Robot motion control method and device based on image recognition processing

Through the method based on image recognition processing, the target to be captured by the robot arm is scanned and recognized and position image are analyzed and obtained, which solves the problem of the automatic learning effect of the robot arm motion control and the optimization of the control data expansion in the existing technology. The best problem achieves adaptive grasping and stability testing for different goals, and improves the autonomous learning and self-optimization capabilities of the robotic arm.

CN118544358BActive Publication Date: 2024-12-03INEXBOT
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410993058.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-07-24
Publication Date
2024-12-03
Estimated Expiration
2044-07-24

AI Technical Summary

Technical Problem

When facing transport targets of different specifications and weights, existing robotic arm motion control solutions cannot independently perform data analysis and stability tests in the grab state, resulting in poor optimization of independent learning effects and control data expansion.

Method used

Using an image recognition processing method, the targets to be captured are scanned and recognized through the Internet of Things technology to obtain the capture feature sequence, analyze and obtain the capture implementation parameters, use the camera device to recognize the position image, control the robotic arm to capture and stabilize it Sex test, dynamically manage crawling parameters.

Benefits of technology

The autonomous learning effect and control data expansion optimization of the robot arm are improved, and the stability testing of adaptive grasping scheme selection and grab state for different goals is achieved, which enhances the autonomous testing ability and self-optimization ability of the robot arm.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118544358B_ABST
    Figure CN118544358B_ABST
Patent Text Reader

Abstract

The invention discloses a robot motion control method and device based on image recognition processing, belonging to the technical field of data recognition processing; the invention is used to solve the technical problems of poor autonomous learning effect of robot arm motion control and poor effect of control data expansion optimization in existing solutions; the basic information of a target to be grasped by the robot arm is scanned and identified, and a grasping feature sequence corresponding to the target is obtained through processing, data analysis is performed on the grasping feature sequence to obtain a corresponding grasping scheme and grasping implementation parameters, position image recognition is performed on the target to be grasped, and basic position features corresponding to the target are obtained through analysis, the robot arm is controlled to perform stability test on the target to be grasped by using the basic grasping implementation parameters obtained through analysis, subsequent grasping of the grasped target is dynamically controlled according to the test results, and the basic grasping implementation parameters belonging to the grasped target are dynamically managed.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of data recognition and processing, and in particular to a robot motion control method and device based on image recognition and processing. Background Art

[0002] Robot (usually in the form of a robotic arm) motion control is the process of precisely adjusting the angle or position of each joint so that it can perform a specific task along a predetermined trajectory and speed.

[0003] The existing robotic arm motion control scheme has certain defects in implementation. When faced with transfer targets of different specifications and weights, it is impossible to independently perform data analysis to implement a variety of grasping and transfer schemes. At the same time, it is impossible to actively test and control the stability of the grasping state of targets with incomplete control parameters, resulting in poor autonomous learning effects of robotic arm motion control and poor effects of control data expansion and optimization. Summary of the invention

[0004] The purpose of the present invention is to provide a robot motion control method and device based on image recognition processing, which is used to solve the technical problems of poor autonomous learning effect of robot arm motion control and poor effect of control data expansion and optimization in existing solutions.

[0005] The purpose of the present invention can be achieved through the following technical solutions:

[0006] A robot motion control method based on image recognition processing, comprising:

[0007] The IoT technology is used to scan and identify the basic information of the target to be grasped by the robot arm and process and obtain the grasping feature sequence corresponding to the target;

[0008] Perform data analysis on the crawling feature sequence to obtain conventional crawling implementation parameters or basic crawling implementation parameters corresponding to different crawling schemes;

[0009] The camera device installed on the robot arm performs position image recognition on the target to be grasped and analyzes and obtains the basic position features corresponding to the target;

[0010] The basic grasping implementation parameters obtained through analysis are used to control the robot arm to grasp the target to be grasped, the grasping state stability test is performed on the grasped target, and the acquired test images are processed and recognized and tested and analyzed in combination with basic position features;

[0011] Based on the test results, the subsequent crawling of the crawling target is dynamically controlled, and the basic crawling implementation parameters of the crawling target are dynamically managed.

