized motion change rule, only the data in the basic data of the motion node that has changed is retained, and then a motion change filter is created to automatically detect and filter useless, erroneous or duplicate motion change data sets in the data, and the entire motion change data set is processed in a batch manner to obtain the motion change data of the robotic arm.

[0096] Step S3: Analyze and process the motion trajectory results according to the motion change data to generate motion planning result data;

[0097] In the embodiment of the present invention, the motion change data is subjected to motion trajectory result analysis processing, wherein the analysis processing sub-step includes: determining a data set where the motion change data is located;

[0098] For the motion change data at each moment, add them up to get the motion trajectory data relative to the starting point. For example, if the initial position is (0,0) and the displacement at the first moment is (2,0), then the position at the first moment is (2,0);

[0099] Visualize the motion trajectory data and draw a motion trajectory diagram for further analysis;

[0100] Further analyze the movement rules and characteristics according to the movement trajectory diagram, such as speed change, turning, etc.;

[0101] According to the laws and characteristics of motion, motion trajectory planning is carried out to determine the object's future motion direction and path, and generate motion planning result data.

[0102] Step S4: performing data dimensionality reduction processing according to the motion planning result data to generate standard two-dimensional planning result data;

[0103] In the embodiment of the present invention, the motion planning result data is subjected to data dimensionality reduction processing, and the data dimensionality reduction processing sub-steps include:

[0104] Determine the data set that needs to be processed for dimensionality reduction;

[0105] Before dimensionality reduction, data needs to be preprocessed, including removing outliers, processing missing values, and normalizing.

[0106] Select dimensionality reduction methods, including principal component analysis (PCA), linear discriminant analysis (LDA), kernel principal component analysis (KPCA), etc.

[0107] Feature selection can be used to select variables that have a significant impact on the target variable and remove irrelevant variables, thereby reducing the dimension.

[0108] Verify the validity of the data after dimensionality reduction, such as cross-validating the classification model, observing the performance of statistical indicators, etc., to generate standard two-dimensional planning result data.

[0109] Step S5: extracting planning time from the standard two-dimensional planning result data to generate total planning time data;

[0110] In an embodiment of the present invention, the types of tasks and resources involved in the standard two-dimensional planning problem are first determined and listed in a table (or matrix) form. For example, tasks can be represented as rows of the problem, and resources can be represented as columns of the problem. In the result data, the number of resources required to complete each task is found and recorded in the corresponding cell, and then the completion time of each task is evaluated, and the evaluated time data is added to the cell corresponding to the original table, and the sorted task table is sorted in the order of completion time. A dynamic programming algorithm is used to scan the task table and determine the minimum time required to complete each task. For each task, a resource that meets the resource requirements and can complete the task in the minimum time is found among all the previous tasks, and then the completion time of the task is calculated from this time point. Repeat this operation until all tasks are completed. Finally, the total time required to complete all tasks is calculated as the total planning time data.

[0111] Step S6: performing robot arm time rounding processing on the planned total time data to generate rounded time data;

[0112] In an embodiment of the present invention, it is first necessary to set the time control cycle of the robot movement, that is, the time required for each cycle, for example, 1 second or 0.1 second, etc., and the planned total time data is rounded according to the time control cycle of the robot movement, usually by rounding down to ensure that the robot can complete all required motions. For example, if the robot movement cycle is 1 second, and the planned total time needs to be executed for 2.5 seconds, the rounded time data after rounding is 2 seconds. The rounded planned total time data is divided according to the robot movement cycle to generate rounded time data. For example, in the above example, the generated rounded time data is 2 cycles, that is, an integer multiple of 1 second.

[0113] Step S7: Perform simulation superimposed motion processing on the rounded time data to generate actual motion values.

[0114] In the embodiment of the present invention, the actual motion value of the type to be generated is determined, such as uniform linear motion, uniformly accelerated linear motion, circular motion, etc. According to the selected motion type, a corresponding mathematical model is established to determine the time interval required for the simulation, that is, the time step of the rounded time data, and the change in the position, speed or acceleration of the moving object is calculated in each time step. At the same time, the motion changes calculated in each time step are superimposed to obtain the change in position, speed or acceleration during the entire motion process. According to the superimposed motion changes, the actual motion value generated at each moment can be obtained.

[0115] Preferably, step S1 comprises the following steps:

[0116] Step S11: Obtaining robot arm rocker instructions;

[0117] Step S12: performing binary transcoding processing on the robot arm remote sensing instruction to generate a robot arm digital signal;

[0118] Step S13: using supervised learning to perform a robot arm behavior logic prediction on the robot arm digital signal, thereby generating robot arm judgment data;

[0119] Step S14: extracting data orientation features from the robot arm judgment data, thereby obtaining robot arm orientation data;

[0120] Step S15: Screen the motion node data of the robot arm orientation data to obtain motion node basic data.

[0121] The present invention can reduce the error and interference caused by the control system by performing binary transcoding processing on the robot arm rocker instructions, thereby improving the accuracy and stability of the control; use supervised learning to perform behavioral logic prediction on the robot arm digital signal, so as to judge the operating state of the robot arm, which is convenient for subsequent data processing; perform data-oriented feature extraction on the robot arm judgment data, and extract the robot arm movement-related data, which is convenient for subsequent processing and identification; perform data screening on the robot arm orientation data, so as to remove some repeated data and invalid data that interfere with the subsequent processing, thereby improving the accuracy of the subsequent processing.

