Collaborative Robot Control Method, System and Storage Medium Based on Deep Learning

Abstract

The present invention relates to the field of robot control technology, and specifically discloses a collaborative robot control method, system and storage medium based on deep learning. The method includes: evaluating the control operation accuracy, determining the coordination control degree, and optimizing the control. First, evaluate the control operation accuracy of the collaborative robot, and comprehensively obtain the coordination control degree of the collaborative robot by combining the visual application data of the collaborative robot, match it with the set coordination control threshold, and finally optimize the control of the collaborative robot through a deep learning model, so as to accurately control the actions and speeds of the robot, thereby improving the coordination ability of the collaborative robot; it can ensure that the collaborative robot has higher accuracy and stability when performing tasks.

Description

Technical Field

[0001] The present invention relates to the technical field of robot control, and specifically to a collaborative robot control method, system, and storage medium based on deep learning. Background Art

[0002] Currently, in order to ensure the safety of interaction with humans, collaborative robots usually need to make compromises in terms of speed and strength, which results in a decrease in the operation accuracy of collaborative robots, and usually has a small load capacity; and due to the certain difficulty of collaborative operations, high skill requirements are imposed on operators; at the same time, in some complex and delicate operations, the reliability of collaborative robots is not high. Therefore, during the process of coordinated work of collaborative robots, certain monitoring and preventive measures need to be taken to improve the coordinated work ability of collaborative robots.

[0003] For example, in the invention patent with the publication number CN105988418B, the above application provides a robot. A user can newly manufacture a driver corresponding to an external machine used and use it; the robot executes a series of operations on each operation point according to operation data based on points as the main body; the external machine assembled to the robot executes the present operation on each operation point through a series of operations; the controller controls the external machine according to control data, and the control data is formed by arranging structured point blocks including a series of operations for each operation point; the external machine uses the driver to perform the conversion of the form of data transmitted and received between the external machine and the controller; the driver generation unit manufactures driver data representing the conversion content of the form of the data in the external machine driver; the driver selection component cooperates with the external machine to select the driver data used in the external machine driver.

[0004] However, in the process of implementing the above application embodiment, it is found that the above technology has at least the following technical problems: when describing the external machine, the above application does not obtain data in terms of graphical interaction with sensors or cameras, and only describes them simply as sensors or cameras, and the obtained data is too monotonous to support the effective control of the robot. Summary of the Invention

[0005] Aiming at the deficiencies of the prior art, the present invention provides a collaborative robot control method, system, and storage medium based on deep learning, which can effectively solve the problems involved in the above background art.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: In the first aspect of the present invention, a collaborative robot control method based on deep learning is provided, including: Control operation accuracy evaluation: Obtain the control operation correlation information of the collaborative robot and evaluate the control operation accuracy of the collaborative robot; Coordination control degree determination: Collect the visual application data of the collaborative robot through the graphical user interface of the collaborative robot, and comprehensively consider the control operation accuracy of the collaborative robot to obtain the coordination control degree of the collaborative robot; Control optimization: Match the coordination control degree of the collaborative robot with the set coordination control threshold. If the coordination control degree of the collaborative robot is less than the coordination control threshold, control optimization of the collaborative robot is performed through a deep learning model.

[0007] As a further method, the process of controlling and optimizing the collaborative robot through the deep learning model is as follows: Match the coordination control degree of the collaborative robot with the coordination control threshold. If the coordination control degree of the collaborative robot is less than the coordination control threshold, control optimization of the collaborative robot is performed through a deep learning model; If the coordination control degree of the collaborative robot is greater than or equal to the coordination control threshold, there is no need to perform control optimization on the collaborative robot.

[0008] As a further method, the specific analysis process of the set coordination control threshold is as follows: Obtain the historical coordination data of the collaborative robot, where the historical coordination data of the collaborative robot includes the number of guiding and positioning times and the number of data matrix decoding times. At the same time, count the number of successful guiding and positioning times and the number of successful data matrix decoding times of the collaborative robot within the set historical coordination period, and set the coordination control threshold through data processing.

[0009] As a further method, the specific analysis process of the coordination control degree of the collaborative robot is as follows: Collect the visual application data of the collaborative robot through the graphical user interface of the collaborative robot, where the visual application data includes the number of recognized targets; According to the number of recognized targets of the collaborative robot, extract the length and width of each recognized target of the collaborative robot; According to the length and width of each recognized target of the collaborative robot, obtain the area of each recognized target of the collaborative robot through the area algorithm, thereby constructing a change curve of the recognized target area of the collaborative robot, extracting the maximum area change angle value from the change curve of the recognized target area, and at the same time, extracting the maximum area change angle definition value from the robot control database and performing comprehensive analysis and processing with the maximum area change angle value to obtain the change influence index of the visual recognition of the collaborative robot ; The comprehensive analysis of the coordination control degree of the collaborative robot is specifically analyzed by the following formula:

[0010] ;

[0011] Where Denoted as the coordination control degree of the collaborative robot, Denoted as the control operation accuracy of the collaborative robot, Denoted as the change impact index of the visual recognition of the collaborative robot, Denoted as the weight factor corresponding to the predefined control operation accuracy, Denoted as the weight factor corresponding to the predefined change impact index of the visual recognition of the collaborative robot, where e is the natural constant.

