Automated quality inspection system based on machine learning

Abstract

The present invention relates to the technical field of robot motion quality detection, and discloses an automated quality detection system based on machine learning, including: an information storage module, which is used to store the robot's joint movement information under standard motion, the power input information of the servo motor under standard motion, and the robot's basic information; an information acquisition module, which is used to obtain the robot's current ambient temperature, the robot's joint movement information, the power output information of the servo motor, and generate the robot's first motion data and the robot's second motion data. When the robot is in the second state, the accuracy of the photoelectric encoder decreases, and the robot's first motion data is corrected by the robot's second motion data. When the accuracy of the photoelectric encoder decreases, the servo motor's input current is used to correct it, so that the detection accuracy of the robot's photoelectric encoder in the second state can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of robot motion quality detection, and more specifically, to an automated quality detection system based on machine learning. Background Art

[0002] At present, the monitoring of robot motion quality mainly relies on setting photoelectric encoding sensors or Hall effect sensors at the joint positions of the robot to evaluate the robot's motion standard. However, when the robot works in a high temperature environment, the accuracy of the photoelectric encoding sensors or Hall effect sensors decreases due to internal thermal expansion, resistance changes, and electronic performance drift, making it impossible to accurately detect the robot's motion standard. Summary of the invention

[0003] The present invention provides an automated quality inspection system based on machine learning to solve the technical problems in the above-mentioned background technology.

[0004] The present invention provides an automated quality detection system based on machine learning, comprising:

[0005] An information storage module, which is used to store the joint movement information of the robot under standard actions, the power input information of the servo motor under standard actions, and the basic information of the robot;

[0006] Among them, the basic information of the robot includes: the robot's usage time, the robot's size information, the weight information of each part of the robot, and the resistance and magnetic permeability information of the servo motors at each joint position of the robot at different temperatures;

[0007] An information acquisition module, which is used to obtain the current ambient temperature of the robot, the movement information of the robot joints, the power output information of the servo motor, and generate the first motion data of the robot and the second motion data of the robot;

[0008] A parameter correction module, which is used to determine the current state of the robot, and correct the first motion data of the robot according to the current state of the robot and the second motion data of the robot, and generate the third motion data of the robot after the first motion data of the robot is corrected;

[0009] The action standard evaluation module uses the first action data of the robot to calculate the action standard score of the robot when the first action data of the robot has not been corrected; and uses the third action data of the robot to calculate the action standard score of the robot when the first action data of the robot has been corrected.

[0010] Furthermore, the information collection module includes:

[0011] The temperature acquisition unit uses a temperature sensor to obtain the current ambient temperature of the robot;

[0012] The motion acquisition unit uses motion sensors to detect the robot joint movement information in real time, which is recorded as the first motion data of the robot;

[0013] Joint movement information includes: joint movement distance, joint movement angle and movement speed;

[0014] The electrical signal acquisition unit uses a power sensor to detect the power input information of the servo motor;

[0015] The power input information includes: the current and voltage of the servo motor;

[0016] The joint movement angle Δθ is calculated based on the robot joint movement information obtained by the motion sensor, where the calculation formula of the joint movement angle Δθ is as follows:

[0017]

[0018] Where ΔN represents the change in the number of pulses of the photoelectric encoder; N P Indicates the number of pulses per revolution of the photoelectric encoder, N P It is a custom parameter, which is determined according to the actual working condition of the photoelectric encoder;

[0019] The calculation formula of joint movement distance ΔL is as follows:

[0020]

[0021] In the formula, r represents the radius from the rotation center of the robot joint to the moving end point;

[0022] Joint movement speed w a The calculation formula is as follows:

[0023]

[0024] Where Δt represents the time it takes for the robot to move.

[0025] Furthermore, the method for calculating the moving distance and angle of the robot joint according to the power input information of the robot joint servo motor is as follows:

[0026] Step S501, using the initial robot joint movement information acquired by the information acquisition module as the initial position of each joint of the robot;

[0027] Step S502, calculating the output torque and angular velocity of the servo motor according to the current ambient temperature of the robot and the power input information of the servo motor, and the calculation formula includes:

[0028] Servo motor output torque T m The calculation formula is as follows:

[0029] T m =K t ×I×(1+α×(TT r ));

[0030] In the formula, K t represents the torque constant of the servo motor; I represents the input current of the servo motor; T represents the ambient temperature when the servo motor is working; T r Indicates the rated operating temperature of the servo motor; α indicates the temperature coefficient of the servo motor resistance; K t , T r and α are custom parameters;

