Robotic Device

Abstract

The present invention relates to a robot device for performing various processes other than laser processing. A robot device comprises: a robot; a motion control unit; a sensor that detects a value related to the position of a control target part of the robot; a low-speed position information acquisition unit that uses the detection value of the sensor when the robot is operated at a low speed to acquire the low-speed position information of the control target part; a calibration unit that performs calibration between the sensor and the robot using the low-speed position information and the command position; and a learning control unit that uses the detection value of the sensor when the robot is operated according to an action program, the low-speed position information or the command position, and the correction data obtained by calibration to learn a correction amount for reducing the deviation between the action described in the action program and the actual action of the control target part.

Description

Technical Field

[0001] The present invention relates to a robot device for performing various processing other than laser processing. Background Art

[0002] The trajectory accuracy of the robot is important in a processing system that performs various processing other than laser processing. However, due to the lack of rigidity of the reducer or robot arm, rebound, friction and other reasons for the deterioration of accuracy, the front end position of the robot will deviate from the target position or vibrate, thereby deteriorating the processing quality. As a corresponding method for this problem, the following method (for example, Japanese Patent Publication No. 2011-167817 and Japanese Patent Publication No. 2012-240142) has been proposed: a sensor is installed at the front end of the robot, the position deviation or vibration in the robot movement is measured by the sensor, and learning control is repeatedly performed to reduce the position deviation or vibration. Summary of the invention

[0003] As described above, when using a sensor to measure the motion trajectory of the robot's front end position and perform learning control, the user of the processing system needs to calibrate the coordinate system of the robot and the coordinate system of the sensor. Generally, calibration takes time, and if the calibration accuracy is not good, the improvement effect of the trajectory accuracy based on learning control will be weakened. A robot device that can automatically perform high-precision calibration between the coordinate system of the robot and the coordinate system of the sensor is desired.

[0004] One embodiment of the present disclosure is a robot device, which includes: a robot; a motion control unit that controls the motion of the robot; a sensor that detects a value related to the position of a control target part of the robot; a low-speed position information acquisition unit that uses the motion control unit to make the robot perform a motion described in the motion program at a speed lower than the speed specified by the motion program, and the detection value of the sensor related to the position of the control target part, and acquires the low-speed position information of the control target part; a calibration unit that uses the low-speed position information and the command position of the motion program to perform calibration between the sensor and the robot; and a learning control unit that uses the detection value of the sensor when the motion control unit causes the robot to perform a motion according to the motion program, the low-speed position information or the command position, and the correction data obtained by the calibration to learn a correction amount for reducing the deviation between the motion described in the motion program and the actual motion of the control target part. BRIEF DESCRIPTION OF THE DRAWINGS

[0005] The purpose, features and advantages of the present invention will be further clarified by the following description of the embodiments associated with the accompanying drawings. In the same drawings:

[0006] Figure 1 It is a diagram showing the overall configuration of a robot device according to one embodiment.

[0007] Figure 2 It is a diagram showing the structure of a robot.

[0008] Figure 3 This is a functional block diagram of a robot control device.

[0009] Figure 4 It is a flowchart which shows the calibration and learning control.

[0010] Figure 5 This is a diagram showing an example of a photographic image of a processing site.

[0011] Figure 6 It means from Figure 5 Figure 3. Edge data obtained from an image.

[0012] Figure 7 This is a flowchart showing an example of calibration processing when the command trajectory is a circle.

[0013] Figure 8 This is a diagram showing a state where the calibrated motion trajectory matches the command trajectory.

[0014] Fig. 9 This is a flowchart showing an example of calibration processing when the command trajectory is a square.

[0015] Fig.10 It is a diagram showing the motion trajectory during low-speed motion and the motion trajectory during high-speed motion.

[0016] Fig.11 This is a graph that plots the error at each angle between the motion trajectory during quick motion and the motion trajectory during high-speed motion.