[0122] As an example of the present invention, refer to Figure 2 As shown, Figure 2 Detailed implementation steps of step S1 in the flowchart, in this example, step S1 includes:

[0123] Step S11: Obtaining robot arm rocker instructions;

[0124] In an embodiment of the present invention, remote sensing control instructions are obtained when the robot arm moves;

[0125] Step S12: performing binary transcoding processing on the robot arm remote sensing instruction to generate a robot arm digital signal;

[0126] In an embodiment of the present invention, it is first necessary to determine the information contained in the remote sensing control instruction through the robot arm remote sensing instruction. For example, the robot arm movement direction, speed and other parameters can be specified by defining an instruction set, and each robot arm rocker instruction is converted into a binary number so that the computer can process and send it. For example, if there are 8 different instructions, a 3-bit binary number (000-111) can be used to encode each instruction, and a coding table is created to map each instruction to a corresponding binary number. In this way, when an instruction is received, the binary code corresponding to the instruction can be found and converted into a digital signal that the robot arm can understand. Use a digital circuit or other electronic device to convert the binary code into a series of digital pulses or other forms of signals to generate a robot arm digital signal.

[0127] Step S13: using supervised learning to perform a robot arm behavior logic prediction on the robot arm digital signal, thereby generating robot arm judgment data;

[0128] In an embodiment of the present invention, a sufficient amount of robotic arm digital signals and corresponding robotic arm behaviors, including input data (robotic arm digital signals) and output data (robotic arm behavior logic) are collected, and then the collected data are cleaned and preprocessed, and Fourier transform, wavelet transform and other methods are used to extract effective features from the collected data, a random forest model is selected for supervised learning, the training data is used to train the random forest model, a prediction model is generated, and the robotic arm digital signals are used for model prediction, thereby generating robotic arm judgment data.

[0129] Step S14: extracting data orientation features from the robot arm judgment data, thereby obtaining robot arm orientation data;

[0130] In an embodiment of the present invention, the robot arm orientation data is generated by performing feature extraction on the robot arm judgment data using the Fast algorithm, including extracting time domain and motion frequency domain features of the robot arm judgment data and evaluating the amount of information on the extracted features.

[0131] Step S15: Screen the motion node data of the robot arm orientation data to obtain motion node basic data.

[0132] In an embodiment of the present invention, the robot arm orientation data screening criteria, such as speed, acceleration, angular velocity, etc., are determined, the data stream is divided according to pre-defined motion nodes, the specified parameter value of each motion node is calculated respectively, and it is compared with the screening criteria. If it meets the criteria, the node data is retained, otherwise the node data is discarded, and for the filtered motion node data, its basic data, such as the mean, variance, maximum value, minimum value, etc., are extracted as needed to obtain the motion node basic data.

[0133] Preferably, step S2 comprises the following steps:

[0134] Step S21: collecting historical data of the motion node basic data to generate historical motion node basic data;

[0135] Step S22: extracting motion change feature data from historical motion node basic data to generate motion feature data;

[0136] Step S23: using deep learning to predict the motion change of the robot arm based on the motion feature data to generate motion dynamic change data;

[0137] Step S24: performing dynamic change data encoding processing on the motion dynamic change data to obtain a motion change data encoding segment;

[0138] Step S25: performing time series extraction processing on the motion change data encoding segment to generate a motion change time series;

[0139] Step S26: performing motion change data segment splicing processing on the motion change data encoding segments according to the motion change time sequence to generate initial robot arm motion change data;

[0140] Step S27: performing effective data extraction processing on the initial robot arm motion change data according to a preset effective time period to generate effective motion change data.

[0141] The present invention can help identify various data types and styles of historical motion nodes by collecting historical data of motion node basic data, improve data accuracy, and thus obtain historical motion node basic data, which can be used for subsequent processing of the robotic arm; extract motion change feature data of the historical motion node basic data, and extract features such as speed and acceleration of the robotic arm's historical motion change data, so as to facilitate subsequent motion prediction of the robotic arm; use deep learning in neural networks to perform motion change prediction on motion feature data, and more accurately and quickly obtain the dynamic motion change data of the robotic arm, saving computing power and time and reducing the load pressure of the system; perform motion change data encoding processing on the motion change data, and obtain motion change data encoding. The code segment can reduce the storage space and transmission bandwidth of data, and improve the transmission efficiency and speed of data; the time series extraction of the motion change data code segment generates the motion change time series, which can effectively understand the changing law of the robot arm motion, discover the key points and characteristics of the robot arm motion, and facilitate subsequent processing, while reducing the complexity of the data and improving the efficiency of data processing and analysis; splicing the motion change time series and the motion change data code segment to obtain more accurate initial robot arm motion change data, which is convenient for subsequent processing and identification, and more reliable for automated processing; the effective data extraction of the initial robot arm motion change data through the preset effective time period can eliminate invalid data segments, thereby obtaining effective motion change data and improving data accuracy.

[0142] As an example of the present invention, refer to Figure 3 As shown, Figure 3 Detailed implementation steps of step S2 in the embodiment are shown in the flowchart. In this embodiment, step S2 includes:

[0143] Step S21: collecting historical data of the motion node basic data to generate historical motion node basic data;

[0144] In the embodiment of the present invention, historical data is collected for the basic data of the motion node, wherein the sub-steps of historical data collection include:

[0145] Determine the sampling time interval based on the time period for collecting historical data;

[0146] At the set time interval, the basic data of the current motion nodes is obtained from the robot arm, and it is saved in the data storage device to generate historical collection data;

[0147] For historically collected data, some simple processing is performed, such as removing abnormal data, and calculating the average value and standard deviation of each historically collected data node;

[0148] By analyzing historical data, we can draw some conclusions, such as a certain node may have periodic changes, or the probability of a robot arm failure is greater when the value of a certain node is high, thereby generating basic data of historical motion nodes.