[0012] As a further method, the control operation accuracy of the collaborative robot, the specific analysis process is as follows: According to the control operation association information of the collaborative robot, obtain the information belonging to the structural characteristics of the collaborative robot and the information belonging to the control algorithm of the collaborative robot from it; comprehensively analyze the influence degree coefficient of the structural characteristics of the collaborative robot and the influence degree coefficient of the control algorithm of the collaborative robot to obtain the control operation accuracy of the collaborative robot.

[0013] The second aspect of the present invention provides a system for a collaborative robot control method based on deep learning, including: a control operation accuracy evaluation module, used to obtain the control operation association information of the collaborative robot and evaluate the control operation accuracy of the collaborative robot; a coordination control degree determination module, used to collect the visual application data of the collaborative robot through the graphical user interface of the collaborative robot, and comprehensively consider the control operation accuracy of the collaborative robot to obtain the coordination control degree of the collaborative robot; a control optimization module, used to match the coordination control degree of the collaborative robot with a set coordination control threshold. If the coordination control degree of the collaborative robot is less than the coordination control threshold, the collaborative robot is controlled and optimized through a deep learning model.

[0014] The third aspect of the present invention provides a computer-readable storage medium for storing a program, and when the program is executed by a processor, it implements a collaborative robot control method based on deep learning.

[0015] Compared with the prior art, the embodiments of the present invention have at least the following advantages or beneficial effects:

[0016] (1) By providing a collaborative robot control method, system and storage medium based on deep learning, the present invention first evaluates the control operation accuracy of the collaborative robot, obtains the coordination control degree of the collaborative robot, matches it with a set coordination control threshold, and controls and optimizes the collaborative robot through a deep learning model. By precisely controlling the actions and speeds of the robot, unnecessary waiting time and operation time can be reduced, thereby improving the overall production efficiency; it can ensure that the collaborative robot has higher accuracy and stability when performing tasks; by adjusting control parameters and algorithms, the flexibility and adaptability of the collaborative robot on the production line are improved;

[0017] (2) By obtaining the control operation correlation information of the collaborative robot and evaluating the control operation accuracy of the collaborative robot, the performance and efficiency of the collaborative robot during task execution can be intuitively understood, improving the collaborative production efficiency of the collaborative robot; and it can enable the collaborative robot to better understand the intentions and needs of humans, thus completing collaborative tasks more efficiently.

[0018] (3) Through the graphical user interface of the collaborative robot, the visual application data of the collaborative robot is collected, and combined with the control operation accuracy of the collaborative robot, the coordinated control degree of the collaborative robot is obtained. By optimizing the coordinated control strategy, the collaborative robot can be made to more flexibly adapt to various complex scenarios, improving its applicability in different industries and application fields; at the same time, ensuring that the collaborative robot has higher safety performance when interacting with humans and avoiding the occurrence of accidental injuries. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The present invention will be further described with reference to the accompanying drawings. However, the embodiments in the drawings do not constitute any limitation to the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained according to the following drawings.

[0020] Figure 1 It is a schematic flow chart of the method steps of the present invention;

[0021] Figure 2 It is a schematic diagram of the connection of the system modules of the present invention;

[0022] Figure 3 It is the recognition target area change curve involved in the present invention.

[0023] Reference numerals: 1. Area growth straight line; 2. First horizontal straight line. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0024] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0025] Refer to Figure 1 As shown, the first aspect of the present invention provides a control method for a collaborative robot based on deep learning, including: Control operation accuracy evaluation: Obtain the control operation correlation information of the collaborative robot and evaluate the control operation accuracy of the collaborative robot.

[0026] Specifically, the influence degree coefficient of the structural characteristics of the collaborative robot, the specific analysis process is:

[0027] According to the control operation correlation information of the collaborative robot, obtain the information belonging to the structural characteristics of the collaborative robot and the information belonging to the control algorithm of the collaborative robot therefrom.

[0028] From the information belonging to the structural characteristics of the collaborative robot, extract the number of joints of the collaborative robot, its own weight, the average load within the control period, and the weight of the end effector, where the control period is set as the duration required for the collaborative robot to receive a control instruction, execute the instruction, and return the execution result; the number of joints, its own weight, and the weight of the end effector of the collaborative robot can all be directly obtained from the technical specifications of the collaborative robot; extract the load values at each control time point from the storage device carried by the collaborative robot itself, and comprehensively process the load at each control time point and the duration corresponding to the control period to obtain the average load of the collaborative robot within the control period.