[0017] Fig.12 It is used to illustrate Fig.11 The angle of the diagram is defined in the diagram. DETAILED DESCRIPTION

[0018] Hereinafter, the embodiments of the present disclosure will be described with reference to the accompanying drawings. In all the drawings, corresponding structural elements are marked with common reference marks. For ease of understanding, the scales of these drawings are appropriately changed. In addition, the method shown in the drawings is an example for implementing the present invention, and the present invention is not limited to the method shown in the drawings.

[0019] Figure 1 FIG. 1 is a diagram showing the overall structure of a robot device 10 according to one embodiment. Figure 1As shown, the robot device 10 includes a robot 1, a robot control device 2 for controlling the movement of the robot 1, and a camera 3 fixed to the installation surface of the robot 1. A processing tool 5 is installed at the front end of the arm of the robot 1. The robot device 10 can be used for various processing such as laser processing (welding, cutting, etc.) and cutting processing based on water jet. The processing tool 5 is a tool for irradiating a laser to a workpiece W and performing laser processing.

[0020] The robot control device 2 may have a configuration as a general computer having a CPU, ROM, RAM, a storage device, etc. The robot control device 2 controls the operation of the robot 1 according to an operation program preloaded into the robot control device 2. Figure 2 As shown, the robot 1 is, for example, a vertical multi-joint robot having first to sixth joints 71 to 76. The structure of the robot 1 according to the present embodiment is not limited to this example. As a coordinate system for defining the position of the robot, a world coordinate system 91 sets the base of the robot 1 as the origin.

[0021] The robot device 10 involved in this embodiment is configured to measure the position of the robot in motion (the position of the control object part) through a sensor, thereby obtaining the position deviation from the target position of the motion trajectory of the robot 1, and performing learning control for reducing the position deviation. Here, the control object part of the robot is, for example, the TCP (tool / center / point) of the robot. Through this structure, the robot device 10 can prevent the front end position of the robot from deviating from the target position (command position) or vibrating due to the deterioration of precision such as insufficient rigidity of the reducer and the robot arm, rebound, friction, etc., thereby affecting the processing quality. As a sensor for measuring the position of the robot 1 in motion, various sensors such as acceleration sensors, force sensors, visual sensors, laser trackers, etc. can be used, but in this embodiment, it is configured to capture and obtain the motion trajectory of the robot 1 through a camera 3 as a visual sensor.

[0022] When using sensors to measure position deviations and perform learning control, it is necessary to calibrate the coordinate system of the robot 1 and the sensor coordinate system. As described below, the robot 10 automatically calibrates the world coordinate system 91 and the coordinate system of the camera 3 (hereinafter also referred to as the camera coordinate system).

[0023] Figure 3 2 is a functional block diagram showing the robot control device 2 . Figure 2Each functional block diagram shown can be implemented as dedicated hardware or by executing software and cooperating with hardware by the CPU of the robot control device 2. In the robot control device 2, the motion control unit 21 controls the motion (position and posture) of the robot 1 according to the motion program 25. More specifically, the motion control unit 21 controls the position and speed of the drive shaft of each joint of the robot 1 according to the motion program 25. The learning control unit 22 compares the command position of the robot 1 based on the motion program 25 with the motion trajectory of the robot 1 captured by the camera 3, thereby obtaining the position deviation of the actual motion trajectory of the robot 1 relative to the command position, and obtains the correction amount to reduce the position deviation through learning control. The low-speed position information acquisition unit 24 uses the detection value of the sensor related to the position of the robot 1 (the image captured by the camera 3 in this embodiment) when the motion control unit 21 causes the robot to perform the motion described in the motion program 25 at a speed lower than the speed specified by the motion program 25, and obtains the low-speed position information of the robot 1 (the motion trajectory of the robot 1 in this embodiment). The calibration unit 23 performs calibration between the world coordinate system 91 and the camera coordinate system.