[0012] Preferably, the basic information includes the number, length, width, height, weight, protection type and protection identification of the target;

[0013] Extracting the values ​​of length, weight and protection mark and marking them as the first grasping feature, the second grasping feature and the third grasping feature respectively;

[0014] Sort and combine several marked grasping features to obtain a grasping feature sequence.

[0015] Preferably, different elements in the captured feature sequence are sequentially obtained and marked as ai, i=1, 2, 3; the marked elements are sequentially input into the feature recognition model for data analysis and output feature recognition values; wherein the feature recognition values ​​include values ​​of 0 or 1;

[0016] The expression of the feature recognition model is: ; In the formula, Ui is the standard grasping feature range corresponding to different grasping features;

[0017] The acquired feature recognition values ​​are arranged and combined according to the output order to obtain a feature recognition sequence.

[0018] Preferably, when determining the corresponding grasping implementation parameters according to the feature recognition sequence, all elements with a value of 0 in the feature recognition sequence and their corresponding element positions are obtained;

[0019] If the values ​​of all elements in the feature recognition sequence are 0, a conventional instruction is generated, and according to the conventional instruction, a conventional clamping distance associated with the standard grasping feature range to which the first element belongs is obtained, and a conventional clamping force associated with the standard grasping feature range to which the third element belongs is obtained;

[0020] The numerical values ​​of the conventional clamping distance and the conventional clamping force are sorted and combined to obtain conventional grasping implementation parameters;

[0021] If there is a second element in the feature recognition sequence whose value is 0, and the total number of elements whose value is 0 is less than three, a verification instruction is generated, and according to the verification instruction, the value of the first element is set as the basic clamping distance, and the conventional clamping force associated with the standard grasping feature range to which the second element belongs is obtained and set as the basic clamping force;

[0022] The basic gripping implementation parameters are obtained by combining the numerical values ​​of the basic clamping distance and the basic clamping force.

[0023] Preferably, the position coordinates of the target to be grasped relative to the position of the robotic arm are determined, a positioning instruction is generated when the clamping mechanism of the robotic arm contacts both sides of the target, and a front image of the clamping mechanism of the robotic arm contacting the target is obtained according to the positioning instruction;

[0024] Perform image processing and feature recognition on the front image, obtain the support objects supporting the target in the front image and construct a support coordinate system;

[0025] According to the support coordinate system, two first contact points between the clamping mechanism of the robot arm and the target and the corresponding first contact point coordinates are obtained, and a first test midpoint of contact between the target and the support and the corresponding first test midpoint coordinates are obtained.

[0026] Preferably, the straight-line distance between the two first contact points and the first test midpoint is calculated based on the coordinates of the two first contact points and the coordinates of the first test midpoint and marked as the first verification value Y(h), and the vertical distance between the two first contact points and the first test midpoint is calculated and marked as the second verification value E(h); the first verification value and the second verification value constitute the basic position feature.

[0027] Preferably, the basic grasping implementation parameters are traversed, and the clamping mechanism of the mechanical arm is controlled to clamp the target from both sides according to the basic clamping distance and basic clamping force obtained through the traversal, and the mechanical arm is controlled to lift the clamped target by using a preset lifting height, so that the mechanical arm is controlled to suspend operation after the target is in a suspended state;

[0028] The target in the suspended state is photographed to obtain a corresponding test image, and the test image is processed and feature recognized to obtain the second contact point and the second contact point coordinates corresponding to the two contact points where the clamping mechanism of the lifted robot arm contacts the target, and the second test midpoint at the lower end of the front surface of the lifted target and the corresponding second test midpoint coordinates.