[0031] Angular velocity w of the servo motor m The calculation formula is as follows:

[0032]

[0033] In the formula, V represents the input voltage of the servo motor; I represents the input current of the servo motor; R represents the internal resistance of the servo motor; K e Indicates the back electromotive force constant of the servo motor, K e is a custom parameter;

[0034] Step S503, construct a robot joint dynamics model, input the robot's usage time, the output torque and angular velocity of the servo motor, the robot's current ambient temperature, the robot's size information, the weight information of each part of the robot, and the resistance and permeability information of the servo motor at each joint position of the robot at different temperatures into the robot joint dynamics model, and output the robot's second motion data, which represents the robot joint movement information.

[0035] Furthermore, the parameter correction module includes:

[0036] A robot state judgment unit, which is used to judge the current state of the robot according to the initial position of the robot, the first motion data of the robot, the second motion data of the robot and the temperature information;

[0037] The current states of the robot include: first state, second state and third state;

[0038] The motion correction unit adopts the first motion data of the robot when the robot is in the first state; corrects the first motion data of the robot to generate the third motion data of the robot when the robot is in the second state; and directly adopts the first motion data of the robot when the robot is in the third state.

[0039] Furthermore, the method for determining the current state of the robot is as follows:

[0040] When the current ambient temperature of the robot acquired by the temperature sensor is within the set temperature threshold, and the difference between the first motion data of the robot and the second motion data of the robot is within the first preset motion threshold range, the robot is in the first state;

[0041] When the current ambient temperature of the robot acquired by the temperature sensor is not within the set temperature threshold, and the difference between the first motion data of the robot and the second motion data of the robot is within the second preset motion threshold range and is not within the first preset motion threshold range, the state of the robot is the second state;

[0042] When the current ambient temperature of the robot acquired by the temperature sensor is within the set temperature threshold, and the difference between the first motion data of the robot and the second motion data of the robot is not within the first preset motion threshold range, the robot is in the third state;

[0043] The first preset motion threshold range and the second preset motion threshold range are both user-defined parameters, and the first preset motion threshold range is included in the second preset motion threshold range.

[0044] Furthermore, the method for correcting the first motion data of the robot is as follows:

[0045] Construct a robot motion correction model, input the sensor usage time, the robot's current ambient temperature, the robot's first motion data and the robot's second motion data into the robot motion correction model, and output the robot's third motion data, which represents the corrected robot joint movement information.

[0046] Further, the robot action correction model includes a first hidden layer, a second hidden layer and a classifier;

[0047] The first hidden layer inputs the sensor usage time, the current ambient temperature of the robot, the first action data of the robot, and the second action data of the robot, and outputs the first updated feature;

[0048] The first updated feature is input into the second hidden layer, and the second updated feature is output;

[0049] The second updated feature is input into the classifier, and the classifier outputs the third action data of the robot.

[0050] Furthermore, the first hidden layer of the robot motion correction model is constructed based on a multi-layer perceptron, and the second hidden layer is constructed based on a transformer.

[0051] Furthermore, the sensor usage time, the robot's first motion data and the robot's second motion data under different temperatures and actions are obtained through experiments as sample data of training samples, and the posture detection model is used to detect the robot's joint movement information as sample labels of the training samples.

[0052] Furthermore, the method for calculating the robot's motion standard is as follows:

[0053] Step S601, when the robot is in the first state, calculating the difference between the first motion data of the robot and the joint movement information under the standard motion;

[0054] Step S602, when the robot is in the second state, calculating the difference between the third motion data and the joint movement information under the standard motion;

[0055] Step S603, when the robot is in the third state, the difference between the first motion data of the robot and the joint movement information under the standard motion;

[0056] Step S604, constructing a first scoring threshold, a second scoring threshold, and a third scoring threshold. When the calculation result falls within the first scoring threshold, the robot's action standard is good; when the calculation result falls within the second scoring threshold, the robot's action standard is medium; when the calculation result falls within the third scoring threshold, the robot's action standard is poor;

[0057] The first scoring threshold, the second scoring threshold and the third scoring threshold are all custom parameters, and the first scoring threshold, the second scoring threshold and the third scoring threshold are arranged in sequence and do not intersect with each other.

