Robot intelligent programming method and system based on deep learning

Abstract

The present invention discloses a robot intelligent programming method and system based on deep learning, relates to the field of industrial robots, and solves the problem of low efficiency of current parameter adjustment methods of industrial robots. The method is specifically as follows: a management terminal uploads specification information of a packaging box; an intelligent learning module intelligently learns the clamping work of the industrial robot to obtain clamping parameters of the packaging box; a mapping relationship is established between the clamping parameters and the industrial robot; the top surface image of the packaging box on the conveyor belt is collected; a parameter comparison module compares and identifies the packaging box transmitted on the conveyor belt, and if the recognition generates an intelligent learning signal, the specification information of the packaging box is uploaded and learned; if the workstation number of the industrial robot is obtained, it is sent to a processor, and the processor generates a work instruction according to the workstation number and loads it to the corresponding industrial robot, and the industrial robot clamps and transfers the packaging box that arrives at the clamping point. The present invention realizes intelligent adjustment of the working parameters of the industrial robot through deep learning.

Description

Technical Field

[0001] The present invention belongs to the field of robot automation technology, and specifically relates to a robot intelligent programming method and system based on deep learning. Background Art

[0002] Industrial robots are multi-joint manipulators or multi-degree-of-freedom machine devices widely used in the industrial field. They have a certain degree of automation and can achieve various industrial processing and manufacturing functions by relying on their own power and control capabilities. Industrial robots are widely used in various industrial fields such as electronics, logistics, and chemicals.

[0003] In the prior art, industrial robots are used for the automated transfer of product packaging in assembly lines. However, when industrial robots clamp product packaging of different specifications, the corresponding clamping parameters are usually adjusted through the parameter panel on the industrial robot. This parameter adjustment method is not only inefficient, but also requires repeated debugging of the parameters to adapt to the corresponding product packaging.

[0004] To this end, we propose a robot intelligent programming method and system based on deep learning. Summary of the invention

[0005] The purpose of the present invention is to propose a robot intelligent programming method and system based on deep learning to solve the problems raised in the above background technology.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] First, the robot intelligent programming method based on deep learning, the robot intelligent programming method is as follows:

[0008] Step S10, the management terminal uploads the specification information of the packaging box to the server, and the server sends the specification information to the intelligent learning module;

[0009] Step S20, the intelligent learning module performs intelligent learning on the clamping work of the industrial robot, obtains the clamping parameters of the packaging box and sends them to the processor via the server, and the processor sends the clamping parameters of the packaging box to the corresponding execution component;

[0010] Step S30, the processor feeds back the mapping relationship between the clamping parameters and the industrial robot to the server, and the server sends the clamping parameters and the mapping relationship between the clamping parameters and the industrial robot to the storage module for storage;

[0011] Step S40, when working, the first acquisition module is used to acquire the top surface image of the packaging box on the conveyor belt, and the top surface image is sent to the parameter comparison module via the processor and the server;

[0012] In step S50, the parameter comparison module compares and identifies the packaging box transmitted on the conveyor belt. If the recognition generates an intelligent learning signal, the specification information of the packaging box is uploaded and learned. If the work station number of the industrial robot is recognized, it is sent to the processor. The processor generates a work instruction based on the work station number and loads it to the corresponding industrial robot. The industrial robot clamps and transfers the packaging box that arrives at the clamping point.

[0013] Preferably, the specification information is the box body length, box body width and box body height of the packaging box.

[0014] Preferably, in step S20, the working process of the intelligent learning module is as follows:

[0015] Obtain the specification information of the packaging box, and obtain the box length, box width and box height of the packaging box;

[0016] Obtain three pairs of reference surfaces of the packaging box according to the box length, box width and box height;

[0017] The three pairs of reference planes are divided into the first reference plane, the second reference plane and the third reference plane in order from the smallest to the largest area;

[0018] Obtaining the maximum side length and the minimum side length of the first reference plane, and using the maximum side length and the minimum side length in the first reference plane as the first clamping parameter of the corresponding packaging box;

[0019] Then, the maximum side length and the minimum side length in another pair of reference surfaces are obtained, and the maximum side length and the minimum side length in the pair of reference surfaces are used as the second clamping parameters of the corresponding packaging box;

[0020] The maximum side length and the minimum side length in the last pair of reference planes are obtained, and the maximum side length and the minimum side length in the pair of reference planes are used as the third clamping parameter of the corresponding packaging box.

[0021] Preferably, in step S20, the corresponding relationship between the clamping parameters and the actuator is specifically:

[0022] The industrial robots are arranged in sequence according to the transmission direction, and the gripping parameters of the industrial robots are increased in sequence along the transmission direction;

[0023] Load the first gripping parameter of the package box to the leftmost industrial robot, and so on, load the second gripping parameter and the third gripping parameter of the package box to the corresponding industrial robots in sequence;

[0024] At the same time, a mapping relationship is established between the clamping parameters and the corresponding industrial robots, and the mapping relationship is the clamping parameters and the station number of the corresponding industrial robot.