[0029] Preferably, the straight-line distances between the two second contact points and the second test midpoint are calculated according to the coordinates of the two second contact points and the coordinates of the second test midpoint and are marked as the first test value Y(c), and the vertical distance between the two second contact points and the second test midpoint is calculated and marked as the second test value E(c);

[0030] When the first test value and the second test value are tested and analyzed, the implementation stability μ corresponding to the clamping target of the robot arm is obtained by calculation through the formula; the calculation formula of the implementation stability μ is: ; In the formula, w1 and w2 are both proportional coefficients greater than zero, and 2×w1=w2; H is the preset lifting height;

[0031] If μ≠w1+w2, a control anomaly label is generated, and the implementation stability calculated based on the control anomaly label is input into the control anomaly identification model for data analysis and output of the corresponding implementation stability mark; wherein the output implementation stability mark contains a value of 1 or 2; the expression of the control anomaly identification model is: .

[0032] Preferably, according to the implementation stability flag with a value of 1, the robot arm is controlled to continue to clamp the target with a preset enhanced force and work according to a preset control process, and according to the implementation stability flag with a value of 1, the conventional clamping force associated with the corresponding target is replaced and updated according to the preset enhanced force;

[0033] And, according to the implementation stability flag with a value of 2, the robotic arm is controlled to lower the clamped target to the original position, and the control parameters of the basic clamping distance and the basic clamping force are adjusted respectively, and the adjusted control parameters are used to control the robotic arm to repeatedly grasp the target and perform a stability test on the grasping state of the grasped target, until the implementation stability flag of the test analysis is 0 or 1, and the adjusted effective control parameters are replaced and updated.

[0034] The present invention also discloses a robot motion control device based on image recognition processing, comprising:

[0035] The target information acquisition and processing module is used to scan and identify the basic information of the target to be grasped by the robot arm through the Internet of Things technology and process and obtain the grasping feature sequence corresponding to the target; perform data analysis on the grasping feature sequence to obtain the conventional grasping implementation parameters or basic grasping implementation parameters corresponding to different grasping schemes; perform position image recognition of the target to be grasped through the camera device installed on the robot arm and analyze and obtain the basic position features corresponding to the target;

[0036] The target grasping test analysis module is used to control the robot arm to grasp the target to be grasped by using the basic grasping implementation parameters obtained through analysis, to test the stability of the grasping state of the grasped target, and to perform image processing and recognition on the acquired test image and perform test analysis in combination with basic position features;

[0037] The target grab test management module is used to dynamically control the subsequent grabbing of the grab target according to the test results, and to dynamically manage the basic grab implementation parameters of the grab target.

[0038] Compared with the existing solutions, the present invention achieves the following beneficial effects:

[0039] The present invention scans and identifies basic information of the target to be grasped by the robot arm and processes it to obtain a grasping feature sequence corresponding to the target. According to the grasping feature sequence, data analysis is performed on grasping features of different aspects of the target to be grasped to obtain corresponding feature identification values. Different grasping features can be digitally processed while providing data support of different dimensions for the implementation of subsequent corresponding target grasping plans, thereby improving the diversity of grasping feature processing and utilization.

[0040] The present invention acquires corresponding grasping schemes and grasping implementation parameters by performing data analysis on grasping feature sequences, thereby realizing adaptive dynamic selection of grasping schemes for different targets and improving the adaptability and flexibility of mechanical arm recognition and grasping.

[0041] The present invention performs position image recognition on the target to be grasped and analyzes the basic position features corresponding to the target, controls the robot arm to grasp the target by utilizing the basic grasping implementation parameters obtained by the analysis, and performs a stability test on the grasping state of the grasped target. It can not only obtain whether the grasping state of the target controlled by the basic grasping implementation parameters is normal, but also automatically share and update the subsequent grasping parameters of the target according to the stability test result, thereby improving the autonomous testing capability and autonomous improvement capability of the robot arm in grasping different targets.

[0042] The present invention dynamically controls the subsequent grasping of the grasping target according to the test results, and dynamically manages the basic grasping implementation parameters to which the grasping target belongs, thereby enabling the robot arm to autonomously implement grasping test adjustments and self-optimize and update grasping test data for targets with incomplete control parameters, thereby improving the autonomous learning effect of the robot arm's motion control and the overall effect of control data expansion and optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The present invention will be further described below in conjunction with the accompanying drawings.

[0044] Figure 1 The present invention is a flowchart of a robot motion control method based on image recognition processing.