[0149] Step S22: extracting motion change feature data from historical motion node basic data to generate motion feature data;

[0150] In an embodiment of the present invention, the basic data of historical motion nodes are extracted to extract the mechanical arm motion data including motion period, motion intensity, motion type, motion frequency, etc., and the extracted mechanical arm motion data is cleaned to remove unnecessary data, and to process missing values ​​and abnormal values. The motion change feature data is extracted using a naive Bayesian machine learning algorithm, including extracting the mean value, standard deviation, maximum value, minimum value, rate of change, etc. After extracting the motion feature data, the feature data is further processed, such as normalization, smoothing, dimensionality reduction, etc., to generate motion feature data.

[0151] Step S23: using deep learning to predict the motion change of the robot arm based on the motion feature data to generate motion dynamic change data;

[0152] In an embodiment of the present invention, by collecting motion feature data of the robot arm, such as the position and posture of the end effector, joint angles, etc., these data can be collected by sensors, encoders and other devices, and these data are labeled, that is, the motion state of the robot arm at this time is labeled for each timestamp, and the motion feature data is preprocessed, and the preprocessing includes denoising, standardizing the data, etc., and selecting an appropriate deep learning model, such as a recurrent neural network (RNN), a long short-term memory neural network (LSTM), etc., and applying it to the problem of predicting the motion change of the robot arm. These models can automatically learn the motion characteristics of the robot arm, thereby predicting the future motion state, and use the prepared training data to train the model. The model parameters are continuously adjusted through the back propagation algorithm and the gradient descent optimization algorithm, so that the model can more accurately predict the motion change of the robot arm. During the training process, some techniques are used to improve the training effect of the model, such as batch normalization, dropout, etc., and the trained model is used to predict new test data to obtain the motion change data of the robot arm in the future. At the same time, some evaluation indicators are used to evaluate the prediction effect of the model, such as the root mean square error (Root Mean Square Error, RMSE), Mean Absolute Error (MAE) and other indicators to generate motion dynamic change data.

[0153] Step S24: performing dynamic change data encoding processing on the motion dynamic change data to obtain a motion change data encoding segment;

[0154] In the embodiment of the present invention, an appropriate sampling rate is selected to sample the original motion dynamic data to obtain a sufficiently dense sampling point sequence, and the sampling point sequence is preprocessed, including noise reduction, filtering, smoothing and other operations. The sampling point sequence is segmented using an appropriate time window to divide it into several continuous time periods, and the sampling point sequence in each time period is compressed and encoded to convert it into fewer code words or symbols. Some common encoding algorithms, such as Huffman coding, arithmetic coding, etc., can be used to combine the encoding results in each time period into a complete encoding segment, which represents the dynamic change characteristics of the entire motion data, and obtain the motion change data encoding segment.

[0155] Step S25: performing time series extraction processing on the motion change data encoding segment to generate a motion change time series;

[0156] In the embodiment of the present invention, the motion change data coding segment is cut into several small segments. For example, the data per second is divided into 10 small segments, each of which contains 6 data points. The data of each small segment is subjected to time series extraction processing using Fourier transform, and is converted into a set of digital sequences to represent the motion change situation in the small segment. The time series of all the small segments are merged into a complete time series to represent the change situation of the entire motion process, thereby generating a motion change time series.

[0157] Step S26: performing motion change data segment splicing processing on the motion change data encoding segments according to the motion change time sequence to generate initial robot arm motion change data;

[0158] In the embodiment of the present invention, the time series of the motion change data is confirmed according to the motion change time series, and data splicing is performed with the motion change data coding segment to generate initial robot arm motion change data.

[0159] Step S27: performing effective data extraction processing on the initial robot arm motion change data according to a preset effective time period to generate effective motion change data.

[0160] In an embodiment of the present invention, the effective working time period of the robot arm is determined based on experiments or actual conditions. For example, for a robot arm, only the motion data within a certain time period may be valid, and the data within other time periods may be that the robot arm is in an idle state or abnormal motion state, and has no reference value. The initial robot arm motion change data is preprocessed, such as removing abnormal values ​​or noise interference, eliminating unnecessary fluctuations by filtering, etc., and valid data is extracted based on the determined effective time period and the data after data preprocessing. Usually, the method of intercepting the time period can be used to exclude the data outside the effective time period to generate effective motion change data.

[0161] Preferably, step S3 comprises the following steps:

[0162] Step S31: Processing the motion change data into planning results, thereby generating planning result feature data;

[0163] Step S32: performing feature vector extraction processing on the planning result feature data, thereby generating a planning result feature vector;

[0164] Step S33: using the robot arm trajectory path coordinate analysis formula to perform motion trajectory coordinate analysis on the planning result feature vector to generate a motion trajectory path coordinate vector;

[0165] Step S34: performing vector superposition processing on the motion trajectory path coordinate vector to generate a motion planning result vector;

[0166] Step S35: Perform feature vector compression encoding processing on the planning result vector to generate motion planning result data.

[0167] The present invention processes the motion change data through planning results, so that the motion change data can be planned according to the result characteristics, generate planning result characteristic data, and reduce data redundancy; extract the data with vector characteristics in the data through feature vector extraction of the planning result characteristic data, generate the planning result characteristic vector, reduce the influence of noise data on the subsequent, prevent overfitting, and improve the reliability of the data; use the robot arm trajectory path coordinates to perform motion trajectory analysis on the planning result characteristic vector, obtain the specific path coordinate vector of the robot arm, reduce the collision risk that may occur during the movement of the robot arm, and improve the data accuracy; perform vector superposition on the motion trajectory path coordinate vector, improve the accuracy of the motion trajectory path coordinate vector, and better represent the motion trajectory path coordinate vector characteristics; perform feature vector compression encoding on the planning result vector, improve the efficiency of vector processing and analysis, improve the privacy and security of the vector, and reduce the storage space of the vector.