[0025] Preferably, in step S50, the working process of the parameter comparison module is as follows:

[0026] Obtain two sets of measured side lengths of the packaging box on the conveyor belt based on the top surface image;

[0027] Compare the two groups of measured side lengths of the packaging box with the clamping parameters of the packaging box in the storage module one by one;

[0028] If the two sets of measured side lengths meet any set of clamping parameters, the station number of the industrial robot corresponding to the clamping parameters is obtained according to the mapping relationship;

[0029] If the two sets of measured side lengths do not meet all clamping parameters, an intelligent learning signal is generated.

[0030] Preferably, the robot intelligent programming method also includes:

[0031] Step S60, the second acquisition module records the time nodes when the packaging boxes on the conveyor belt arrive at different measurement lines and the real-time positions of the packaging boxes on the conveyor belt, and sends the time nodes and real-time positions to the control analysis module via the server;

[0032] Step S70, the control analysis module intelligently controls the industrial robot corresponding to the packaging box on the conveyor belt, and obtains a control packet of the industrial robot corresponding to the packaging box and sends it to the processor via the server;

[0033] Step S80: the processor performs a clamping and transferring operation on the industrial robot corresponding to the packaging box according to the control package.

[0034] Preferably, in step S70, the operation process of the manipulation analysis module is specifically as follows:

[0035] Acquire the top surface image of the packaging box on the conveyor belt;

[0036] If the outer edge of the top surface image is perpendicular to the conveyor belt, there is no offset angle;

[0037] If the outer edge of the top surface image is not perpendicular to the conveyor belt, the obtained offset angle is recorded as the adjustment offset angle of the industrial robot corresponding to the packaging box, and the adjustment direction of the industrial robot corresponding to the packaging box is established at the same time;

[0038] Then, the time nodes when the packaging box on the conveyor belt arrives at different measuring lines are obtained. The time node when the packaging box arrives at the first measuring line is the first time node, and the time node when the packaging box arrives at the second measuring line is the second time node. The time node is the time when the geometric center in the top surface image arrives at the measuring line.

[0039] The interval distance between the first measuring line and the second measuring line is known, the time difference is obtained by subtracting the first time node from the second time node, and the moving speed of the packaging box on the conveyor belt is obtained by dividing the interval distance by the time difference;

[0040] Obtain the real-time position of the packaging box on the conveyor belt, and then obtain the gripping point of the corresponding industrial robot where the packaging box is to reach; wherein the real-time position is the real-time position of the geometric center in the top surface image;

[0041] Calculate the distance between the real-time position of the packaging box and the clamping point and record it as the remaining transmission distance. Divide the remaining transmission distance by the moving speed to get the transmission time for the packaging box to reach the clamping point.

[0042] The transmission time is used as the control reaction time of the industrial robot corresponding to the packaging box;

[0043] The adjustment of the offset angle, the adjustment direction and the control response time are packaged and integrated into a control package for the industrial robot corresponding to the packaging box.

[0044] Preferably, in step S80, the working process of the industrial robot is specifically as follows:

[0045] The geometric center of the packaging box is obtained based on the top surface image. When the geometric center reaches the calibration line, the dynamic calibration point under the tracking calibration device moves along with the geometric center.

[0046] The industrial robot corresponding to the packaging box is regulated according to the adjustment of the offset angle and the adjustment direction and completed within the control reaction time;

[0047] When the packaging box reaches the corresponding clamping point of the industrial robot, the dynamic calibration point corresponds to the static calibration point up and down, and the industrial robot clamps and transfers the packaging box.

[0048] Preferably, the process of obtaining the offset angle and the adjustment direction is specifically as follows:

[0049] Connect two sets of diagonal lines of the top surface image, the intersection of the two sets of diagonal lines is the geometric center of the top surface image, and construct a first rectangular coordinate system with the geometric center of the top surface image. Then adjust the top surface image so that any outer edge of the top surface image is perpendicular to the conveyor belt. At this time, construct a second rectangular coordinate system with the geometric center of the top surface image. The angle between the Y' axis in the first rectangular coordinate system and the Y axis in the second rectangular coordinate system is the offset angle.

[0050] The initial state of the industrial robot is the same as the second rectangular coordinate system. If the second rectangular coordinate system is rotated counterclockwise, the adjustment direction of the industrial robot is clockwise. If the second rectangular coordinate system is rotated clockwise, the adjustment direction of the industrial robot is counterclockwise.

[0051] In the second aspect, a robot intelligent programming system based on deep learning includes a processor and a server connected to the processor, a first acquisition module and a plurality of industrial robots, the server includes a second acquisition module, an intelligent learning module, a storage module, a parameter comparison module and a control analysis module, and the server is also connected to a management terminal;

[0052] A management terminal is used to upload the specification information of the packaging box to the server, and the server sends the specification information to the intelligent learning module;

[0053] An intelligent learning module is used to intelligently learn the clamping work of the industrial robot, obtain the clamping parameters of the packaging box and send them to the corresponding execution components through the server and processor;

[0054] A processor, used for feeding back the mapping relationship between the clamping parameters and the industrial robot to the server, and the server sends the clamping parameters and the mapping relationship between the clamping parameters and the industrial robot to the storage module for storage;

[0055] When working;

[0056] The first acquisition module is used to acquire the top surface image of the packaging box on the conveyor belt, and the top surface image is sent to the parameter comparison module and the control analysis module via the processor and the server;

[0057] The parameter comparison module is used to compare and identify the packaging boxes transmitted on the conveyor belt. If the identification generates an intelligent learning signal and feeds it back to the server, the management terminal uploads and learns the specification information of the packaging boxes on the conveyor belt. If the station number of the industrial robot is identified, it is sent to the processor. The processor generates a work instruction based on the station number and loads it to the corresponding industrial robot. The industrial robot clamps and transfers the packaging box that arrives at the clamping point.