[0045] Figure 2 This is a module block diagram of a robot motion control device based on image recognition processing of the present invention. DETAILED DESCRIPTION

[0046] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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 ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0047] Example 1: Figure 1 As shown, the present invention is a robot motion control method based on image recognition processing, comprising:

[0048] The IoT technology is used to scan and identify the basic information of the target to be grasped by the robot arm and process the grasping feature sequence corresponding to the target; including:

[0049] The Internet of Things technology can be radio frequency identification technology, which uses radio frequency tags installed on the outer surface of the target to identify and obtain information. The basic information includes the target's number, length, width, height, weight, protection type and protection mark; the target can be logistics goods of different shapes to be transported;

[0050] Among them, the protection type includes hard protection type and soft protection type;

[0051] The specific hard protection type can be a carton-type protective outer packaging, and the specific soft protection type can be a plastic-type protective outer packaging; it is used to provide external protection data support for the subsequent dynamic adjustment of the robot arm to the target clamping of different protection types;

[0052] The values ​​of length, weight and protection mark are extracted and marked as the first grab feature, the second grab feature and the third grab feature respectively; the protection mark is used to digitally represent and identify different protection types. The specific value is not limited and can be customized according to the actual situation;

[0053] Sort and combine several marked grasping features to obtain a grasping feature sequence;

[0054] In the embodiment of the present invention, by scanning and identifying basic information of the target to be grasped by the robot arm and processing and obtaining the grasping feature sequence corresponding to the target, reliable target digital data support can be provided for the subsequent automatic recognition and analysis of different grasping schemes implemented by the robot arm.

[0055] Perform data analysis on the crawling feature sequence to obtain conventional crawling implementation parameters or basic crawling implementation parameters corresponding to different crawling schemes; including:

[0056] Sequentially obtain different elements in the captured feature sequence and mark them as ai, i=1, 2, 3;

[0057] The elements obtained by marking are sequentially input into the feature recognition model for data analysis and output of feature recognition values; the feature recognition model is constructed based on the isolation forest algorithm combined with all standard capture feature ranges; wherein the feature recognition value includes a value of 0 or 1, which respectively indicates that the corresponding element exists in history or does not exist in history;

[0058] The expression of the feature recognition model is: ; In the formula, Ui is the standard crawling feature range corresponding to different crawling features, which is determined based on the historical crawling big data corresponding to the same target type;

[0059] Arrange and combine the acquired feature recognition values ​​according to the output order to obtain a feature recognition sequence;

[0060] In the embodiment of the present invention, by performing data analysis on the grasping features of different aspects of the grasping target to obtain the corresponding feature recognition value, different grasping features can be digitized and data support of different dimensions can be provided for the implementation of subsequent corresponding target grasping solutions, thereby improving the diversity of grasping feature processing and utilization.

[0061] When determining the corresponding grasping implementation parameters according to the feature recognition sequence, all elements with a value of 0 in the feature recognition sequence and their corresponding element positions are obtained;

[0062] If the values ​​of all elements in the feature recognition sequence are 0, a conventional instruction is generated, and according to the conventional instruction, a conventional clamping distance associated with the standard grasping feature range to which the first element belongs is obtained, and a conventional clamping force associated with the standard grasping feature range to which the third element belongs is obtained;

[0063] It should be explained that the conventional instruction indicates that the robot arm has grasped the target with the same parameters in the historical grasping behavior, so the grasping parameters that have been grasped and implemented in the past can be used to control the grasping of the target to be grasped; in addition, different standard grasping feature ranges are pre-associated with a corresponding conventional clamping distance or conventional clamping force, which is determined based on the grasping big data of different targets in the history;

[0064] The numerical values ​​of the conventional clamping distance and the conventional clamping force are sorted and combined to obtain conventional grasping implementation parameters;

[0065] If there is a second element in the feature recognition sequence whose value is 0, and the total number of elements whose value is 0 is less than three, a verification instruction is generated, and according to the verification instruction, the value of the first element is set as the basic clamping distance, and the conventional clamping force associated with the standard grasping feature range to which the second element belongs is obtained and set as the basic clamping force;

[0066] It is worth noting that the standard grasping feature range to which the second element belongs is also associated with a conventional clamping force. When the first element and the second element are valid, the conventional clamping force associated with the standard grasping feature range to which the third element belongs has a higher priority than the conventional clamping force associated with the standard grasping feature range to which the second element belongs; otherwise, the conventional clamping force associated with the standard grasping feature range to which the second element belongs shall prevail.