[0058] The beneficial effects of the present invention are as follows: when the robot is in the second state, the accuracy of the photoelectric encoder decreases, and the first motion data of the robot is corrected by the second motion data of the robot. When the accuracy of the photoelectric encoder decreases, it is corrected by using the input current of the servo motor, which can improve the detection accuracy of the photoelectric encoder of the robot in the second state. In addition, the current state of the robot is calculated based on the first motion data and the second motion threshold, which is convenient for the operator to quickly troubleshoot the robot's faults. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 It is the overall flow chart of the present invention;

[0060] Figure 2 is a flowchart of calculating the moving distance and angle of the robot joint of the present invention;

[0061] Figure 3 It is a flowchart of the present invention for calculating the standard degree of robot action.

[0062] In the figure: 101, information storage module; 104, action standard evaluation module; 201, temperature acquisition unit; 202, action acquisition unit; 203, electrical signal acquisition unit; 301, robot state judgment unit; 302, action correction unit. DETAILED DESCRIPTION

[0063] The subject matter described herein will now be discussed with reference to example embodiments. It should be understood that the discussion of these embodiments is only to enable those skilled in the art to better understand and implement the subject matter described herein, and the functions and arrangements of the elements discussed may be changed without departing from the scope of protection of the contents of this specification. Each example may omit, replace or add various processes or components as needed. In addition, the features described relative to some examples may also be combined in other examples.

[0064] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in one or more embodiments of the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in one or more embodiments of the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0065] like Figure 1-Figure 3 As shown in the figure, the automated quality inspection system based on machine learning includes:

[0066] An information storage module 101 is used to store the joint movement information of the robot under standard actions, the power input information of the servo motor under standard actions, and the basic information of the robot;

[0067] Among them, the basic information of the robot includes: the robot's usage time, the robot's size information, the weight information of each part of the robot, and the resistance and magnetic permeability information of the servo motors at each joint position of the robot at different temperatures;

[0068] An information acquisition module, which is used to obtain the current ambient temperature of the robot, the movement information of the robot joints, the power output information of the servo motor, and generate the first motion data of the robot and the second motion data of the robot;

[0069] A parameter correction module, which is used to determine the current state of the robot, and correct the first motion data of the robot according to the current state of the robot and the second motion data of the robot, and generate the third motion data of the robot after the first motion data of the robot is corrected;

[0070] The action standard evaluation module 104 uses the first action data of the robot to calculate the action standard score of the robot when the first action data of the robot has not been corrected; and uses the third action data of the robot to calculate the action standard score of the robot when the first action data of the robot has been corrected.

[0071] In one embodiment of the present invention, the information collection module includes:

[0072] The temperature acquisition unit 201 uses a temperature sensor to obtain the current ambient temperature of the robot. The temperature sensor is preferably an infrared sensor. The infrared sensor measures the temperature by detecting the infrared energy radiated by the object. Through the high temperature resistance design, its accuracy generally does not fluctuate greatly due to high temperature. Infrared sensors are relatively mature existing technologies and are not described in detail here.

[0073] The motion collection unit 202 uses a motion sensor to detect the robot joint movement information in real time, which is recorded as the first motion data of the robot;

[0074] The joint movement information includes: joint movement distance, joint movement angle and movement speed. The movement sensor preferably adopts a photoelectric encoder, which is installed on the rotation axis of the robot's joint. As the joint moves, the photoelectric encoder outputs a pulse signal corresponding to the position, thereby calculating the movement angle and movement distance of the joint. The photoelectric encoder belongs to the prior art and will not be described in detail here.

[0075] The electric signal acquisition unit 203 detects the electric power input information of the servo motor using a power sensor;

[0076] The power input information includes: the current and voltage of the servo motor; the power sensor detects the current and voltage of the servo motor at the same time, and can calculate the movement distance and angle of the robot joint based on the input current and voltage information of the servo motor. The power sensor belongs to the existing technology and will not be elaborated here.

[0077] In one embodiment of the present invention, the joint movement angle Δθ can be directly calculated based on the robot joint movement information obtained by the motion sensor, wherein the calculation formula of the joint movement angle Δθ is as follows:

[0078]

[0079] Where ΔN represents the change in the number of pulses of the photoelectric encoder; N P Indicates the number of pulses per revolution of the photoelectric encoder, N P It is a custom parameter, which is determined according to the actual working condition of the photoelectric encoder;

[0080] The calculation formula of joint movement distance ΔL is as follows:

[0081]

[0082] In the formula, r represents the radius from the rotation center of the robot joint to the moving end point;

[0083] Joint movement speed w a The calculation formula is as follows:

[0084]

[0085] Where Δt represents the time it takes for the robot to move.