[0058] The second acquisition module is used to record the time nodes when the packaging boxes on the conveyor belt arrive at different measurement lines and the real-time position of the packaging boxes on the conveyor belt. The time nodes and real-time positions are sent to the control analysis module via the server;

[0059] A control analysis module is used to intelligently control the industrial robot corresponding to the packaging box on the conveyor belt, and obtain a control package of the industrial robot corresponding to the packaging box and send it to the processor via the server;

[0060] The processor is also used to perform clamping and transferring work on the packaging box corresponding to the industrial robot according to the control package.

[0061] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0062] The present invention uses a management terminal to upload the specification information of the packaging box, and then uses an intelligent learning module to intelligently learn the clamping work of the industrial robot, obtains the clamping parameters of the packaging box and sends them to the corresponding execution component, establishes a mapping relationship between the clamping parameters and the corresponding industrial robot, obtains the top surface image of the packaging box on the conveyor belt during operation and sends it to the parameter comparison module, the parameter comparison module compares and identifies the packaging box transmitted on the conveyor belt, if the recognition generates an intelligent learning signal, the specification information of the packaging box is uploaded and learned, if the work station number is recognized, a work instruction is generated and loaded into the corresponding industrial robot, the industrial robot clamps and transfers the packaging box that arrives at the clamping point, and the present invention realizes intelligent adjustment of the working parameters of the industrial robot through deep learning. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In order to facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0064] Figure 1 is a flow chart of the method of the present invention;

[0065] Figure 2 is another method flow chart of the present invention;

[0066] Figure 3 is a system block diagram of the present invention;

[0067] Figure 4 It is a structural schematic diagram of the packaging box in the present invention;

[0068] Figure 5 It is a schematic diagram of the structure of the transmission belt in the present invention;

[0069] Figure 6 It is a side view of the tracking calibration device of the present invention;

[0070] Figure 7 It is a structural schematic diagram of the offset angle in the present invention. DETAILED DESCRIPTION

[0071] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than 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.

[0072] Embodiment 1:

[0073] See also Figure 1-Figure 7 As shown, the present invention provides a technical solution: a robot intelligent programming method based on deep learning, and the robot intelligent programming method is specifically as follows:

[0074] Step S10, the management terminal uploads the specification information of the packaging box to the server, and the server sends the specification information to the intelligent learning module;

[0075] The specification information includes the length, width and height of the packaging box.

[0076] Step S20, the intelligent learning module performs intelligent learning on the clamping work of the industrial robot, obtains the clamping parameters of the packaging box and sends them to the processor via the server, and the processor sends the clamping parameters of the packaging box to the corresponding execution component;

[0077] In step S20, the working process of the intelligent learning module is as follows:

[0078] Step S201, obtaining specification information of the packaging box, and obtaining the box length, box width and box height of the packaging box;

[0079] Step S202, obtaining three pairs of reference surfaces of the packaging box according to the box length, box width and box height;

[0080] Step S203, dividing the three pairs of reference planes into a first reference plane, a second reference plane and a third reference plane in order from small to large areas;

[0081] Step S204, obtaining the maximum side length and the minimum side length of the first reference plane, and using the maximum side length and the minimum side length of the first reference plane as the first clamping parameter of the corresponding packaging box;

[0082] Step S205, then obtaining the maximum side length and the minimum side length in another pair of reference surfaces, and using the maximum side length and the minimum side length in the pair of reference surfaces as the second clamping parameter of the corresponding packaging box;

[0083] Step S206, obtaining the maximum side length and the minimum side length in the last pair of reference surfaces, and using the maximum side length and the minimum side length in the pair of reference surfaces as the third clamping parameter of the corresponding packaging box;

[0084] For example, a box like Figure 4 As shown, A3-A4-A5-A6 is a first reference plane, the first reference plane and another first reference plane corresponding thereto form a pair of reference planes, the side length of the A3-A4 side and the side length of the A3-A6 side are used as the second clamping parameter of the packaging box; A1-A2-A3-A4 is a second reference plane, the second reference plane and another second reference plane corresponding thereto form a pair of reference planes, the side length of the A1-A2 side and the side length of the A1-A4 side are used as the first clamping parameter of the packaging box; A2-A3-A6-A7 is a third reference plane, the third reference plane and another third reference plane corresponding thereto form a pair of reference planes, the side length of the A2-A7 side and the side length of the A2-A3 side are used as the third clamping parameter of the packaging box;