[0024] Figure 4 is a flowchart showing the process of reducing position deviation by learning control. Figure 4 Processing. First, the robot control device 2 causes the robot 1 to perform the target action at a low speed (step S1). Here, the low-speed action refers to a speed lower than the action speed specified by the action program, and is a speed so low that the influence of the accuracy deterioration factors such as vibration on the robot 1 can be ignored. The speed during the low-speed action can be set to a speed sufficiently lower than the speed specified by the action program (for example, a speed of 1 / 10). The robot control device 2 uses the camera 3 to shoot the action trajectory of the workpiece W processed by the robot 1 at a low speed and obtains the point column data of the processed part (step S2). In this way, by causing the robot 1 to move at a low speed, a motion trajectory with no error (extremely small error) relative to the command trajectory described in the action program 25 can be obtained. In addition, the processing of steps S1-S2 is performed under the control of the new position acquisition unit 24 at low speed.

[0025] Next, in step S3, the coordinate system of the robot 1 (world coordinate system 91) and the coordinate system of the camera 3 (camera coordinate system) are calibrated using the command value data during the low-speed operation and the point array data obtained by the camera 3. The calibration is performed under the control of the calibration unit 23. Assuming that the command recorded in the action program is a command for laser processing a circle with a diameter of 150 mm, an example of the calibration process is described below. After the command for laser processing a circle with a diameter of 150 mm is executed by low-speed operation in step S1, the image obtained by photographing the workpiece W by the camera 3 in step S2 is Figure 5 The image shown. Figure 5 The portion indicated by a blank in the image is a portion processed by laser (hereinafter referred to as a processed portion 101). Due to the installation error of the camera 3 and the like, the image of the processed portion 101 photographed by the camera 3 may not be circular.

[0026] To perform the calibration process, first, Figure 5 The edge data can be extracted by the low-speed position information acquisition unit 24. As the edge extraction result, Figure 6 The edge data (point sequence data) 102 shown. In addition, Figure 6 In , the vertical axis represents the horizontal pixel position and the horizontal axis represents the vertical pixel position. Figure 6 The camera 3 is calibrated based on the relationship between the edge data 102 and the command trajectory during low-speed operation. Figure 7 An example of calibration processing when the command trajectory is a circle is shown. First, the geometric center of the edge data 102 is aligned with the center of the circle of the command trajectory, thereby focusing (step S101). Next, respective magnification factors are applied to the edge data 102 in the horizontal direction and the vertical direction, and the edge data 102 is aligned with the circle of the command trajectory (step S102). The offset amount when the edge data 102 is offset for focusing, and the magnification factors in the horizontal direction and the vertical direction for aligning the edge data 102 with the circle based on the command trajectory constitute calibration data (i.e., correction data for converting the camera coordinate system into the world coordinate system).

[0027] Figure 8 The motion trajectory 102A shown in FIG. 1 is obtained by using the calibration data obtained as described above and converting the motion trajectory during low-speed motion, that is, the edge data 102, into the coordinate system of the robot 1. Figure 8 As shown, the calibrated motion trajectory 102A is consistent with the command trajectory 103 of the robot 1.

[0028] Figure 7 This is an example of calibration processing when the command trajectory is a circle, but here Fig. 9The following is an example of calibration processing when the command trajectory is a quadrilateral. At this time, first, the edge data obtained by moving the robot 1 at a low speed and photographing the processed motion trajectory is focused on the quadrilateral of the command trajectory (step S201). Next, the edge data is rotated (step S202), and the corresponding sides of the quadrilateral are parallelized (step S203). Next, the aspect ratio of the edge data is adjusted and the quadrilateral of the edge data is consistent with the quadrilateral of the command trajectory (step S204). The offset amount, rotation amount, and aspect ratio adjustment amount used for focusing obtained by the above processing constitute the calibration data. In addition, when the command trajectory is other shapes, it can also be achieved by comparing with Figure 7 or Fig. 9 The calibration data is obtained by performing the same process as shown.

[0029] The above calibration process enables automatic and highly accurate calibration of the camera 3 .