[0168] As an example of the present invention, refer to Figure 4 As shown, Figure 4 Detailed implementation steps of step S3 in the flowchart, in this example, step S3 includes:

[0169] Step S31: Processing the motion change data into planning results, thereby generating planning result feature data;

[0170] In the embodiment of the present invention, by collecting motion change data, the motion change data often needs to be matched with a predefined target curve to achieve more accurate motion planning. The fitting method can use quadratic curve fitting, cubic curve fitting, etc., and use dimensionality reduction technology to convert high-dimensional data into low-dimensional feature vectors. Commonly used dimensionality reduction methods include principal component analysis (PCA) and linear discriminant analysis (LDA), thereby generating planning result feature data.

[0171] Step S32: performing feature vector extraction processing on the planning result feature data, thereby generating a planning result feature vector;

[0172] In an embodiment of the present invention, feature vector extraction processing is performed on the planning result feature data, wherein the feature vector extraction processing includes: encoding the planning result feature data according to the Unicode encoding method, thereby converting it into a two-dimensional matrix, wherein the rows in the two-dimensional matrix represent data instances, and the columns represent different features, each row contains a separate data instance, and each column corresponds to a feature, performing minimum-maximum range scaling, regularization and standardization processing on different features, selecting the planning result feature value, and generating the planning result feature vector.

[0173] Step S33: using the robot arm trajectory path coordinate analysis formula to perform motion trajectory coordinate analysis on the planning result feature vector to generate a motion trajectory path coordinate vector;

[0174] In an embodiment of the present invention, the planning result feature vector uses the robot trajectory path analysis coordinate formula to perform motion trajectory coordinate analysis, wherein the specific operation of the motion trajectory coordinate analysis includes: extracting the position and speed information in the planning result feature vector, calculating the displacement of each moment corresponding to the robot path coordinate by the robot trajectory path coordinate analysis formula, taking the starting point as the reference point, calculating the position coordinate of each moment according to the displacement of each moment, performing interpolation processing between each two adjacent position coordinates, generating a set of relatively smooth intermediate points, so that the generated path is more continuous. Put all the position coordinates into an array to obtain the motion trajectory path coordinate vector.

[0175] Step S34: performing vector superposition processing on the motion trajectory path coordinate vector to generate a motion planning result vector;

[0176] In the embodiment of the present invention, the motion trajectory path coordinate vector is determined, which is recorded as p0, p1, ..., p n . Where p0 represents the starting point, p n Represents the end point. By finding the difference between two adjacent vectors, a new vector sequence is generated as δp0, δp1, ..., δp n-1 , where δp i =p i+1 -p i (i=0,1,……,n-1), starting from δp0, the accumulator is used to continuously add the differences of adjacent vectors to obtain a new vector sequence v0,v1,……,v n-1 .in Finally, v n-1 It is the motion planning result vector.

[0177] Step S35: Perform feature vector compression encoding processing on the planning result vector to generate motion planning result data.

[0178] In an embodiment of the present invention, the robot position information, velocity information, acceleration information, etc. in the planning result vector are compressed, including using a feature vector compression algorithm including a PCA algorithm, an LDA algorithm, etc. to generate motion planning result data.

[0179] Preferably, the robot arm trajectory path coordinate analysis formula in step S33 is as follows:

[0180]

[0181] In the formula, q represents the path coordinate of the robot arm, A i Indicates the length of the motion timestamp of the i-th robot terminal state, a i represents the axis motion angle of the i-th robot arm in the final state, B i Indicates the length of the initial motion timestamp of the i-th robot arm, b i represents the initial state axis motion angle of the i-th robot arm, e represents the power loss constant of the robot arm motion, s represents the robot arm motion space coordinate system, w represents the robot arm space coordinate projection value, m represents the upper limit of the robot arm motion acceleration, t represents the time from the initial state to the terminal state of the robot arm, θ represents the robot arm direction offset, and ε represents the abnormal adjustment value of the robot arm's path coordinate.

[0182] The present invention provides a robot arm trajectory path coordinate analysis formula, which fully considers the path coordinate q of the robot arm, the length A of the motion timestamp of the i-th robot arm terminal state, and the i , the final axis motion angle a of the i-th robot arm i , length of the initial motion timestamp of the i-th robot arm B i , the initial axis motion angle b of the i-th robot arm i , the power loss constant e of the robot movement, the space coordinate system s of the robot movement, the projection value w of the space coordinate of the robot movement, the upper limit m of the acceleration of the robot movement, the time t of the robot from the initial state to the terminal state, the direction offset θ of the robot, the abnormal adjustment value ε of the path coordinate of the robot, and the interaction between the motion timestamp length and the abnormal adjustment value ε of the path coordinate of the robot, a functional relationship has been formed Through the interaction between the motion timestamp length and the projection value of the robot space coordinate, the spatial feature extraction is performed while ensuring the data accuracy, and the robot motion space coordinate system is extracted. The robot direction offset is generated by using the robot terminal state axis motion angle and the robot terminal state motion timestamp length, which reduces data redundancy while ensuring data accuracy, saves computing power, and enables the calculation to converge quickly. The robot direction offset is corrected by the abnormal adjustment value ε of the robot path coordinate, so that the robot trajectory path coordinates are generated more accurately by calculating the acceleration upper limit of the motion power loss constant, thereby improving the accuracy and reliability of the robot motion trajectory. At the same time, the parameters such as the robot direction offset and the robot motion power loss constant in the formula can be adjusted according to actual conditions to adapt to different robot motion scenarios, thereby improving the applicability and flexibility of the algorithm. Preferably, step S4 includes the following steps:

[0183] Step S41: obtaining motion planning result data;

[0184] Step S42: performing three-dimensional mapping of the motion planning result data to obtain a three-dimensional data block of the planning result;

[0185] Step S43: performing linear decision analysis and dimensionality reduction processing on the three-dimensional data block of the planning result to obtain two-dimensional planning result data;

[0186] Step S44: The two-dimensional planning result data is judged and processed with the preset two-dimensional data standard, and the two-dimensional planning result data that does not meet the two-dimensional data standard is eliminated, so as to generate standard two-dimensional planning result data.