[0085] In step S20, the corresponding relationship between the clamping parameters and the actuator components is specifically as follows:

[0086] like Figure 5 and Figure 6 As shown, the industrial robots are arranged in sequence according to the transmission direction, and the gripping parameters of the industrial robots increase in sequence along the transmission direction;

[0087] Load the first gripping parameter of the package box to the leftmost industrial robot, and so on, load the second gripping parameter and the third gripping parameter of the package box to the corresponding industrial robots in sequence;

[0088] At the same time, a mapping relationship is established between the clamping parameters and the corresponding industrial robots, and the mapping relationship is the clamping parameters and the station number of the corresponding industrial robots;

[0089] Step S30, the processor feeds back the mapping relationship between the clamping parameters and the industrial robot to the server, and the server sends the clamping parameters and the mapping relationship between the clamping parameters and the industrial robot to the storage module for storage;

[0090] Step S40: during operation, the first acquisition module acquires the top surface image of the packaging box on the conveyor belt, and the top surface image is sent to the parameter comparison module and the control analysis module via the processor and the server;

[0091] Step S50, the parameter comparison module compares and identifies the packaging box transmitted on the conveyor belt. If the recognition generates an intelligent learning signal, the specification information of the packaging box is uploaded and learned. If the station number of the industrial robot is recognized, it is sent to the processor. The processor generates a work instruction according to the station number and loads it to the corresponding industrial robot. The industrial robot clamps and transfers the packaging box that reaches the clamping point.

[0092] In step S50, the working process of the parameter comparison module is as follows:

[0093] Step S501, obtaining two sets of measured side lengths of the packaging box on the conveyor belt according to the top surface image;

[0094] Step S502, comparing the two groups of measured side lengths of the packaging box with the clamping parameters of the packaging box in the storage module one by one;

[0095] Step S503, if the two groups of measured side lengths meet any group of clamping parameters, the station number of the industrial robot corresponding to the clamping parameters is obtained according to the mapping relationship;

[0096] Step S504: if the two sets of measured side lengths do not meet all clamping parameters, an intelligent learning signal is generated.

[0097] like Figure 2 , Figure 5-Figure 7As shown, as a further solution of the present invention, the robot intelligent programming method based on deep learning also includes:

[0098] Step S60, the second acquisition module records the time nodes when the packaging boxes on the conveyor belt arrive at different measurement lines and the real-time positions of the packaging boxes on the conveyor belt, and sends the time nodes and real-time positions to the control analysis module via the server;

[0099] Step S70, the control analysis module intelligently controls the industrial robot corresponding to the packaging box on the conveyor belt, and obtains a control packet of the industrial robot corresponding to the packaging box and sends it to the processor via the server;

[0100] In step S70, the operation process of the manipulation analysis module is as follows:

[0101] Step S701, acquiring a top surface image of a packaging box on a conveyor belt;

[0102] Step S702: if the outer edge of the top surface image is perpendicular to the transmission belt, then there is no offset angle;

[0103] Step S703, if the outer edge of the top surface image is not perpendicular to the conveyor belt, the angle of the offset angle is obtained and recorded as the adjustment offset angle of the industrial robot corresponding to the packaging box, and the adjustment direction of the industrial robot corresponding to the packaging box is determined;

[0104] According to Figure 7 As shown, two groups of diagonal lines of the top surface image are connected, and the intersection of the two groups of diagonal lines is the geometric center of the top surface image. The first rectangular coordinate system is constructed with the geometric center of the top surface image, and then the top surface image is adjusted so that any outer edge of the top surface image is perpendicular to the conveyor belt. At this time, the second rectangular coordinate system is constructed with the geometric center of the top surface image. The angle formed by the Y' axis in the first rectangular coordinate system and the Y axis in the second rectangular coordinate system is the offset angle. The initial state of the industrial robot is the same as the second rectangular coordinate system. If the second rectangular coordinate system is obtained by counterclockwise rotation, the adjustment direction of the industrial robot is clockwise adjustment. If the second rectangular coordinate system is obtained by clockwise rotation, the adjustment direction of the industrial robot is counterclockwise adjustment.

[0105] Step S704, then obtaining the time nodes when the packaging box on the conveyor belt arrives at different measurement lines, the time node when the packaging box arrives at the first measurement line is the first time node, and the time node when the packaging box arrives at the second measurement line is the second time node; the time node is the time when the geometric center in the top surface image arrives at the measurement line;

[0106] Step S705: Given the interval between the first measurement line and the second measurement line, the time difference is obtained by subtracting the first time node from the second time node, and the moving speed of the packaging box on the conveyor belt is obtained by dividing the interval by the time difference;

[0107] Step S706, obtaining the real-time position of the packaging box on the conveyor belt, and then obtaining the gripping point of the corresponding industrial robot where the packaging box is to arrive; wherein the real-time position is the real-time position of the geometric center in the top surface image;

[0108] Step S707, using the distance formula to calculate the distance between the real-time position of the packaging box and the clamping point and record it as the remaining transmission distance, and dividing the remaining transmission distance by the moving speed to obtain the transmission time for the packaging box to reach the clamping point;