[0086] In one embodiment of the present invention, the method for calculating the moving distance and angle of the robot joint according to the power input information of the robot joint servo motor is as follows:

[0087] Step S501, using the initial robot joint movement information acquired by the information acquisition module as the initial position of each joint of the robot;

[0088] Step S502, calculating the output torque and angular velocity of the servo motor according to the current ambient temperature of the robot and the power input information of the servo motor, and the calculation formula includes:

[0089] Servo motor output torque T m The calculation formula is as follows:

[0090] T m =K t ×I×(1+α×(TT r ));

[0091] In the formula, K t represents the torque constant of the servo motor; I represents the input current of the servo motor; T represents the ambient temperature when the servo motor is working; T r Indicates the rated operating temperature of the servo motor; α indicates the temperature coefficient of the servo motor resistance; K t , T r and α are custom parameters, which should be confirmed according to the actual situation of the servo motor;

[0092] Angular velocity w of the servo motor m The calculation formula is as follows:

[0093]

[0094] In the formula, V represents the input voltage of the servo motor; I represents the input current of the servo motor; R represents the internal resistance of the servo motor; K e Indicates the back electromotive force constant of the servo motor, K e For custom parameters, please confirm according to the actual situation of the motor;

[0095] Step S503, constructing a robot joint dynamics model, inputting the robot use time, the output torque and angular velocity of the servo motor, the current ambient temperature of the robot, the size information of the robot, the weight information of each part of the robot, and the resistance and permeability information of the servo motor at each joint position of the robot at different temperatures into the robot joint dynamics model, and outputting the second action data of the robot, indicating the movement information of the robot joint;

[0096] Through experiments, the power input information of the servo motor when the current model robot performs an action at a certain temperature is obtained, and its output torque and angular velocity are calculated. The robot usage time, the resistance and magnetic permeability information of the servo motor at the corresponding temperature, the output torque and angular velocity are used as a data sample for training the robot joint dynamics model. According to the above method, the temperature and action are adjusted respectively to generate several data samples;

[0097] The robot's motion detection results using the posture detection model are used as sample labels for training the robot's joint dynamics model;

[0098] The posture detection model is an existing technology and will not be described in detail here.

[0099] In one embodiment of the present invention, the parameter correction module includes:

[0100] A robot state judgment unit 301 is used to judge the current state of the robot according to the initial position of the robot, the first motion data of the robot, the second motion data of the robot and the temperature information;

[0101] The current states of the robot include: first state, second state and third state;

[0102] The motion correction unit 302 adopts the first motion data of the robot when the robot is in the first state; corrects the first motion data of the robot to generate the third motion data of the robot when the robot is in the second state; and directly adopts the first motion data of the robot when the robot is in the third state.

[0103] In one embodiment of the present invention, the method for determining the current state of the robot is as follows:

[0104] When the current ambient temperature of the robot acquired by the temperature sensor is within the set temperature threshold, and the difference between the first motion data of the robot and the second motion data of the robot is within the first preset motion threshold range, the robot is in the first state, indicating that the robot is in a normal state;

[0105] When the current ambient temperature of the robot acquired by the temperature sensor is not within the set temperature threshold, and the difference between the first motion data of the robot and the second motion data of the robot is within the second preset motion threshold range and is not within the first preset motion threshold range, the robot is in the second state, indicating that the temperature is high at this time, there is an error in the accuracy of the robot's sensor output signal, and the robot's joint connection position is fine;

[0106] When the current ambient temperature of the robot acquired by the temperature sensor is within the set temperature threshold, and the difference between the first motion data of the robot and the second motion data of the robot is not within the first preset motion threshold range, the robot is in the third state, indicating that the accuracy of the sensor output signal of the robot is normal, but the joint connection position of the robot has a connection deviation;

[0107] The first preset action threshold range and the second preset action threshold range are both custom parameters, and the first preset action threshold range is included in the second preset action threshold range;

[0108] Through the above method, the problems of decreased sensor accuracy and joint position connection deviation of the robot can be analyzed, which can assist the operator to quickly find the problem of non-standard robot movements and improve the efficiency of subsequent maintenance.

[0109] In one embodiment of the present invention, the method for correcting the first motion data of the robot is as follows:

[0110] Construct a robot motion correction model, input the sensor usage time, the robot's current ambient temperature, the robot's first motion data and the robot's second motion data into the robot motion correction model, and output the robot's third motion data, which represents the corrected robot joint movement information.