[0187] The present invention acquires motion planning result data, performs three-dimensional mapping of the motion planning result data, and establishes clearer and more accurate three-dimensional data blocks of planning results, so that the changing trend of the motion planning result data can be better understood, thereby better discovering the law of the motion planning result and improving the efficiency and accuracy of data processing; the three-dimensional data blocks of planning results are subjected to linear decision analysis by a linear decision analysis algorithm, so that the motion planning result data blocks are mapped to a low-dimensional space, which helps to identify abnormal values ​​and outliers in the data blocks, facilitates subsequent processing, and improves the accuracy of data processing; the two-dimensional planning results are judged and processed with a preset two-dimensional data standard, the two-dimensional planning result data that does not meet the standard is eliminated, and standard two-dimensional planning result data is generated, so that the planning result data that meets the standard can be screened out, thereby improving the accuracy and reliability of the planning results, and the two-dimensional planning result data that does not meet the standard is eliminated, which can reduce the workload and time cost of subsequent processing.

[0188] In an embodiment of the present invention, motion planning result data is obtained, and the motion planning result data includes the position, velocity and acceleration of the robot, as well as the execution trajectory, etc. The coordinate points in the motion planning data are replaced with three-dimensional vector points, so as to map the motion planning result data to a three-dimensional coordinate system to create a three-dimensional data block of the planning result. The three-dimensional data block is converted into two-dimensional planning result data, which is usually achieved using dimensionality reduction technology. For example, linear discriminant analysis (LDA) can be used to obtain principal components, and the motion planning data is projected into a two-dimensional subspace. The generated two-dimensional planning result data is further processed to determine whether it meets the preset two-dimensional data standard. If the data standard does not meet the requirements, the data can be eliminated until standard two-dimensional planning result data that meets the standard is generated.

[0189] Preferably, step S5 comprises the following steps:

[0190] Step S51: Dynamically time warping the standard two-dimensional planning result data to obtain a planning result timestamp;

[0191] Step S52: Perform time series sampling according to the planning result timestamp to generate a planning result time node;

[0192] Step S53: Perform time period synchronization processing on the planning result time node to generate the planning result time period;

[0193] Step S54: Use the preset time threshold and the planning result time period to perform proofreading. When the planning result time period is greater than the time threshold, generate a valid planning result time period. When the planning result time period is not greater than the time threshold, generate an invalid planning result time period, and remove the invalid planning result time period.

[0194] Step S55: Perform affine transformation on the effective planning result time period to obtain the total planning time data.

[0195] The present invention collects standard two-dimensional planning result data for dynamic time regularization, arranges the standard planning result data in chronological order, and can improve the accuracy and efficiency of subsequent analysis and processing; performs time series sampling on the planning result timestamp, samples the planning result data at a certain time interval, thereby obtaining the planning result time node, which is convenient for subsequent data analysis and processing; through time period synchronization processing, the planning result time node is synchronized according to a certain time period to obtain the planning result time period, which can better perform periodic analysis and processing on the planning result, realize subsequent periodicity-related functions, and reduce the time and cost of manual intervention; by using a preset time threshold and the planning result time period for proofreading processing, the effective planning result time period can be screened out, thereby improving the accuracy and reliability of the planning result, and at the same time, eliminating invalid planning result time periods can also reduce the workload and time cost of subsequent processing; the effective planning result time period is subjected to affine transformation processing to obtain the total planning time data, which can help understand the total planning time and improve the efficiency and quality of subsequent data processing.

[0196] In an embodiment of the present invention, the standard two-dimensional planning result data contains information of two dimensions, one dimension of information is time, and the other dimension of information is the operation status and state of the robotic arm. The standard two-dimensional planning result data is adjusted using a greedy algorithm and a dynamic programming algorithm, wherein the dynamic regularization step includes: representing the allocation of each resource as a time series: arranging the allocation of a resource in the two-dimensional planning result data in chronological order to obtain a time series representing the allocation of the resource, and calculating the distance between the two time series: using the DTW algorithm to calculate the distance between the two time series, the DTW algorithm uses a distance matrix to store the distance between the two time series, wherein the element in the i-th row and j-th column of the distance matrix represents the distance between the i-th element of the first time series and the j-th element of the second time series. You can choose different distance measurement methods, such as Euclidean distance, Manhattan distance, etc., for time alignment. Use the DTW algorithm to align two time series to find the similar parts between them. In the alignment process, you need to define a path to indicate how to align the two time series. The path is composed of a series of points, each of which represents an element in the first time series and the second time series. The starting point of the path is the point representing the first element of the first time series and the first element of the second time series, and the end point of the path is the point representing the last element of the first time series and the last element of the second time series. In the DTW algorithm, by calculating a cumulative distance The path is determined by the distance matrix, so as to obtain the normalized planning result timestamp, and some data points are extracted from the time series at fixed time intervals to obtain a sequence with a smaller interval. The obtained planning result timestamp is sampled at a certain sampling interval to generate the planning result time node, and the planning result time node is marked according to the required time period (such as week, month, quarter, etc.) to generate the planning result time period sequence, and the planning result time period is proofread by the preset time threshold to determine whether each planning result time period is valid or invalid. If the planning result time period is greater than the time threshold, the time period is considered to be valid, otherwise it is invalid. For invalid time periods, they can be removed from the time period sequence.