[0067] The basic gripping implementation parameters are obtained by combining the numerical values ​​of the basic gripping distance and the basic gripping force;

[0068] In the embodiment of the present invention, by performing data analysis on the grasping feature sequence to obtain the corresponding grasping scheme and grasping implementation parameters, dynamic selection of grasping schemes adaptive to different targets is achieved, thereby improving the adaptability and flexibility of robot arm recognition and grasping.

[0069] The camera device installed on the robot arm is used to identify the position image of the target to be grasped and analyze the basic position features corresponding to the target; including:

[0070] Determine the position coordinates of the target to be grasped relative to the position of the robotic arm. The position coordinates of the target are determined based on the coordinate system constructed by the existing robotic arm and the existing positioning technology. When the clamping mechanism of the robotic arm contacts both sides of the target, a positioning instruction is generated. According to the positioning instruction, a front image of the clamping mechanism of the robotic arm contacting the target is obtained. The clamping mechanism is used to clamp the target from both sides.

[0071] Perform image processing and feature recognition on the front image, obtain the support that supports the target in the front image and construct a support coordinate system; the support coordinate system can set the midpoint of the front surface of the support as the origin, and establish the support coordinate system according to the preset coordinate spacing and coordinate direction, and the preset coordinate spacing and coordinate direction are determined according to the actual application scenario;

[0072] It should be noted that image processing and feature recognition of the front image are both existing conventional technical means, and the specific steps are not repeated here; the support may be a conveyor belt for intermittent transmission, or other supporting structures;

[0073] According to the support coordinate system, two first contact points where the clamping mechanism of the robot arm contacts the target and the corresponding first contact point coordinates are obtained, and the first test midpoint of contact between the target and the support and the corresponding first test midpoint coordinates are obtained; the contact point is the midpoint of the contact surface where the clamping mechanism of the robot arm contacts the two side surfaces of the target;

[0074] According to the coordinates of the two first contact points and the coordinates of the first test midpoint, the straight-line distances between the two first contact points and the first test midpoint are calculated respectively and marked as the first verification value Y(h), and the vertical distances between the two first contact points and the first test midpoint coordinates are calculated and marked as the second verification value E(h);

[0075] The first verification value and the second verification value constitute the basic position feature;

[0076] In an embodiment of the present invention, by performing position image recognition on the target to be grasped and analyzing the basic position features corresponding to the target, it is possible to digitally represent the relative state of the target before grasping, and provide reliable data support for the subsequent grasping state analysis of the target after grasping, thereby improving the diversity of feature data processing in different dimensions when the target is stationary.

[0077] The basic grasping implementation parameters obtained by analysis are used to control the robot arm to grasp the target to be grasped, the grasping state stability test is performed on the grasped target, and the acquired test image is processed and recognized and tested and analyzed in combination with the basic position features; including:

[0078] Traverse the basic grasping implementation parameters, control the clamping mechanism of the robot arm to clamp the target from both sides according to the basic clamping distance and basic clamping force obtained through the traversal, and use the preset lifting height to control the robot arm to lift the clamped target, so that the robot arm is controlled to suspend operation after the target is in a suspended state; wherein the specific value of the preset lifting height can be determined according to the height of the target, which can be 1.5 times the height of the target;

[0079] The target in the suspended state is photographed to obtain a corresponding test image, where the photographing direction is the same as that of the previous front image, and the test image is processed and feature recognized, where the technical scheme for processing and feature recognition is the same as that for the corresponding processing of the front image, and the second contact point and the coordinates of the second contact point corresponding to the two contact points where the clamping mechanism of the lifted robot arm contacts the target are obtained, and the second test midpoint at the lower end of the front surface of the lifted target and the corresponding coordinates of the second test midpoint are obtained;