[0029] Subtract the weight of the end effector from the average load of the collaborative robot within the control period to obtain the effective load value of the collaborative robot within the control period, and conduct comprehensive analysis and processing with the set effective load threshold to obtain the effective load impact index of the collaborative robot, where the set effective load threshold can be extracted from the storage device carried by the collaborative robot itself.

[0030] Extract the influencing factors corresponding to the defined number of joints, the defined own weight, and the effective load impact index from the robot control database.

[0031] The data in the above robot control database is obtained by fitting based on the statistically obtained multiple historical collaborative robot control data. For example, for the defined number of joints, through different control processes of the collaborative robot, different numbers of joints of the collaborative robot are obtained, and the control manager determines whether the control state of the collaborative robot under different numbers of joints meets the requirement standards. The number of joints corresponding to the control state of the collaborative robot that meets the requirement standards is averaged to obtain the defined number of joints. In this way, the data in the robot control database is obtained, and these data will vary due to factors such as the working environment, task requirements, and the own state of the collaborative robot. This embodiment does not make special limitations on this.

[0032] Comprehensively analyze the influence degree coefficient of the structural characteristics of the collaborative robot. In a specific embodiment, it can be obtained by evaluating the various performances of the collaborative robot, including its motion range, load capacity, accuracy, repeat positioning accuracy, etc. These performance parameters can reflect the influence of the structural characteristics of the robot on its working ability, so as to obtain the influence degree coefficient of the structural characteristics of the collaborative robot.

[0033] The influence degree coefficient of the structural characteristics of the above collaborative robot is, in this embodiment, obtained through comprehensive analysis of the number of joints, its own weight, and the effective load influence index of the collaborative robot, so as to determine the value of the influence degree of the structural characteristics of the collaborative robot. In this embodiment, a more accurate calculation method is used to obtain it, and the specific expression is:

[0034] ;

[0035] where represents the influence degree coefficient of the structural characteristics of the collaborative robot. In this embodiment, if the number of joints of the collaborative robot is large, it will cause the increase of its own weight, which will affect the load capacity and stability of the collaborative robot. The relationship between these three parameters restricts each other. Therefore, correlation analysis is carried out on this expression to ensure that the collaborative robot can meet the structural requirements in practical applications;

[0036] represents the number of joints of the collaborative robot, which refers to the number of joints that can be independently controlled in its mechanical structure; common collaborative robots have different numbers of joints, including but not limited to 6-axis robots, 7-axis robots, and underactuated robots;

[0037] represents the defined number of joints, which refers to the maximum value allowed by the number of joints.

[0038] In this embodiment, it is set that the number of joints of the collaborative robot is 4, the defined number of joints is 6, and the correction factor corresponding to the number of joints is 0.5. Then, in this expression the operation result of the part is 0.33;

[0039] represents the own weight of the collaborative robot, which refers to the weight of the collaborative robot body. This determines whether the collaborative robot can easily move and change the working position, or whether additional manual assistance is required to complete the moving work;

[0040] represents the defined own weight, which refers to the maximum value specified by the own weight;

[0041] represents the effective load influence index of the collaborative robot. The effective load of the collaborative robot refers to the weight that the collaborative robot can carry, and the effective load influence index refers to the degree of its influence on the motion accuracy and stability of the collaborative robot;

[0042] represents the correction factor corresponding to the predefined number of joints, represents the correction factor corresponding to the predefined own weight, Denoted as the influence factor corresponding to the payload influence index, where the correction factors corresponding to the number of joints and the self-weight are both obtained by establishing an accurate dynamic model for the collaborative robot, calculating the joint distribution, mass distribution, and inertia parameters of the collaborative robot, and these parameters can be used as correction factors. The dynamic model can be obtained through theoretical derivation; the influence factor corresponding to the payload influence index is obtained by using robot dynamics simulation software to construct a virtual model of the collaborative robot, simulating the motion and working states of the robot under different payloads in the simulation environment, analyzing the influence of the payload on the robot's performance indicators through the simulation data, and calculating the corresponding influence factor; in this embodiment, the value range of the correction factor corresponding to the number of joints is from 0.47 to 0.52; the value range of the correction factor corresponding to the self-weight is from 0.37 to 0.47; the value range of the influence factor corresponding to the payload influence index is from 0.45 to 0.6.

[0043] Furthermore, the specific analysis process of the influence degree coefficient of the control algorithm of the collaborative robot is as follows:

[0044] From the information of the control algorithm of the collaborative robot, the number of temperature sensors and pressure sensors carried by the collaborative robot itself is extracted, and the number of sensors can be obtained from the user manual of the collaborative robot.