[0197] Preferably, step S6 comprises the following steps:

[0198] Step S61: Obtaining planning total time data;

[0199] Step S62: using a timer to measure the planned total time data to obtain actual control time data;

[0200] Step S63: performing time axis segmentation processing on the actual control time data according to the preset time comparison priority to obtain a time control cycle;

[0201] Step S64: rounding the time control period to generate rounded time data.

[0202] The present invention obtains planned total time data, measures the planned total time data using a timer, and obtains actual control time data, thereby improving the accuracy and reliability of data and reducing data processing costs and risks; performs time axis segmentation processing on the actual control time data through a preset time comparison priority, cuts the continuous time axis into time discrete points, eliminates useless time control data, obtains the required time control period, reduces the coupling of data, and reduces the data storage space of the robot arm; performs integer processing on the time control period to generate rounded time data, reduces the risk of robot arm movement jitter during continuous trajectory movement, optimizes time management efficiency, and reduces data resource usage.

[0203] As an example of the present invention, refer to Figure 4 As shown, Figure 1 Detailed implementation steps of step S6 in the flowchart, in this example, step S6 includes:

[0204] Step S61: Obtaining planning total time data;

[0205] In the embodiment of the present invention, the total planning time data of obtaining the robot arm movement is marked as mp;

[0206] Step S62: using a timer to measure the planned total time data to obtain actual control time data;

[0207] In an embodiment of the present invention, a timer is used to set the starting and ending times of the measurement time, and time measurement is performed. The measured time is compared and analyzed with the estimated time. If the actual control time is significantly different from the estimated time, the schedule is re-planned to perform time measurement until the actual control time is close to the preset time.

[0208] Step S63: performing time axis segmentation processing on the actual control time data according to the preset time comparison priority to obtain a time control cycle;

[0209] In an embodiment of the present invention, the actual control time data is compared according to the set time priority value, the actual control time data is divided according to the preset period, and sorted according to the priority. According to the actual control time data and the preset period, the specific time control period is determined. For example, data is obtained from the robot arm rocker command, and the task that receives the rocker command first is processed in order of priority, thereby generating a time control period.

[0210] Step S64: rounding the time control period to generate rounded time data.

[0211] In the embodiment of the present invention, the code formula involved in rounding the time control period includes:

[0212] t_plan_round=((int)(t_plan / ct))*ct

[0213] t_plan_round is the rounding time, ct is the abbreviation of cycle_time, cycle_time is the minimum time slice allowed by the control system, such as 1 millisecond, the control system generates control information once every 1ms. The total planned time divided by cycle_time is the number of control cycles required. For example, if the total planned time is 1265ms and cycle_time is 2ms, the control cycle is 632.5 cycles. The control cycle cannot be a decimal. At this time, the control cycle needs to be rounded up. If the rounding down strategy is adopted, the control cycle is 632, and the total time after rounding is 1264ms.

[0214] Preferably, step S7 comprises the following steps:

[0215] Step S71: using a difference processing formula to perform data difference processing on the rounded time data to obtain difference motion data;

[0216] Step S72: performing phase calculation on the difference motion data to obtain phase motion data;

[0217] Step S73: performing adversarial motion data generation processing on the phase motion data according to the adversarial neural network model to generate stable phase motion data;

[0218] Step S74: performing data cleaning processing on the steady phase motion data to obtain steady phase velocity data;

[0219] Step S75: performing Cartesian coordinate conversion on the difference motion data and the steady phase velocity data to generate a difference motion velocity vector and a steady phase velocity vector respectively;

[0220] Step S76: Perform vector superposition processing on the difference motion velocity vector and the steady phase velocity vector through a synthesis method to obtain an actual velocity value.

[0221] The present invention obtains rounded time data and performs data segmentation processing on the rounded time data using a difference processing formula, so as to more accurately grasp the utilization of time, improve the efficiency of time data processing, and reduce data redundancy; the difference motion data is subjected to phase solution calculation to obtain more accurate phase motion data, improve the accuracy and reliability of data analysis, and facilitate subsequent processing and identification; the phase motion data is subjected to adversarial motion data generation processing by an adversarial neural network model, so as to generate stable phase motion data, improve the quality and accuracy of the data, reduce the noise and interference of the data, make the data smoother and more stable, and improve the interpretability and visualization of the data; the fluctuation and error of the data are reduced by data cleaning processing, so that the data is smoother and more stable, obtain more accurate and reliable stable phase velocity data, improve the accuracy and credibility of data analysis and research, and achieve more accurate and stable motion control effect; the difference motion data and the stable phase velocity data are converted into a Cartesian coordinate system, and through the conversion, the difference motion velocity vector and the stable phase velocity vector can be generated respectively, so as to better describe the motion characteristics and motion laws, improve the data logic and accuracy, and achieve a stable motion control effect.

[0222] As an example of the present invention, refer to Figure 5 As shown, Figure 1 Detailed implementation steps of step S7 in the flowchart, in this example, step S7 includes:

[0223] Step S71: using a difference processing formula to perform data difference processing on the rounded time data to obtain difference motion data;

[0224] In the embodiment of the present invention, the difference processing formula is used to perform difference processing on the rounding time, wherein the difference processing includes: obtaining the speed, acceleration, and distance information corresponding to the rounding time t_plan_round, and using the original planning result mp to obtain the time point information of the rounding time:

[0225] Ve_round=mp(t_plan_round);

[0226] Ae_round=mp(t_plan_round);

[0227] L_round = mp(t_plan_round);

[0228] Where Ve_round is the speed, Ae_round is the acceleration, and L_round is the distance information. The speed, acceleration, and distance information are respectively subjected to difference processing to obtain the speed difference V_diff, the acceleration difference A_diff, and the distance difference L_diff.