[0109] Step S708, using the transmission time as the control reaction time of the industrial robot corresponding to the packaging box;

[0110] Step S709, packaging and integrating the adjustment of the offset angle, the adjustment direction and the control reaction time into a control package of the industrial robot corresponding to the packaging box;

[0111] Step S80, the processor performs a gripping and transferring operation on the industrial robot corresponding to the packaging box according to the control package;

[0112] In step S80, the working process of the industrial robot is specifically as follows:

[0113] Step S801, obtaining the geometric center of the packaging box according to the top surface image, when the geometric center reaches the calibration line, the dynamic calibration point under the tracking calibration device moves along with the geometric center;

[0114] Step S802, the industrial robot corresponding to the packaging box is regulated according to the adjustment offset angle and the adjustment direction and completed within the control reaction time;

[0115] Step S803, when the packaging box reaches the corresponding gripping point of the industrial robot, the dynamic calibration point corresponds to the static calibration point up and down, and the industrial robot grips and transfers the packaging box;

[0116] In this application, if corresponding calculation formulas appear, the above calculation formulas are all dimensionless and take their numerical calculations. The weight coefficients, proportional coefficients and other coefficients in the formulas are set to a result value obtained by quantifying each parameter. The size of the weight coefficient and the proportional coefficient can be determined as long as it does not affect the proportional relationship between the parameter and the result value.

[0117] Embodiment 2:

[0118] Based on another concept of the same invention, please refer to Figure 3-Figure 7 As shown, a robot intelligent programming system based on deep learning is proposed, including a processor and a server connected to the processor, a first acquisition module and a plurality of industrial robots, the server including a second acquisition module, an intelligent learning module, a storage module, a parameter comparison module and a control analysis module, and the server is also connected to a management terminal;

[0119] In this embodiment, the robot intelligent programming system is used to clamp and transfer the packaging boxes in the conveyor belt. Here, only the packaging boxes are considered to be regular rectangular bodies. The system involves a conveyor belt, multiple groups of industrial robots, and tracking and calibration devices, such as Figure 5 and Figure 6 As shown, the conveyor belt is provided with a transmission center line, and correction devices are provided on both sides of the conveyor belt. The conveyor belt is provided with a calibration line, a first measuring line, and a second measuring line in sequence along the transmission direction. The correction device can make the geometric center of the top surface image of the packaging box move along the transmission center line all the time when the packaging box is transmitted. At the same time, the conveyor belt is controlled with clamping points corresponding to the industrial robot, and the corresponding clamping head of the industrial robot is located directly above the clamping points. At the same time, a tracking calibration device is provided at the top of the conveyor belt and the industrial robot. The tracking calibration device can reciprocate along the transmission center line. A dynamic calibration point is provided directly below the tracking calibration device. The dynamic calibration point is used to track the geometric center of the top surface image of the packaging box. A static calibration point is provided directly above the industrial robot, and the static calibration point is adapted to the dynamic calibration point.

[0120] The management terminal is used to upload the specification information of the packaging box to the server, and the server sends the specification information to the intelligent learning module;

[0121] The specification information includes the length, width and height of the packaging box.

[0122] The intelligent learning module is used to perform intelligent learning on the clamping work of the industrial robot. The working process is as follows:

[0123] Obtain the specification information of the packaging box, and obtain the box length, box width and box height of the packaging box;

[0124] Obtain three pairs of reference surfaces of the packaging box according to the box length, box width and box height;

[0125] The three pairs of reference planes are divided into the first reference plane, the second reference plane and the third reference plane in order from the smallest to the largest area;

[0126] Obtaining the maximum side length and the minimum side length of the first reference plane, and using the maximum side length and the minimum side length in the first reference plane as the first clamping parameter of the corresponding packaging box;

[0127] Then, the maximum side length and the minimum side length in another pair of reference surfaces are obtained, and the maximum side length and the minimum side length in the pair of reference surfaces are used as the second clamping parameters of the corresponding packaging box;

[0128] Obtain the maximum side length and the minimum side length in the last pair of reference surfaces, and use the maximum side length and the minimum side length in the pair of reference surfaces as the third clamping parameter of the corresponding packaging box;

[0129] For example, a box like Figure 4 As shown, A3-A4-A5-A6 is a first reference plane, the first reference plane and another first reference plane corresponding thereto form a pair of reference planes, the side length of the A3-A4 side and the side length of the A3-A6 side are used as the second clamping parameter of the packaging box; A1-A2-A3-A4 is a second reference plane, the second reference plane and another second reference plane corresponding thereto form a pair of reference planes, the side length of the A1-A2 side and the side length of the A1-A4 side are used as the first clamping parameter of the packaging box; A2-A3-A6-A7 is a third reference plane, the third reference plane and another third reference plane corresponding thereto form a pair of reference planes, the side length of the A2-A7 side and the side length of the A2-A3 side are used as the third clamping parameter of the packaging box;