[0045] The control period is divided into each control time point, and the temperature values corresponding to each temperature sensor of the collaborative robot at each control time point and the pressure values corresponding to each pressure sensor at each control time point are extracted. The data of the sensors are also obtained from the user manual of the collaborative robot, including the temperature values and pressure values, and are compared with the preset temperature threshold and pressure threshold respectively, and the temperature average error value and pressure average error value of the collaborative robot at each control time point are integrated.

[0046] The above preset temperature threshold and pressure threshold are extracted from the robot database.

[0047] Multiply the temperature average error value of the collaborative robot at each control time point by the influence factor corresponding to the set temperature compensation to obtain the temperature influence index of the collaborative robot at each control time point; multiply the pressure average error value of the collaborative robot at each control time point by the influence factor corresponding to the set pressure compensation to obtain the pressure influence index of the collaborative robot at each control time point.

[0048] The above temperature compensation and pressure compensation refer to a method of correcting errors caused by temperature changes and pressure changes in order to make the temperature value and pressure value more accurate; the influence factors corresponding to temperature compensation and the influence factors corresponding to pressure compensation are both used to predict and estimate the influence of temperature and pressure changes on the collaborative robot control algorithm through computer simulation, so as to determine the influence factors; in this embodiment, the value range of the influence factor corresponding to temperature compensation is from 0.5 to 0.65; the value range of the influence factor corresponding to pressure compensation is from 0.53 to 0.67.

[0049] Add and integrate the temperature influence index and pressure influence index of the collaborative robot at each control time point to obtain the influence degree coefficient of the collaborative robot control algorithm 。

[0050] Specifically, the control operation accuracy of the collaborative robot, in a specific embodiment, can be recorded and analyzed by using high-precision measurement tools and equipment for key parameters such as the movement trajectory, speed, and position of the collaborative robot, and detailed analysis and statistics are carried out. Compare the actual performance of the collaborative robot with the expected goal to evaluate the value of its control operation accuracy.

[0051] The above control operation accuracy of the collaborative robot, in this embodiment, is obtained through comprehensive analysis of the influence degree coefficient of the collaborative robot's structural characteristics and the influence degree coefficient of the collaborative robot's control algorithm. Therefore, the value of the control operation accuracy of the collaborative robot is determined. In this embodiment, a more accurate calculation method is used to obtain it. The specific analysis process is as follows:

[0052] ;

[0053] Among them represents the control operation accuracy of the collaborative robot. In this embodiment, the structural characteristics and control algorithm will also affect each other. An optimized control algorithm may be able to partially compensate for the performance degradation caused by mechanical structure limitations; similarly, an excellent mechanical structure design may also reduce the requirements for the accuracy of the control algorithm; at the same time, the influence degree coefficient of the collaborative robot's structural characteristics and the influence degree coefficient of the collaborative robot's control algorithm each have a direct impact on the control performance of the collaborative robot. They jointly determine the control operation accuracy of the collaborative robot and need to be considered comprehensively in the design or optimization process;

[0054] represents the influence degree coefficient of the collaborative robot's structural characteristics, which represents the value of the influence degree of the collaborative robot's structural characteristics obtained through comprehensive analysis of the number of joints, its own weight, and the effective load influence index of the collaborative robot;

[0055] It represents the influence degree coefficient of the control algorithm of the collaborative robot, which is a numerical value representing the influence degree of the control algorithm of the collaborative robot obtained through comprehensive analysis of the temperature values and pressure values corresponding to each temperature sensor and each pressure sensor of the collaborative robot at each control time point;

[0056] It represents the weight factor corresponding to the predefined influence degree coefficient of the control algorithm. Among them, the weight factor corresponding to the influence degree coefficient of the structural feature and the weight factor corresponding to the influence degree coefficient of the control algorithm are both obtained by collecting data related to the structural feature and control algorithm of the collaborative robot, including experimental data, simulation data, or historical data, etc., and using the method of factor analysis to process the collected data. Factor analysis is a statistical method used to extract common factors from multiple variables, that is, to extract the data that jointly acts on the influencing factors from multiple related data, so as to calculate the weight factor corresponding to each structural feature. e is the natural constant.

[0057] In this embodiment, the value range of the weight factor corresponding to the influence degree coefficient of the structural feature is from 0.45 to 0.6, and the weight factor corresponding to the influence degree coefficient of the structural feature is set to 0.49; the value range of the weight factor corresponding to the influence degree coefficient of the control algorithm is from 0.5 to 0.6, and the weight factor corresponding to the influence degree coefficient of the control algorithm is set to 0.57.

[0058] In this embodiment, the control operation accuracy of the above-mentioned collaborative robot is shown in Table 1 as follows:

[0059] Table 1 Control operation accuracy of the collaborative robot:

[0060]

[0061] In this embodiment, when the influence degree coefficient of the structural feature and the influence degree coefficient of the control algorithm increase in value, the control operation accuracy of the collaborative robot decreases accordingly; conversely, when the influence degree coefficient of the structural feature and the influence degree coefficient of the control algorithm decrease in value, the control operation accuracy of the collaborative robot increases accordingly. Therefore, if we want to improve the control operation accuracy of the collaborative robot , we need to take effective measures to reduce the negative impacts brought by the structural feature and control algorithm of the collaborative robot.