[0229] Step S72: performing phase calculation on the difference motion data to obtain phase motion data;

[0230] In the embodiment of the present invention, the velocity difference V_diff, the acceleration difference A_diff, and the distance difference L_diff are subjected to superimposed phase processing to generate phase motion data, wherein the superimposed phase processing process includes:

[0231] Velocity phase solution: If (V_diff>0) V_sign=1.0; Else V_sign=-1.0;

[0232] Acceleration phase solution: If (A_diff>0) A_sign = 1.0; Else A_sign = -1.0;

[0233] Distance phase solution: If (L_diff>0) L_sign = 1.0; Else L_sign = -1.0;

[0234] The velocity phase V_sign, acceleration phase A_sign, and distance phase L_sign are obtained.

[0235] Step S73: performing adversarial motion data generation processing on the phase motion data according to the adversarial neural network model to generate stable phase motion data;

[0236] In an embodiment of the present invention, the original phase motion data is preprocessed to eliminate noise and outliers in the data, and normalization is performed to ensure the stability and effectiveness of the subsequent GAN model training and data generation process. The specific preprocessing method may include: removing noise in the motion data (such as filtering, removing interference, etc.), smoothing the data, ensuring the smoothness of the data, normalization, etc. In the GAN model, there are two networks: a generator and a discriminator. The function of the generator is to convert the noise data into phase motion data with similar time-frequency characteristics to the original data, and the function of the discriminator is to determine whether the generated data is real data or generated data. During the training process, the generator and the discriminator are alternately iteratively updated until the data generated by the generator can deceive the discriminator so that the discriminator cannot effectively identify the generated data. After the GAN model is trained, the noise data can be used as the input of the generator to generate new phase motion data. When generating new data, different data sequences can be generated by setting different initial conditions to generate stable phase motion data.

[0237] Step S74: performing data cleaning processing on the steady phase motion data to obtain steady phase velocity data;

[0238] In an embodiment of the present invention, data cleaning is performed on the steady phase motion data to eliminate invalid and repeated data, and these noises are removed by some smoothing methods, such as using sliding average or Kalman filter methods, to obtain steady phase velocity data.

[0239] Step S75: performing Cartesian coordinate conversion on the difference motion data and the steady phase velocity data to generate a difference motion velocity vector and a steady phase velocity vector respectively;

[0240] In the embodiment of the present invention, it is necessary to determine the direction of the Cartesian coordinate system, use a right-handed coordinate system, where the x-axis points to the right, the y-axis points upward, and the z-axis points outside the screen, determine the position of each sample point, and represent it as a three-dimensional vector in the Cartesian coordinate system. For each sample point in the differential motion data, it can be represented as two three-dimensional vectors, namely, the differential motion lateral velocity vector and the differential motion longitudinal velocity vector. The differential motion lateral velocity vector and the differential motion longitudinal velocity vector are added to obtain the differential motion velocity vector. This vector represents the differential motion velocity of each sample point, and is represented as a three-dimensional vector in the Cartesian coordinate system to obtain the differential motion velocity vector and the steady phase velocity vector.

[0241] Step S76: Perform vector superposition processing on the difference motion velocity vector and the steady phase velocity vector through a synthesis method to obtain an actual velocity value.

[0242] In the embodiment of the present invention, a vector superposition process is performed on the difference motion velocity vector and the steady phase velocity vector by a synthesis method to obtain an actual velocity value, wherein the vector superposition process includes:

[0243] The original planned speed of mp at point t_i is V_1;

[0244] The velocity of the velocity superposition curve at point t_i is V_2;

[0245] Then the actual speed at point t_i is v_i=V_1+V_sign*V_2.

[0246] In this specification, a time rounding system based on velocity superposition is provided, characterized in that:

[0247] at least one processor; and,

[0248] a memory communicatively connected to the at least one processor; wherein,

[0249] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the time rounding method based on velocity superposition as described in any one of the above items.

[0250] The beneficial effect of the present invention is that the basic data of motion nodes are obtained by encoding the rocker commands of the robot arm. Since the basic data of motion nodes obtained are too large, direct identification and use will cause processing pressure on the robot arm system and slow down the response speed of the robot arm. Therefore, we need to clean the basic data of motion nodes to reduce the impact of useless, redundant and abnormal data, and predict and analyze the motion trajectory of the robot arm through the motion trajectory algorithm, so as to find out the optimal path for the robot arm to perform corresponding motion operations after receiving the command, and perform dimensionality reduction operation on the generated motion planning result data, converting it from three-dimensional to two-dimensional, and extracting the required motion-related features after the conversion. Data, useless data is eliminated, computing power and time are saved, and the load pressure of the system is reduced. The time slice time series sampling of the generated standard two-dimensional planning result data can accurately extract the time period of the robot movement, improve the accuracy and stability of the robot movement, and perform time rounding processing on the total planning time data. By rounding the minimum time period, the interference and error of the robot movement are reduced, and the accuracy and movement efficiency of the robot movement are improved. Then, the rounded time data after rounding is calculated by difference to obtain the robot movement value, and the robot movement value is vector decomposed, simulated and superimposed, and the motion data of each vector are superimposed to obtain the actual movement speed of the robot. Therefore, the time rounding method based on speed superposition of the present invention performs motion data analysis and processing on the obtained robot rocker command, the robot movement time is rounded, and then the processed robot movement data is vector superimposed to eliminate the robot arm jitter caused by the continuous trajectory movement of the robot, and at the same time improve the control speed and accuracy of the robot arm.