[0080] According to the coordinates of the two second contact points and the coordinates of the second test midpoint, the straight-line distances between the two second contact points and the second test midpoint are calculated and marked as the first test value Y(c), and the vertical distance between the two second contact points and the second test midpoint is calculated and marked as the second test value E(c);

[0081] When the first test value and the second test value are tested and analyzed, the implementation stability μ corresponding to the clamping target of the robot arm is obtained by calculation through the formula; the calculation formula of the implementation stability μ is: ; In the formula, w1 and w2 are both proportional coefficients greater than zero, and 2×w1=w2; H is the preset lifting height;

[0082] If μ = w1 + w2, then generate a control fully valid label and set its associated implementation stable flag to 0;

[0083] If μ≠w1+w2, a control anomaly label is generated, and the implementation stability calculated based on the control anomaly label is input into the control anomaly recognition model for data analysis and output of the corresponding implementation stability mark; the control anomaly recognition model is constructed based on the isolation forest algorithm combined with the standard capture design requirement data of the target;

[0084] The output implementation stability flag includes a value of 1 or 2, which respectively indicates that the corresponding test result is a slight abnormality in grasping or a severe abnormality in grasping; specifically, there is a slight decline or a severe decline in the corresponding target after being grasped;

[0085] The expression of the control anomaly recognition model is: .

[0086] In the embodiment of the present invention, by utilizing the basic grasping implementation parameters obtained through analysis to control the robotic arm to grasp the target to be grasped, and performing a stability test on the grasping state of the grasped target, it is possible to obtain whether the grasping state of the target controlled by the basic grasping implementation parameters is normal, and to automatically share and update the subsequent grasping parameters of the target according to the stability test results, thereby improving the autonomous testing capability and autonomous improvement capability of the robotic arm in grasping different targets.

[0087] Dynamically control the subsequent crawling of the crawling target based on the test results, and dynamically manage the basic crawling implementation parameters of the crawling target; including:

[0088] According to the implementation stability mark with a value of 0, the robot arm is controlled to continue to clamp the target and work according to the preset control process;

[0089] According to the implementation stability flag with a value of 1, the robot arm is controlled to continue to clamp the target with a preset enhanced force and work according to a preset control process, and according to the implementation stability flag with a value of 1, the conventional clamping force associated with the corresponding target is replaced and updated according to the preset enhanced force; the preset enhanced force is determined according to the weight of the target, for example, according to the median or maximum value of all conventional clamping forces of other types of targets of the same weight;

[0090] And, according to the implementation stability flag with a value of 2, the robotic arm is controlled to lower the clamped target to the original position, and the control parameters of the basic clamping distance and the basic clamping force are adjusted respectively through the preset enhanced force and the preset clamping adjustment distance. The clamping adjustment distance is determined according to the median or minimum value of the conventional clamping distances corresponding to all targets of the same length and the same weight, and the adjusted clamping distance and the adjusted clamping force are obtained. The adjusted clamping distance and the adjusted clamping force are used to control the robotic arm to repeat the above-mentioned grasping of the target and to perform a stability test on the grasping state of the grasped target until the implementation stability flag of the test analysis is 0 or 1, and the last implemented adjusted clamping distance and adjusted clamping force are used to replace and update the data of the basic clamping distance and basic clamping force belonging to the clamped target.

[0091] In the embodiment of the present invention, the subsequent grasping of the grasping target is dynamically controlled according to the test results, and the basic grasping implementation parameters of the grasping target are dynamically managed, so that the robot arm can autonomously implement grasping test adjustments and self-optimize and update grasping test data for targets with incomplete control parameters. This can avoid manual intervention in passively setting parameters for testing and adjustment, and improve the autonomous learning effect of the robot arm motion control and the overall effect of control data expansion and optimization.