[0130] The intelligent learning module feeds back the clamping parameters of the packaging box to the server, and the server also sends the clamping parameters of the packaging box to the processor, and the processor sends the clamping parameters of the packaging box to the corresponding execution component, specifically:

[0131] like Figure 5 As shown, the industrial robots are arranged in sequence according to the transmission direction, and the gripping parameters of the industrial robots increase in sequence along the transmission direction;

[0132] Load the first gripping parameter of the package box to the leftmost industrial robot, and so on, load the second gripping parameter and the third gripping parameter of the package box to the corresponding industrial robots in sequence;

[0133] At the same time, a mapping relationship is established between the clamping parameters and the corresponding industrial robots, and the mapping relationship is the clamping parameters and the station number of the corresponding industrial robots;

[0134] The processor feeds back the mapping relationship between the clamping parameters and the industrial robot to the server, and the server sends the clamping parameters and the mapping relationship between the clamping parameters and the industrial robot to the storage module for storage.

[0135] In actual operation, the first acquisition module is used to acquire the top surface image of the packaging box on the conveyor belt and send the top surface image to the processor, the processor sends the top surface image to the server, and the server sends the top surface image to the parameter comparison module and the manipulation analysis module;

[0136] The parameter comparison module is used to compare and identify the packaging boxes transmitted on the conveyor belt. The working process is as follows:

[0137] Obtain two sets of measured side lengths of the packaging box on the conveyor belt based on the top surface image;

[0138] Compare the two groups of measured side lengths of the packaging box with the clamping parameters of the packaging box in the storage module one by one;

[0139] If the two sets of measured side lengths meet any set of clamping parameters, the station number of the industrial robot corresponding to the clamping parameters is obtained according to the mapping relationship;

[0140] If the two sets of measured side lengths do not meet all clamping parameters, an intelligent learning signal is generated;

[0141] The parameter comparison module feeds back the work station number or intelligent learning signal of the industrial robot to the server. If the server receives the intelligent learning signal, it will forward it to the management terminal. The management terminal is used to upload the specification information of the packaging box on the conveyor belt after receiving the intelligent learning signal, and then repeat the above steps through the intelligent learning module. If the server receives the work station number of the industrial robot, it will forward it to the processor. The processor is used to generate a work instruction based on the work station number and load it to the corresponding industrial robot. The industrial robot is used to clamp and transfer the packaging box that arrives at the clamping point after receiving the work instruction.

[0142] As a further solution of the present invention, the second acquisition module is used to record the time nodes when the packaging boxes on the conveyor belt arrive at different measurement lines and the real-time positions of the packaging boxes on the conveyor belt, and send the time nodes and real-time positions to the server, and the server sends the time nodes and real-time positions to the control analysis module;

[0143] The control analysis module is used to intelligently control the industrial robot corresponding to the packaging box on the conveyor belt. The working process is as follows:

[0144] Acquire the top surface image of the packaging box on the conveyor belt;

[0145] If the outer edge of the top surface image is perpendicular to the conveyor belt, there is no offset angle;

[0146] If the outer edge of the top surface image is not perpendicular to the conveyor belt, the obtained offset angle is recorded as the adjustment offset angle of the industrial robot corresponding to the packaging box, and the adjustment direction of the industrial robot corresponding to the packaging box is established at the same time; Figure 7 As shown, two groups of diagonal lines of the top surface image are connected, and the intersection of the two groups of diagonal lines is the geometric center of the top surface image. The first rectangular coordinate system is constructed with the geometric center of the top surface image, and then the top surface image is adjusted so that any outer edge of the top surface image is perpendicular to the conveyor belt. At this time, the second rectangular coordinate system is constructed with the geometric center of the top surface image. The angle formed by the Y' axis in the first rectangular coordinate system and the Y axis in the second rectangular coordinate system is the offset angle. The initial state of the industrial robot is the same as the second rectangular coordinate system. If the second rectangular coordinate system is obtained by counterclockwise rotation, the adjustment direction of the industrial robot is clockwise adjustment. If the second rectangular coordinate system is obtained by clockwise rotation, the adjustment direction of the industrial robot is counterclockwise adjustment.

[0147] Then, the time nodes when the packaging box on the conveyor belt arrives at different measuring lines are obtained. The time node when the packaging box arrives at the first measuring line is the first time node, and the time node when the packaging box arrives at the second measuring line is the second time node. The time node is the time when the geometric center in the top surface image arrives at the measuring line.

[0148] The interval distance between the first measuring line and the second measuring line is known, the time difference is obtained by subtracting the first time node from the second time node, and the moving speed of the packaging box on the conveyor belt is obtained by dividing the interval distance by the time difference;

[0149] Obtain the real-time position of the packaging box on the conveyor belt, and then obtain the gripping point of the corresponding industrial robot where the packaging box is to reach; wherein the real-time position is the real-time position of the geometric center in the top surface image;

[0150] The distance between the real-time position of the packaging box and the clamping point is calculated using the distance formula and recorded as the remaining transmission distance. The remaining transmission distance is divided by the moving speed to obtain the transmission time for the packaging box to reach the clamping point.