[0251] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0252] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A time rounding method based on velocity superposition, characterized in that: The method comprises: Step S1: Obtain the remote sensing control instructions of the robot arm, encode the remote sensing control instructions, and obtain basic data of the motion node; Step S2: Filter the basic data of the motion node to obtain the motion change data of the robot arm; Step S3: Analyze and process the motion trajectory results according to the motion change data to generate motion planning result data; Step S4: performing data dimensionality reduction processing according to the motion planning result data to generate standard two-dimensional planning result data; Step S5: extracting planning time from the standard two-dimensional planning result data to generate total planning time data; Step S6: performing robot arm time rounding processing on the planned total time data to generate rounded time data; Step S7: Perform simulation superposition motion processing on the rounded time data to generate actual motion values; Step S7 includes the following steps: Step S71: using a difference processing formula to perform data difference processing on the rounded time data to obtain difference motion time data; Step S72: performing phase calculation on the difference motion time data to obtain phase motion data; Step S73: performing adversarial motion data generation processing on the phase motion data according to the adversarial neural network model to generate stable phase motion data; Step S74: performing data cleaning processing on the steady phase motion data to obtain steady phase velocity data; Step S75: performing Cartesian coordinate conversion on the difference motion data and the steady phase velocity data to generate a difference motion velocity vector and a steady phase velocity vector respectively; Step S76: Perform vector superposition processing on the difference motion velocity vector and the steady phase velocity vector through a synthesis method to obtain an actual velocity value.

2. The time rounding method based on velocity superposition according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: Obtaining robot arm rocker instructions; Step S12: performing binary transcoding processing on the robot arm remote sensing instruction to generate a robot arm digital signal; Step S13: using supervised learning to perform a robot arm behavior logic prediction on the robot arm digital signal, thereby generating robot arm judgment data; Step S14: extracting data orientation features from the robot arm judgment data, thereby obtaining robot arm orientation data; Step S15: Screen the motion node data of the robot arm orientation data to obtain motion node basic data.

3. The time rounding method based on velocity superposition according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: collecting historical data of the motion node basic data to generate historical motion node basic data; Step S22: extracting motion change feature data from historical motion node basic data to generate motion feature data; Step S23: using deep learning to predict the motion change of the robot arm based on the motion feature data to generate motion dynamic change data; Step S24: performing dynamic change data encoding processing on the motion dynamic change data to obtain a motion change data encoding segment; Step S25: performing time series extraction processing on the motion change data encoding segment to generate a motion change time series; Step S26: performing motion change data segment splicing processing on the motion change data encoding segments according to the motion change time sequence to generate initial robot arm motion change data; Step S27: performing effective data extraction processing on the initial robot arm motion change data according to a preset effective time period to generate effective motion change data.

4. The time rounding method based on velocity superposition according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: Processing the motion change data into planning results, thereby generating planning result feature data; Step S32: performing feature vector extraction processing on the planning result feature data, thereby generating a planning result feature vector; Step S33: using the robot arm trajectory path coordinate analysis formula to perform motion trajectory coordinate analysis on the planning result feature vector to generate a motion trajectory path coordinate vector; Step S34: performing vector superposition processing on the motion trajectory path coordinate vector to generate a motion planning result vector; Step S35: Perform feature vector compression encoding processing on the planning result vector to generate motion planning result data.

5. The time rounding method based on velocity superposition according to claim 4, characterized in that: The formula for analyzing the robot arm trajectory path coordinates in step S33 is as follows: Among them, q represents the path coordinate of the robot arm, A i Indicates the length of the motion timestamp of the i-th robot terminal state, a i represents the axis motion angle of the i-th robot arm in the final state, B i Indicates the length of the initial motion timestamp of the i-th robot arm, b i represents the initial state axis motion angle of the i-th robot arm, e represents the power loss constant of the robot arm motion, s represents the robot arm motion space coordinate system, w represents the robot arm space coordinate projection value, m represents the upper limit of the robot arm motion acceleration, t represents the time from the initial state to the terminal state of the robot arm, θ represents the robot arm direction offset, and ε represents the abnormal adjustment value of the robot arm's path coordinate.

6. The time rounding method based on velocity superposition according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: obtaining motion planning result data; Step S42: performing three-dimensional mapping of the motion planning result data to obtain a three-dimensional data block of the planning result; Step S43: performing linear decision analysis and dimensionality reduction processing on the three-dimensional data block of the planning result to obtain two-dimensional planning result data; Step S44: The two-dimensional planning result data is judged and processed with the preset two-dimensional data standard, and the two-dimensional planning result data that does not meet the two-dimensional data standard is eliminated, so as to generate standard two-dimensional planning result data.

7. The time rounding method based on velocity superposition according to claim 1, characterized in that: Step S5 includes the following steps: Step S51: Dynamically time warping the standard two-dimensional planning result data to obtain a planning result timestamp; Step S52: Perform time series sampling according to the planning result timestamp to generate a planning result time node; Step S53: Perform time period synchronization processing on the planning result time node to generate the planning result time period; Step S54: Use the preset time threshold and the planning result time period to perform proofreading. When the planning result time period is greater than the time threshold, generate a valid planning result time period. When the planning result time period is not greater than the time threshold, generate an invalid planning result time period, and remove the invalid planning result time period. Step S55: Perform affine transformation on the effective planning result time period to obtain the total planning time data.

8. The time rounding method based on velocity superposition according to claim 1, characterized in that: The planning total time data is rounded off, and step S6 includes the following steps: Step S61: Obtaining planning total time data; Step S62: using a timer to measure the planned total time data to obtain actual control time data; Step S63: performing time axis segmentation processing on the actual control time data according to the preset time comparison priority to obtain a time control cycle; Step S64: rounding the time control period to generate rounded time data.

9. A time rounding system based on velocity superposition, characterized in that: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor. The computer program is executed by the at least one processor so that the at least one processor can perform the time rounding method based on velocity superposition according to any one of claims 1 to 8.

Images