[0062] Determination of coordination control degree: Through the graphical user interface of the collaborative robot, collect the visual application data of the collaborative robot, and comprehensively consider the control operation accuracy of the collaborative robot to obtain the coordination control degree of the collaborative robot.

[0063] Furthermore, the specific analysis process of the coordination control degree of the collaborative robot is as follows:

[0064] Through the graphical user interface of the collaborative robot, collect the visual application data of the collaborative robot, where the visual application data includes the number of recognized targets.

[0065] The above graphical user interface is an intuitive and easy-to-understand way to interact with the collaborative robot. It uses visual elements such as graphics, icons, buttons, sliders, as well as animations and feedback mechanisms, enabling non-professional users to easily understand and operate the collaborative robot; the displayed content includes but is not limited to the status of the collaborative robot, visual applications, the 3D model of the robot, parameter settings, etc.

[0066] According to the number of recognized targets of the collaborative robot, extract the length and width of the recognized targets of the collaborative robot, where the length and width of each recognized target can be measured by the length sensor carried by the graphical user interface of the collaborative robot.

[0067] According to the length and width of each recognized target of the collaborative robot, obtain the area of each recognized target of the collaborative robot through an area algorithm, thereby constructing a curve of the area change of the recognized targets of the collaborative robot. Extract the maximum area change angle value from the curve of the area change of the recognized targets. At the same time, extract the maximum area change angle definition value from the robot control database, and conduct comprehensive analysis and processing with the maximum area change angle value to obtain the change impact index of the visual recognition of the collaborative robot .

[0068] According to the length and width of each recognized target of the collaborative robot, obtain the area of each recognized target of the collaborative robot through an area algorithm, thereby constructing a curve of the area change of the recognized targets of the collaborative robot, as Figure 3 shown. The abscissa of the curve of the area change of the recognized targets is the recognized target, with the unit being "each"; the ordinate of the curve of the area change of the recognized targets is the area, with the unit being square centimeters.

[0069] According to the curve of the area change of the recognized targets, locate a straight line 1 of area growth between the minimum area and the maximum area of each recognized target of the collaborative robot, and according to Figure 3 the first horizontal straight line 2 in, obtain the minimum angle value of the intersection between the straight line 1 of area growth and the first horizontal straight line 2, which is recorded as the maximum area change angle value of each recognized target of the collaborative robot.

[0070] When a collaborative robot can identify multiple targets simultaneously, it indicates that its recognition function is more accurate and can respond quickly within a short period of time, meaning that the collaborative robot can complete more tasks in a shorter time; when the maximum change angle value of the area is larger, it shows that the collaborative robot can identify more types of target objects, indicating that the collaborative robot may have stronger capabilities in object recognition.

[0071] Comprehensively analyzing the coordination control degree of the collaborative robot, in a specific embodiment, a dynamic model can be constructed to enable the collaborative robot to adjust its own motion trajectory and speed according to the current state and task requirements to achieve the coordinated control of the collaborative robot, and the numerical value of the coordination control degree of the collaborative robot can be determined based on the model data.

[0072] The above-mentioned coordination control degree of the collaborative robot, in this embodiment, is obtained through the comprehensive analysis of the control operation accuracy of the collaborative robot and the change influence index of the collaborative robot's visual recognition, so as to determine the numerical value of the coordination control degree of the collaborative robot. In this embodiment, a more accurate calculation method is used to obtain it. The specific analysis formula is:

[0073] ;

[0074] where represents the coordination control degree of the collaborative robot. In this embodiment, the accuracy of visual recognition directly affects the control operation accuracy of the collaborative robot; if the change in visual recognition is too large, it may cause the collaborative robot to be unable to complete tasks accurately; in order to improve the control operation accuracy of the collaborative robot, it is necessary to continuously optimize and improve its visual recognition system. There is a mutual relationship between the two, so this expression is simplified to better grasp the change trend of the coordinated control of the collaborative robot, enabling the coordination management personnel to take effective measures in a timely manner;

[0075] represents the control operation accuracy of the collaborative robot, which is obtained through the comprehensive analysis of the influence degree coefficient of the structural characteristics of the collaborative robot and the influence degree coefficient of the control algorithm of the collaborative robot, and is the numerical value of the control operation accuracy of the collaborative robot;

[0076] represents the change influence index of the collaborative robot's visual recognition, which refers to the influence degree of the change in the area of each target object recognized by the collaborative robot on the collaborative robot's visual recognition;

[0077] represents the weight factor corresponding to the predefined control operation accuracy, represents the weight factor corresponding to the predefined change influence index of the collaborative robot's visual recognition, where e is the natural constant.