[0151] The transmission time is used as the control reaction time of the industrial robot corresponding to the packaging box;

[0152] The adjustment of the offset angle, the adjustment direction and the control reaction time are packaged and integrated into a control package corresponding to the industrial robot in the packaging box;

[0153] The control analysis module feeds back the control package of the industrial robot corresponding to the packaging box to the server, and the server sends the control package of the industrial robot corresponding to the packaging box to the processor, and the processor is used to perform the clamping and transfer work on the industrial robot corresponding to the packaging box according to the control package, specifically:

[0154] The geometric center of the packaging box is obtained based on the top surface image. When the geometric center reaches the calibration line, the dynamic calibration point under the tracking calibration device moves along with the geometric center.

[0155] The industrial robot corresponding to the packaging box is regulated according to the adjustment of the offset angle and the adjustment direction and completed within the control reaction time;

[0156] When the packaging box reaches the corresponding clamping point of the industrial robot, the dynamic calibration point corresponds to the static calibration point up and down, and the industrial robot clamps and transfers the packaging box.

[0157] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A robot intelligent programming method based on deep learning, characterized in that: The specific method of robot intelligent programming is as follows: Step S10, the management terminal uploads the specification information of the packaging box to the server, and the server sends the specification information to the intelligent learning module; Step S20, the intelligent learning module performs intelligent learning on the clamping work of the industrial robot, obtains the clamping parameters of the packaging box and sends them to the processor via the server, and the processor sends the clamping parameters of the packaging box to the corresponding execution component; In step S20, the corresponding relationship between the clamping parameters and the actuator components is specifically as follows: The industrial robots are arranged in sequence according to the transmission direction, and the gripping parameters of the industrial robots are increased in sequence along the transmission direction; Load the first gripping parameter of the package box to the leftmost industrial robot, and so on, load the second gripping parameter and the third gripping parameter of the package box to the corresponding industrial robots in sequence; At the same time, a mapping relationship is established between the clamping parameters and the corresponding industrial robots, and the mapping relationship is the clamping parameters and the station number of the corresponding industrial robots; Step S30, the processor feeds back the mapping relationship between the clamping parameters and the industrial robot to the server, and the server sends the clamping parameters and the mapping relationship between the clamping parameters and the industrial robot to the storage module for storage; Step S40, when working, the first acquisition module is used to acquire the top surface image of the packaging box on the conveyor belt, and the top surface image is sent to the parameter comparison module via the processor and the server; Step S50, the parameter comparison module compares and identifies the packaging box transmitted on the conveyor belt. If the recognition generates an intelligent learning signal, the specification information of the packaging box is uploaded and learned. If the station number of the industrial robot is recognized, it is sent to the processor. The processor generates a work instruction based on the station number and loads it to the corresponding industrial robot. The industrial robot clamps and transfers the packaging box that reaches the clamping point. In step S50, the working process of the parameter comparison module is as follows: Obtain two sets of measured side lengths of the packaging box on the conveyor belt based on the top surface image; Compare the two groups of measured side lengths of the packaging box with the clamping parameters of the packaging box in the storage module one by one; If the two sets of measured side lengths meet any set of clamping parameters, the station number of the industrial robot corresponding to the clamping parameters is obtained according to the mapping relationship; If the two sets of measured side lengths do not meet all clamping parameters, an intelligent learning signal is generated.

2. The robot intelligent programming method based on deep learning according to claim 1, characterized in that: The specification information is the box length, box width and box height of the packaging box.

3. The robot intelligent programming method based on deep learning according to claim 2, characterized in that: In step S20, the working process of the intelligent learning module is as follows: Obtain the specification information of the packaging box, and obtain the box length, box width and box height of the packaging box; Obtain three pairs of reference surfaces of the packaging box according to the box length, box width and box height; The three pairs of reference planes are divided into the first reference plane, the second reference plane and the third reference plane in order from the smallest to the largest area; Obtaining the maximum side length and the minimum side length of the first reference plane, and using the maximum side length and the minimum side length in the first reference plane as the first clamping parameter of the corresponding packaging box; Then, the maximum side length and the minimum side length in another pair of reference surfaces are obtained, and the maximum side length and the minimum side length in the pair of reference surfaces are used as the second clamping parameters of the corresponding packaging box; The maximum side length and the minimum side length in the last pair of reference planes are obtained, and the maximum side length and the minimum side length in the pair of reference planes are used as the third clamping parameter of the corresponding packaging box.

4. The robot intelligent programming method based on deep learning according to claim 3 is characterized in that: Robot intelligent programming methods also include: Step S60, the second acquisition module records the time nodes when the packaging boxes on the conveyor belt arrive at different measurement lines and the real-time positions of the packaging boxes on the conveyor belt, and sends the time nodes and real-time positions to the control analysis module via the server; Step S70, the control analysis module intelligently controls the industrial robot corresponding to the packaging box on the conveyor belt, and obtains a control packet of the industrial robot corresponding to the packaging box and sends it to the processor via the server; Step S80: the processor performs a clamping and transferring operation on the industrial robot corresponding to the packaging box according to the control package.