[0030] return Figure 4 As described above, learning control of repeated learning is then performed through loop processing L1. Learning control is performed under the control of the learning control unit 22. If learning control is started, processing is performed at the action speed specified by the action program (hereinafter also referred to as high-speed action) (step S4). At this time, when the correction amount is calculated by the internal processing of the previous learning control (step S6), processing is performed on the basis of applying the correction amount to the action instruction of the action program and correcting it. Next, the processed trajectory is photographed by the camera 3 and the edge data (point column data) of the processed part is obtained (step S5).

[0031] Next, a learning control is performed to obtain a correction amount for reducing the position deviation based on the motion trajectory during high-speed motion obtained in step S5 (step S6). As a method of learning control for obtaining a correction amount for reducing the position deviation between the actual motion trajectory and the target trajectory, various methods known in the art can be used. Here, the process of obtaining the correction amount is performed as follows. As a target trajectory that serves as a reference for obtaining the position deviation, a motion trajectory during low-speed motion or a command trajectory for the robot 1 can be used, but the following description will be given using the case of using the motion trajectory during low-speed motion as an example.

[0032] Fig.10 The motion trajectory 102 of a low-speed motion and the motion trajectory 111 of a high-speed motion obtained by shooting with the camera 3 are shown. Fig.10 In FIG. 1 , the vertical axis and the horizontal axis represent the pixel position in the vertical direction and the pixel position in the horizontal direction, respectively. Based on the calibration data used to calibrate the motion trajectory 102 during the low-speed motion and the motion trajectory 111 during the high-speed motion, the error of each angle of the motion trajectory 102 and the motion trajectory 111 is plotted. Fig.11.here, Fig.11 The angle of the horizontal axis is defined as Fig.12 In this way, a straight line extending downward in parallel with the vertical axis from the geometric center P of the motion trajectory 102 is used as a reference, and the direction rotating counterclockwise in the figure around the center P is set as a positive angle θ.

[0033] In the calculation of the correction amount for learning control, the elimination Fig.11 The correction amount of the error shown is stored in the memory in the learning control unit 22. The correction amount is used to correct the command position of the next processing action (step S4). Figure 4 This learning control can be repeated until the position deviation is zero or converges to a negligible level. Figure 4 processing.

[0034] Through the above processing, the position deviation can be obtained using the output (photographic image) of the calibrated camera 5 and correct learning control can be performed. In particular, according to the above processing, the calibration process is automatically performed using the command value data during low-speed operation and the point sequence data obtained by the camera 3. Therefore, the concern about poor calibration accuracy due to user's operational errors or lack of experience can be eliminated, thereby obtaining stable learning control performance.

[0035] That is, according to this embodiment, high-precision calibration between the sensor coordinate system (e.g., the camera coordinate system) and the world coordinate system can be automatically performed. In addition, accurate learning control can be performed by using the position deviation obtained by the sensor output (e.g., the image of the camera) that has been calibrated with high precision.

[0036] As mentioned above, although embodiment of this disclosure is described, it will be understood by those skilled in the art that various corrections and changes can be made without departing from the disclosure scope of the claims described later.

[0037] The above-mentioned embodiment is configured to measure the processing site and obtain the robot's motion trajectory by using the camera 3 as a visual sensor, but the measurement site for obtaining the robot's motion trajectory may also be a part on the robot 1. For example, the front end of the arm of the robot 1 may be used as the measurement site. In addition, it may also be configured to install an acceleration sensor, a force sensor, or a laser tracker on the front end of the robot's arm when these sensors are used as sensors for obtaining the robot's motion trajectory. For example, by installing an acceleration sensor on the front end of the arm of the robot 1, the motion trajectory of the front end of the arm of the robot 1 during its motion can be obtained. As an example, Figure 2The state where the sensor 81 is attached to the fore-end portion of the arm (wrist) is shown. At this time, the coordinate system 93 fixed to the fore-end portion of the arm is set as the coordinate system of the sensor 81.