[0111] In one embodiment of the present invention, the robot motion correction model includes a first hidden layer, a second hidden layer and a classifier;

[0112] The first hidden layer inputs the sensor usage time, the current ambient temperature of the robot, the first action data of the robot, and the second action data of the robot, and outputs the first updated feature;

[0113] The first updated feature is input into the second hidden layer, and the second updated feature is output;

[0114] The second updated feature is input into the classifier, and the classifier outputs the third action data of the robot.

[0115] In one embodiment of the present invention, the first hidden layer of the robot motion correction model is constructed based on a multi-layer perceptron, and the second hidden layer is constructed based on a transformer.

[0116] In one embodiment of the present invention, the sensor usage time, the robot's first motion data and the robot's second motion data under different temperatures and actions are obtained through experiments as sample data of training samples, and a posture detection model is used to detect the robot's joint movement information as sample labels of the training samples.

[0117] In one embodiment of the present invention, the method for calculating the robot action standard is as follows:

[0118] Step S601, when the robot is in the first state, calculating the difference between the first motion data of the robot and the joint movement information under the standard motion;

[0119] Step S602, when the robot is in the second state, calculating the difference between the third motion data and the joint movement information under the standard motion;

[0120] Step S603, when the robot is in the third state, the difference between the first motion data of the robot and the joint movement information under the standard motion;

[0121] Step S604, constructing a first scoring threshold, a second scoring threshold, and a third scoring threshold. When the calculation result falls within the first scoring threshold, the robot's action standard is good; when the calculation result falls within the second scoring threshold, the robot's action standard is medium; when the calculation result falls within the third scoring threshold, the robot's action standard is poor;

[0122] The first scoring threshold, the second scoring threshold and the third scoring threshold are all custom parameters, and the first scoring threshold, the second scoring threshold and the third scoring threshold are arranged in sequence and do not intersect with each other.

[0123] The above describes an embodiment of the present embodiment, but the present embodiment is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present embodiment, ordinary technicians in this field can also make many forms, all of which are within the protection of the present embodiment.

Claims

1. An automated quality inspection system based on machine learning, characterized in that: include: An information storage module (101) is used to store joint movement information of the robot under standard motion, power input information of the servo motor under standard motion, and basic information of the robot; Among them, the basic information of the robot includes: the robot's usage time, the robot's size information, the weight information of each part of the robot, and the resistance and magnetic permeability information of the servo motors at each joint position of the robot at different temperatures; An information acquisition module, which is used to obtain the current ambient temperature of the robot, the movement information of the robot joints, the power output information of the servo motor, and generate the first motion data of the robot and the second motion data of the robot; The information collection module includes: A temperature acquisition unit (201) uses a temperature sensor to acquire the current ambient temperature of the robot; The motion collection unit (202) uses a motion sensor to detect the robot joint movement information in real time, and records it as the robot's first motion data; Joint movement information includes: joint movement distance, joint movement angle and movement speed; The electric signal acquisition unit (203) detects the electric power input information of the servo motor using a power sensor; The power input information includes: the current and voltage of the servo motor; Calculate the joint movement angle based on the robot joint movement information obtained by the motion sensor , where the joint movement angle The calculation formula is as follows: ； In the formula, ∆N represents the change in the number of pulses of the photoelectric encoder; Indicates the number of pulses per revolution of the photoelectric encoder. It is a custom parameter, which is determined according to the actual working condition of the photoelectric encoder; Joint movement distance The calculation formula is as follows: ； In the formula, Indicates the radius from the rotation center of the robot joint to the moving end point; Joint movement speed The calculation formula is as follows: ； In the formula, represents the time taken by the robot to move; A parameter correction module, which is used to determine the current state of the robot, and correct the first motion data of the robot according to the current state of the robot and the second motion data of the robot, and generate the third motion data of the robot after the first motion data of the robot is corrected; The action standard evaluation module (104) uses the first action data of the robot to calculate the action standard score of the robot when the first action data of the robot has not been corrected; and uses the third action data of the robot to calculate the action standard score of the robot when the first action data of the robot has been corrected.