[0092] Example 2: Figure 2 As shown, the present invention is a robot motion control device based on image recognition processing, including a target information acquisition and processing module, a target grasping test analysis module and a target grasping test management module;

[0093] The target information acquisition and processing module is used to scan and identify the basic information of the target to be grasped by the robot arm through the Internet of Things technology and process and obtain the grasping feature sequence corresponding to the target; perform data analysis on the grasping feature sequence to obtain the conventional grasping implementation parameters or basic grasping implementation parameters corresponding to different grasping schemes; perform position image recognition of the target to be grasped through the camera device installed on the robot arm and analyze and obtain the basic position features corresponding to the target;

[0094] The target grasping test analysis module is used to control the robot arm to grasp the target to be grasped by using the basic grasping implementation parameters obtained through analysis, to test the stability of the grasping state of the grasped target, and to perform image processing and recognition on the acquired test image and perform test analysis in combination with basic position features;

[0095] The target grab test management module is used to dynamically control the subsequent grabbing of the grab target according to the test results, and to dynamically manage the basic grab implementation parameters of the grab target.

[0096] In addition, the formulas involved in the above are all dimensionless and numerical calculations. They are a formula that is closest to the actual situation obtained by collecting a large amount of data and simulating it with simulation software.

[0097] In the several embodiments provided by the present invention, it should be understood that the disclosed device can be implemented in other ways. For example, the above-described embodiments of the invention are only illustrative, for example, the division of modules is only a logical function division, and there may be other division methods in actual implementation.

[0098] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, and may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0099] In addition, each functional module in each embodiment of the present invention may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0100] It is obvious to a person skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the essential characteristics of the present invention.

[0101] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A robot motion control method based on image recognition processing, It is characterized in that include: The IoT technology is used to scan and identify the basic information of the target to be grasped by the robot arm and process and obtain the grasping feature sequence corresponding to the target; Basic information includes the target number, length, width, height, weight, protection type and protection mark; Extracting the values ​​of length, weight and protection mark and marking them as the first grasping feature, the second grasping feature and the third grasping feature respectively; Sort and combine several marked grasping features to obtain a grasping feature sequence; Perform data analysis on the crawling feature sequence to obtain conventional crawling implementation parameters or basic crawling implementation parameters corresponding to different crawling schemes; Wherein, different elements in the captured feature sequence are obtained in sequence and marked as ai, i=1, 2, 3; the marked elements are input into the feature recognition model in sequence for data analysis and the feature recognition value is output; wherein the feature recognition value includes a value of 0 or 1; The expression of the feature recognition model is: ; In the formula, Ui is the standard grasping feature range corresponding to different grasping features; Arrange and combine the acquired feature recognition values ​​according to the output order to obtain a feature recognition sequence; When determining the corresponding grasping implementation parameters according to the feature recognition sequence, all elements with a value of 0 in the feature recognition sequence and their corresponding element positions are obtained; If the values ​​of all elements in the feature recognition sequence are 0, a conventional instruction is generated, and according to the conventional instruction, a conventional clamping distance associated with the standard grasping feature range to which the first element belongs is obtained, and a conventional clamping force associated with the standard grasping feature range to which the third element belongs is obtained; The numerical values ​​of the conventional clamping distance and the conventional clamping force are sorted and combined to obtain conventional grasping implementation parameters; If there is a second element in the feature recognition sequence whose value is 0, and the total number of elements whose value is 0 is less than three, a verification instruction is generated, and according to the verification instruction, the value of the first element is set as the basic clamping distance, and the conventional clamping force associated with the standard grasping feature range to which the second element belongs is obtained and set as the basic clamping force; The basic gripping implementation parameters are obtained by combining the numerical values ​​of the basic gripping distance and the basic gripping force; The camera device installed on the robot arm performs position image recognition on the target to be grasped and analyzes and obtains the basic position features corresponding to the target; The position coordinates of the target to be grasped relative to the position of the robot arm are determined, and a positioning instruction is generated when the clamping mechanism of the robot arm contacts both sides of the target, and a front image of the contact between the clamping mechanism of the robot arm and the target is obtained according to the positioning instruction; Perform image processing and feature recognition on the front image, obtain the support objects supporting the target in the front image and construct a support coordinate system; Acquire two first contact points between the clamping mechanism of the manipulator and the target and the corresponding first contact point coordinates according to the support coordinate system, and acquire a first test midpoint of contact between the target and the support and the corresponding first test midpoint coordinates; According to the coordinates of the two first contact points and the coordinates of the first test midpoint, the straight-line distance between the two first contact points and the first test midpoint is calculated and marked as the first verification value Y(h), and the vertical distance between the two first contact points and the first test midpoint coordinates is calculated and marked as the second verification value E(h); the first verification value and the second verification value constitute the basic position feature; The basic grasping implementation parameters obtained through analysis are used to control the robot arm to grasp the target to be grasped, the grasping state stability test is performed on the grasped target, and the acquired test images are processed and recognized and tested and analyzed in combination with basic position features; Among them, the basic grasping implementation parameters are traversed, and the clamping mechanism of the robot arm is controlled to clamp the target from both sides according to the basic clamping distance and basic clamping force obtained through the traversal, and the robot arm is controlled to lift the clamped target using the preset lifting height, so that the robot arm is controlled to suspend operation after the target is in a suspended state; The target in the suspended state is photographed to obtain a corresponding test image, the test image is processed and feature recognized, a second contact point and the coordinates of the second contact point corresponding to two contact points where the clamping mechanism of the lifted robot arm contacts the target are obtained, and a second test midpoint at the lower end of the front surface of the lifted target and the corresponding coordinates of the second test midpoint are obtained; According to the coordinates of the two second contact points and the coordinates of the second test midpoint, the straight-line distances between the two second contact points and the second test midpoint are calculated and marked as the first test value Y(c), and the vertical distance between the two second contact points and the second test midpoint is calculated and marked as the second test value E(c); When the first test value and the second test value are tested and analyzed, the implementation stability μ corresponding to the clamping target of the robot arm is obtained by calculation through the formula; the calculation formula of the implementation stability μ is: ; In the formula, w1 and w2 are both proportional coefficients greater than zero, and 2×w1=w2; H is the preset lifting height; If μ≠w1+w2, a control anomaly label is generated, and the implementation stability calculated based on the control anomaly label is input into the control anomaly identification model for data analysis and output of the corresponding implementation stability mark; wherein the output implementation stability mark contains a value of 1 or 2; the expression of the control anomaly identification model is: ; Based on the test results, the subsequent crawling of the crawling target is dynamically controlled, and the basic crawling implementation parameters of the crawling target are dynamically managed.