[0038] It is also possible to install a force sensor between the front end of the arm of the robot 1 and the processing tool, and detect the load accompanying the movement of the processing tool 5 through the force sensor. In this case, a three-axis force sensor is used as the force sensor to detect the three components of XYZ force when the robot 1 moves. According to the motion equation F=ma and the mass m of the processing tool installed in front of the force sensor, the detected force is converted into the acceleration of the processing tool. Then, based on the converted acceleration and the acceleration obtained by performing a second-order differential on the command position of the robot, data representing the relationship between the coordinate system of the robot and the coordinate system of the sensor can be obtained, and the force sensor can be calibrated by using this data.

[0039] In order to solve the problem of the present disclosure, the following various modes and their effects can be provided. In addition, the numbers in the parentheses of the description of the following modes correspond to the reference numerals of the drawings of the present disclosure.

[0040] A first mode of the present disclosure is a robot device (10), comprising: a robot (1); a motion control unit (21) for controlling the motion of the robot (1); a sensor (3) for detecting a value related to the position of a control target part of the robot (1); and a low-speed position information acquisition unit (24) for acquiring the low-speed position information of the control target part by using the motion control unit (21) to cause the robot (1) to perform a motion described in the motion program at a speed lower than the speed specified by the motion program, based on the detection value of the sensor related to the position of the control target part. a calibration unit (23) which uses the low-speed position information and the command position of the action program to perform calibration between the sensor (3) and the robot (1); and a learning control unit (22) which uses the detection value of the sensor when the action control unit (21) causes the robot (1) to act according to the action program, the low-speed position information or the command position, and the correction data obtained by the calibration to learn a correction amount for reducing the deviation between the action described in the action program and the actual action of the control object part.

[0041] According to the first aspect, high-precision calibration between the sensor coordinate system (eg, camera coordinate system) and the world coordinate system can be automatically performed. In addition, accurate learning control can be performed by using position deviations obtained from high-precision calibrated sensor outputs (eg, camera images).

[0042] In addition, a second aspect of the present disclosure is the robot device (10) of the above aspect, wherein the correction data obtained by the calibration includes data for converting the coordinate system of the sensor into the coordinate system of the robot.

[0043] In addition, a third aspect of the present disclosure is the robot device (10) of the first aspect or the second aspect, wherein the sensor includes a camera (3) for photographing a trajectory of the actual action.

[0044] In addition, a fourth aspect of the present disclosure is the robot device (10) of the first aspect or the second aspect, wherein the sensor is mounted on a wrist of the robot and detects a value indicating a position of the wrist.

Claims

1. A robot device, characterized in that: The robot device has: robot; A motion control unit, which controls the motion of the robot; A sensor that detects a value related to a position of a control target portion of the robot; a low-speed position information acquisition unit for acquiring the low-speed position information of the control target part by using the detection value of the sensor related to the position of the control target part when the motion control unit causes the robot to perform the motion described in the motion program at a speed lower than the speed specified by the motion program; a calibration unit that performs calibration between the sensor and the robot using the low-speed position information and the command position of the motion program; as well as a learning control unit that uses the detection value of the sensor when the motion control unit causes the robot to move according to the motion program, the low-speed position information or the command position, and the correction data obtained by the calibration to learn a correction amount for reducing a deviation between the motion described in the motion program and the actual motion of the control target part, The correction data obtained by the calibration includes data for converting the coordinate system of the sensor into the coordinate system of the robot. The sensor includes a force sensor for detecting a load accompanying the movement of a processing tool mounted on the robot, and converting the detected load into an acceleration of the processing tool. The calibration unit acquires data indicating a relationship between a coordinate system of the robot and a coordinate system of the sensor based on the acceleration and a command position of the operation program, and performs calibration between the sensor and the robot using the data.

2. The robot device according to claim 1, characterized in that: The sensor is mounted on a wrist of the robot and detects a value indicating a position of the wrist.

Images