[0078] In this embodiment, the value range of the weight factor corresponding to the control operation accuracy is from 0.52 to 0.6, and the weight factor corresponding to the control operation accuracy is set to 0.6; the value range of the weight factor corresponding to the change influence index of the collaborative robot vision recognition is from 0.61 to 0.69, and the weight factor corresponding to the change influence index of the collaborative robot vision recognition is set to 0.67.

[0079] In this embodiment, the coordination control degree of the above-mentioned collaborative robot is shown in Table 2:

[0080] Table 2 Coordination control degree of the collaborative robot:

[0081]

[0082] In this embodiment, as the change influence index of the collaborative robot vision recognition decreases, the coordination control degree of the collaborative robot will gradually increase. When there are slight differences in the control operation accuracy of the collaborative robot , even if the change influence index of the collaborative robot vision recognition only decreases by 1, it will cause a significant increase in the coordination control degree of the collaborative robot . Therefore, to improve the coordination control degree of the collaborative robot , corresponding measures need to be taken to reduce the change influence index of the collaborative robot vision recognition to the adaptation range.

[0083] Specifically, the set coordination control threshold, the specific analysis process is as follows:

[0084] Obtain the historical coordination data of the collaborative robot, where the historical coordination data of the collaborative robot includes the number of guiding and positioning times and the number of data matrix decoding times. The number of guiding and positioning times and the number of data matrix decoding times can be directly obtained from the historical coordination data report. At the same time, count the number of successful guiding and positioning times and the number of successful data matrix decoding times of the collaborative robot within the set historical coordination period. The number of successful guiding and positioning times and the number of successful data matrix decoding times can be obtained from the coordination control instruction manual. Through data processing, the coordination control threshold is set.

[0085] Furthermore, in a specific embodiment, the coordination control threshold can be obtained by comprehensively analyzing the speed, pressure value, and perception of the collaborative robot according to the evaluation of the task requirements, so as to obtain the specific value of the coordination control threshold.

[0086] The above coordination control threshold is obtained through comprehensive analysis of the number of times the collaborative robot is guided for positioning, the number of times the data matrix is decoded, the number of successful positioning guidance times of the collaborative robot within the set historical coordination period, and the number of successful data matrix decoding times. Therefore, in this embodiment, a more accurate calculation method is adopted to obtain the specific value of the determination coordination control threshold. The specific setting formula is:

[0087] ;

[0088] Where represents the coordination control threshold. In this embodiment, if the collaborative robot frequently performs positioning guidance within the historical coordination period, this may mean that the interaction between the collaborative robot and the operator or the environment is very frequent. Therefore, the coordination control threshold may need to be set to a lower value to ensure that the robot can accurately respond to the operator's instructions or environmental changes in a short time; if the collaborative robot frequently decodes the data matrix within the historical coordination period, this may mean that the task has a high complexity or variability. In this case, the coordination control threshold may need to be set to a higher value to allow the collaborative robot to have more time and flexibility in processing complex tasks;

[0089] represents the number of times the collaborative robot is guided for positioning, which refers to the number of times the collaborative robot is guided by a human or a coordination control algorithm for positioning;

[0090] represents the number of successful positioning guidance times of the collaborative robot within the set historical coordination period, which refers to the number of times the collaborative robot is successfully guided by a human or a coordination control algorithm for positioning within the historical coordination period;

[0091] represents the number of times the collaborative robot decodes the data matrix, which refers to the number of times the collaborative robot decodes or identifies a specific data matrix, such as a QR code, a barcode, etc.;

[0092] represents the number of successful data matrix decoding times of the collaborative robot within the set historical coordination period, which refers to the number of times the collaborative robot successfully decodes or identifies a specific data matrix, such as a QR code, a barcode, etc. within the historical coordination period.

[0093] In this embodiment, the number of times the collaborative robot decodes the data matrix is set to 8 times, the number of successful data matrix decoding times of the collaborative robot within the set historical coordination period is set to 6, and the correction factor corresponding to the data matrix decoding success rate is set to 0.73. Then, the operation result of the part in this expression is 0.55.

[0094] Is expressed as a correction factor corresponding to a predefined guiding positioning success rate, Is expressed as a correction factor corresponding to a predefined data matrix decoding success rate, where the correction factor corresponding to the guiding positioning success rate and the correction factor corresponding to the data matrix decoding success rate are both obtained by collecting data related to the number of guiding positionings and the number of data matrix decodings, which may include system response time, user feedback, task completion time, etc., using regression analysis to analyze the collected data, and determining the correction factor corresponding to the guiding positioning success rate and the correction factor corresponding to the data matrix decoding success rate according to the results of the data analysis; in this embodiment, the value range of the correction factor corresponding to the guiding positioning success rate is from 0.65 to 0.72; the value range of the correction factor corresponding to the data matrix decoding success rate is set to be from 0.71 to 0.77.