2. The automated quality inspection system based on machine learning according to claim 1, characterized in that: The method for calculating the movement distance and angle of the robot joint based on the power input information of the robot joint servo motor is as follows: Step S501, using the initial robot joint movement information acquired by the information acquisition module as the initial position of each joint of the robot; Step S502, calculating the output torque and angular velocity of the servo motor according to the current ambient temperature of the robot and the power input information of the servo motor, and the calculation formula includes: Output torque of servo motor The calculation formula is as follows: ； In the formula, Indicates the torque constant of the servo motor; I indicates the input current of the servo motor; T indicates the ambient temperature when the servo motor is working; Indicates the rated operating temperature of the servo motor; Indicates the temperature coefficient of the servo motor resistance; , and All are custom parameters; Angular velocity of the servo motor The calculation formula is as follows: ； In the formula, V represents the input voltage of the servo motor; I represents the input current of the servo motor; R represents the internal resistance of the servo motor; represents the back electromotive force constant of the servo motor, is a custom parameter; Step S503, construct a robot joint dynamics model, input the robot's usage time, the output torque and angular velocity of the servo motor, the robot's current ambient temperature, the robot's size information, the weight information of each part of the robot, and the resistance and permeability information of the servo motor at each joint position of the robot at different temperatures into the robot joint dynamics model, and output the robot's second motion data, which represents the robot joint movement information.

3. The automated quality inspection system based on machine learning according to claim 2, characterized in that: The parameter correction module includes: A robot state judgment unit (301), which is used to judge the current state of the robot according to the initial position of the robot, the first motion data of the robot, the second motion data of the robot and the temperature information; The current states of the robot include: first state, second state and third state; The motion correction unit (302) adopts the first motion data of the robot when the robot is in the first state; corrects the first motion data of the robot to generate the third motion data of the robot when the robot is in the second state; and directly adopts the first motion data of the robot when the robot is in the third state.

4. The automated quality inspection system based on machine learning according to claim 3, characterized in that: The method to determine the current state of the robot is as follows: When the current ambient temperature of the robot acquired by the temperature sensor is within the set temperature threshold, and the difference between the first motion data of the robot and the second motion data of the robot is within the first preset motion threshold range, the robot is in the first state; When the current ambient temperature of the robot acquired by the temperature sensor is not within the set temperature threshold, and the difference between the first motion data of the robot and the second motion data of the robot is within the second preset motion threshold range and is not within the first preset motion threshold range, the state of the robot is the second state; When the current ambient temperature of the robot acquired by the temperature sensor is within the set temperature threshold, and the difference between the first motion data of the robot and the second motion data of the robot is not within the first preset motion threshold range, the robot is in the third state; The first preset motion threshold range and the second preset motion threshold range are both user-defined parameters, and the first preset motion threshold range is included in the second preset motion threshold range.

5. The automated quality inspection system based on machine learning according to claim 4, characterized in that: The method to correct the robot's first action data is as follows: Construct a robot motion correction model, input the sensor usage time, the robot's current ambient temperature, the robot's first motion data and the robot's second motion data into the robot motion correction model, and output the robot's third motion data, which represents the corrected robot joint movement information.

6. The automated quality inspection system based on machine learning according to claim 5, characterized in that: The robot action correction model includes a first hidden layer, a second hidden layer and a classifier; The first hidden layer inputs the sensor usage time, the current ambient temperature of the robot, the first action data of the robot, and the second action data of the robot, and outputs the first updated feature; The first updated feature is input into the second hidden layer, and the second updated feature is output; The second updated feature is input into the classifier, and the classifier outputs the third action data of the robot.

7. The automated quality inspection system based on machine learning according to claim 6, characterized in that: The first hidden layer of the robot motion correction model is built based on a multi-layer perceptron, and the second hidden layer is built based on a transformer.

8. The automated quality inspection system based on machine learning according to claim 5, characterized in that: Through experiments, the sensor usage time, the robot's first action data and the robot's second action data under different temperatures and actions are obtained as sample data of training samples, and the posture detection model is used to detect the robot's joint movement information as sample labels of training samples.

9. The automated quality inspection system based on machine learning according to claim 5, characterized in that: The method for calculating the robot's motion standard is as follows: Step S601, when the robot is in the first state, calculating the difference between the first motion data of the robot and the joint movement information under the standard motion; Step S602, when the robot is in the second state, calculating the difference between the third motion data and the joint movement information under the standard motion; Step S603, when the robot is in the third state, calculating the difference between the first motion data of the robot and the joint movement information under the standard motion; Step S604, constructing a first scoring threshold, a second scoring threshold, and a third scoring threshold. When the calculation result falls within the first scoring threshold, the robot's action standard is good; when the calculation result falls within the second scoring threshold, the robot's action standard is medium; when the calculation result falls within the third scoring threshold, the robot's action standard is poor; The first scoring threshold, the second scoring threshold and the third scoring threshold are all custom parameters, and the first scoring threshold, the second scoring threshold and the third scoring threshold are arranged in sequence and do not intersect with each other.

Images