5. The robot intelligent programming method based on deep learning according to claim 4 is characterized in that: In step S70, the operation process of the manipulation analysis module is as follows: Acquire the top surface image of the packaging box on the conveyor belt; If the outer edge of the top surface image is perpendicular to the conveyor belt, there is no offset angle; If the outer edge of the top surface image is not perpendicular to the conveyor belt, the obtained offset angle is recorded as the adjustment offset angle of the industrial robot corresponding to the packaging box, and the adjustment direction of the industrial robot corresponding to the packaging box is established at the same time; Then, the time nodes when the packaging box on the conveyor belt arrives at different measuring lines are obtained. The time node when the packaging box arrives at the first measuring line is the first time node, and the time node when the packaging box arrives at the second measuring line is the second time node. The time node is the time when the geometric center in the top surface image arrives at the measuring line. The interval distance between the first measuring line and the second measuring line is known, the time difference is obtained by subtracting the first time node from the second time node, and the moving speed of the packaging box on the conveyor belt is obtained by dividing the interval distance by the time difference; Obtain the real-time position of the packaging box on the conveyor belt, and then obtain the gripping point of the corresponding industrial robot where the packaging box is to reach; wherein the real-time position is the real-time position of the geometric center in the top surface image; Calculate the distance between the real-time position of the packaging box and the clamping point and record it as the remaining transmission distance. Divide the remaining transmission distance by the moving speed to get the transmission time for the packaging box to reach the clamping point. The transmission time is used as the control reaction time of the industrial robot corresponding to the packaging box; The adjustment of the offset angle, adjustment of the direction and control reaction time are packaged and integrated into a control package for the industrial robot corresponding to the packaging box.

6. The robot intelligent programming method based on deep learning according to claim 5 is characterized in that: In step S80, the working process of the industrial robot is specifically as follows: The geometric center of the packaging box is obtained based on the top surface image. When the geometric center reaches the calibration line, the dynamic calibration point under the tracking calibration device moves along with the geometric center. The industrial robot corresponding to the packaging box is regulated according to the adjustment of the offset angle and the adjustment direction and completed within the control reaction time; When the packaging box reaches the corresponding clamping point of the industrial robot, the dynamic calibration point corresponds to the static calibration point up and down, and the industrial robot clamps and transfers the packaging box.

7. The robot intelligent programming method based on deep learning according to claim 6 is characterized in that: The specific process of obtaining the offset angle and adjustment direction is as follows: Connect two sets of diagonal lines of the top surface image, the intersection of the two sets of diagonal lines is the geometric center of the top surface image, and construct a first rectangular coordinate system with the geometric center of the top surface image. Then adjust the top surface image so that any outer edge of the top surface image is perpendicular to the conveyor belt. At this time, construct a second rectangular coordinate system with the geometric center of the top surface image. The angle between the Y' axis in the first rectangular coordinate system and the Y axis in the second rectangular coordinate system is the offset angle. The initial state of the industrial robot is the same as the second rectangular coordinate system. If the second rectangular coordinate system is rotated counterclockwise, the adjustment direction of the industrial robot is clockwise. If the second rectangular coordinate system is rotated clockwise, the adjustment direction of the industrial robot is counterclockwise.

8. A robot intelligent programming system based on deep learning, comprising a processor, a server connected to the processor, a first acquisition module, and a plurality of industrial robots, characterized in that: In combination with the robot intelligent programming method based on deep learning as described in claim 7, the server includes a second acquisition module, an intelligent learning module, a storage module, a parameter comparison module and a control analysis module, and the server is also connected to a management terminal; A management terminal is used to upload the specification information of the packaging box to the server, and the server sends the specification information to the intelligent learning module; An intelligent learning module is used to intelligently learn the clamping work of the industrial robot, obtain the clamping parameters of the packaging box and send them to the corresponding execution components through the server and processor; A processor, used for feeding back the mapping relationship between the clamping parameters and the industrial robot to the server, and the server sends the clamping parameters and the mapping relationship between the clamping parameters and the industrial robot to the storage module for storage; When working; The first acquisition module is used to acquire the top surface image of the packaging box on the conveyor belt, and the top surface image is sent to the parameter comparison module and the control analysis module via the processor and the server; The parameter comparison module is used to compare and identify the packaging boxes transmitted on the conveyor belt. If the identification generates an intelligent learning signal and feeds it back to the server, the management terminal uploads and learns the specification information of the packaging boxes on the conveyor belt. If the station number of the industrial robot is identified, it is sent to the processor. The processor generates a work instruction based on the station number and loads it to the corresponding industrial robot. The industrial robot clamps and transfers the packaging box that arrives at the clamping point. The second acquisition module is used to record the time nodes when the packaging boxes on the conveyor belt arrive at different measurement lines and the real-time position of the packaging boxes on the conveyor belt. The time nodes and real-time positions are sent to the control analysis module via the server; The control analysis module is used to intelligently control the industrial robot corresponding to the packaging box on the conveyor belt, and obtain the control package of the industrial robot corresponding to the packaging box and send it to the processor through the server; The processor is also used to perform clamping and transferring work on the packaging box corresponding to the industrial robot according to the control package.

Images