[0095] Specifically, the control optimization of the collaborative robot by the deep learning model is as follows:

[0096] Match the coordination control degree of the collaborative robot with the coordination control threshold. If the coordination control degree of the collaborative robot is less than the coordination control threshold, control optimization of the collaborative robot is performed through the deep learning model; if the coordination control degree of the collaborative robot is greater than or equal to the coordination control threshold, there is no need to perform control optimization on the collaborative robot.

[0097] The specific control optimization method is as follows: According to the tasks required by the collaborative robot, it is necessary to clarify the characteristics, requirements, and constraints of the tasks; according to the task requirements, adjust the parameters of the collaborative robot control algorithm to achieve the best control effect; use an efficient path planning algorithm to plan the optimal motion path for the collaborative robot; utilize the multi-sensor fusion technology in the control algorithm to fuse and process data from different sensors to improve the collaborative robot's perception ability and decision-making ability of the environment; introduce machine learning and deep learning technologies to enable the collaborative robot to improve its control strategy and behavior pattern through learning and adaptation.

[0098] Referring to Figure 2 As shown, the second aspect of the present invention provides a system for a collaborative robot control method based on deep learning, including: a control operation accuracy evaluation module, a coordination control degree determination module, and a control optimization module.

[0099] The second aspect of the present invention provides a system for a collaborative robot control method based on deep learning, further including: a robot control database, where the robot control database stores influence factors corresponding to the defined number of joints, the defined weight of itself, the influence index of the payload, and the defined value of the maximum change angle of the area.

[0100] The control operation accuracy evaluation module is connected to the coordination control degree determination module, the coordination control degree determination module is connected to the control optimization module, and both the control operation accuracy evaluation module and the coordination control degree determination module are connected to the robot control database.

[0101] The control operation accuracy evaluation module is used to obtain the control operation correlation information of the collaborative robot and evaluate the control operation accuracy of the collaborative robot.

[0102] The coordination control degree determination module is used to collect the visual application data of the collaborative robot through the graphical user interface of the collaborative robot and obtain the coordination control degree of the collaborative robot by synthesizing the control operation accuracy of the collaborative robot.

[0103] The control optimization module is used to match the coordination control degree of the collaborative robot with a set coordination control threshold. If the coordination control degree of the collaborative robot is less than the coordination control threshold, the collaborative robot is controlled and optimized through a deep learning model.

[0104] The third aspect of the present invention provides a computer-readable storage medium for storing a program, and when the program is executed by a processor, it implements a method for controlling a collaborative robot based on deep learning.

[0105] The above content is only an example and explanation of the structure of the present invention. Those skilled in the art of the present technology can make various modifications or supplements to the described specific embodiments or use similar methods to replace them, as long as they do not deviate from the structure of the invention or exceed the scope defined by the claims of the present invention, they should fall within the protection scope of the present invention.

Claims

1. A collaborative robot control method based on deep learning, characterized in that: include: Control operation accuracy evaluation: obtain the control operation related information of the collaborative robot and evaluate the control operation accuracy of the collaborative robot; Coordination control degree determination: Through the collaborative robot's graphical interactive interface, the collaborative robot's visual application data is collected, and the collaborative robot's control operation accuracy is integrated to obtain the collaborative robot's coordination control degree; Control optimization: Match the coordination control degree of the collaborative robot with the set coordination control threshold. If the coordination control degree of the collaborative robot is less than the coordination control threshold, the control of the collaborative robot is optimized through the deep learning model; The coordination control degree of the collaborative robot is analyzed in detail as follows: Through the collaborative robot's graphical interactive interface, collect the collaborative robot's visual application data, where the visual application data includes the number of recognized targets; According to the number of recognition targets of the collaborative robot, the length and width of each recognition target of the collaborative robot are extracted; According to the length and width of each recognition target of the collaborative robot, the area of ​​each recognition target of the collaborative robot is obtained through the area algorithm, and the recognition target area change curve of the collaborative robot is constructed. The maximum area change angle value is extracted from the recognition target area change curve. At the same time, the maximum area change angle boundary value is extracted from the robot control database, and a comprehensive analysis is performed with the maximum area change angle value to obtain the change impact index of the collaborative robot visual recognition. ; Comprehensively analyze the coordination control degree of the collaborative robot, the specific analysis formula is: ； in Expressed as the coordination control degree of the collaborative robot, Expressed as the control operation accuracy of the collaborative robot, It is expressed as the impact index of changes in collaborative robot visual recognition, It is expressed as the weight factor corresponding to the predefined control operation accuracy, It is represented as the weight factor corresponding to the predefined change impact index of collaborative robot visual recognition, and e is a natural constant; The control operation accuracy of the collaborative robot is analyzed in the following specific process: According to the control operation related information of the collaborative robot, the structural feature information of the collaborative robot and the control algorithm information of the collaborative robot are obtained; The influence coefficients of the structural characteristics of the collaborative robot and the control algorithm of the collaborative robot are comprehensively analyzed to obtain the control operation accuracy of the collaborative robot.