2. A robot motion control method based on image recognition processing according to claim 1, It is characterized in that According to the implementation stability flag with a value of 1, the robot arm is controlled to continue to clamp the target with a preset enhanced force and work according to a preset control process, and according to the implementation stability flag with a value of 1, the conventional clamping force associated with the corresponding target is replaced and updated according to the preset enhanced force; And, according to the implementation stability flag with a value of 2, the robotic arm is controlled to lower the clamped target to the original position, and the control parameters of the basic clamping distance and the basic clamping force are adjusted respectively, and the adjusted control parameters are used to control the robotic arm to repeatedly grasp the target and perform a stability test on the grasping state of the grasped target, until the implementation stability flag of the test analysis is 0 or 1, and the adjusted effective control parameters are replaced and updated.

3. A robot motion control device based on image recognition processing, using a robot motion control method based on image recognition processing according to any one of claims 1-2, It is characterized in that include: The target information acquisition and processing module is used to scan and identify the basic information of the target to be grasped by the robot arm through the Internet of Things technology and process and obtain the grasping feature sequence corresponding to the target; perform data analysis on the grasping feature sequence to obtain the conventional grasping implementation parameters or basic grasping implementation parameters corresponding to different grasping schemes; perform position image recognition of the target to be grasped through the camera device installed on the robot arm and analyze and obtain the basic position features corresponding to the target; The target grasping test analysis module is used to control the robot arm to grasp the target to be grasped by using the basic grasping implementation parameters obtained through analysis, to test the stability of the grasping state of the grasped target, and to perform image processing and recognition on the acquired test image and perform test analysis in combination with basic position features; The target grab test management module is used to dynamically control the subsequent grabbing of the grab target according to the test results, and to dynamically manage the basic grab implementation parameters of the grab target.