2. The collaborative robot control method based on deep learning according to claim 1, characterized in that: The control optimization of the collaborative robot is performed by using a deep learning model, and the specific optimization process is as follows: The coordination control degree of the collaborative robot is matched with the coordination control threshold. If the coordination control degree of the collaborative robot is less than the coordination control threshold, the control of the collaborative robot is optimized through the deep learning model; if the coordination control degree of the collaborative robot is greater than or equal to the coordination control threshold, there is no need to optimize the control of the collaborative robot.

3. The collaborative robot control method based on deep learning according to claim 1, characterized in that: The specific analysis process of the set coordination control threshold is as follows: The historical coordination data of the collaborative robot is obtained, where the historical coordination data of the collaborative robot includes the number of guided positioning times and the number of data matrix decoding times. At the same time, the number of successful guided positioning times and the number of successful data matrix decoding times of the collaborative robot within the set historical coordination period are counted, and the coordination control threshold is set through data processing.

4. The collaborative robot control method based on deep learning according to claim 2, characterized in that: The coordination control threshold, the specific data processing process is: The number of guided positioning times of the collaborative robot is processed by ratio processing with the number of successful guided positioning times of the collaborative robot in a set historical coordination cycle to obtain the guided positioning success rate of the collaborative robot, and the ratio is multiplied by the correction factor corresponding to the predefined guided positioning success rate to obtain the guiding positioning success impact index of the collaborative robot; The number of data matrix decoding times of the collaborative robot is processed by ratio with the number of successful data matrix decoding times of the collaborative robot within a set historical coordination period to obtain the data matrix decoding success rate of the collaborative robot, and the ratio is multiplied by the correction factor corresponding to the predefined data matrix decoding success rate to obtain the data matrix successful decoding influence index of the collaborative robot; The coordination control threshold is obtained by adding the collaborative robot's guidance positioning success impact index and the collaborative robot's data matrix successful decoding impact index.

5. The collaborative robot control method based on deep learning according to claim 1, characterized in that: The structural characteristics of the collaborative robot influence the degree coefficient, and the specific analysis process is as follows: From the information of the structural characteristics of the collaborative robot, the number of joints of the collaborative robot, its own weight, the average load value in the control cycle, and the weight of the end effector are extracted; The effective load value of the collaborative robot in the control period is obtained by subtracting the weight of the end effector from the load mean value of the collaborative robot in the control period, and the effective load value of the collaborative robot in the control period is comprehensively analyzed and processed with the set effective load threshold to obtain the effective load influence index of the collaborative robot; The influencing factors corresponding to the joint definition number, self-defined weight and effective load influencing indicators are extracted from the robot control database, and the influence coefficient of the structural characteristics of the collaborative robot is comprehensively analyzed.

6. The collaborative robot control method based on deep learning according to claim 1, characterized in that: The control algorithm influence coefficient of the collaborative robot is analyzed in detail as follows: The number of temperature sensors and pressure sensors carried by the collaborative robot are extracted from the information of the collaborative robot's control algorithm; The control cycle is divided into control time points, and the temperature value corresponding to each temperature sensor of the collaborative robot at each control time point and the pressure value corresponding to each pressure sensor at each control time point are extracted, and compared with the preset temperature threshold and pressure threshold respectively, and the average temperature error value and the average pressure error value of the collaborative robot at each control time point are obtained by integration; The average temperature error value of the collaborative robot at each control time point is multiplied by the influence factor corresponding to the set temperature compensation to obtain the temperature influence index of the collaborative robot at each control time point; the average pressure error value of the collaborative robot at each control time point is multiplied by the influence factor corresponding to the set pressure compensation to obtain the pressure influence index of the collaborative robot at each control time point; The temperature influence index and pressure influence index of the collaborative robot at each control time point are added and integrated to obtain the influence degree coefficient of the control algorithm of the collaborative robot.

7. A system using the collaborative robot control method based on deep learning as described in any one of claims 1 to 6, characterized in that: A control operation accuracy evaluation module is used to obtain the control operation related information of the collaborative robot and evaluate the control operation accuracy of the collaborative robot; The coordination control degree determination module collects the visual application data of the collaborative robot through the collaborative robot's graphical interactive interface, and integrates the control operation accuracy of the collaborative robot to obtain the coordination control degree of the collaborative robot; The control optimization module matches the coordination control degree of the collaborative robot with the set coordination control threshold. If the coordination control degree of the collaborative robot is less than the coordination control threshold, the control of the collaborative robot is optimized through the deep learning model.

8. A computer-readable storage medium for storing a program, which, when executed by a processor, implements the deep learning-based collaborative robot control method as described in any one of claims 1 to 6.

Images