JP2018001331A - Correction device, correction method, and correction program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a correction device, a correction method, and a correction program which can correct positional deviation with simple processing.SOLUTION: A correction device includes: a sensor which detects a prescribed position of an arm robot in a prescribed posture or a prescribed position of an article held by an arm robot in the prescribed position; an estimation section which estimates positional deviation amount in the prescribed posture using time series data of the position detected by the sensor when the arm robot takes the prescribed posture; and a correction section which corrects positional deviation of the arm robot on the basis of the positional deviation amount estimated by the estimation section.SELECTED DRAWING: Figure 1

Description

  This case relates to a correction apparatus, a correction method, and a correction program.

  Articulated robots are used as industrial robots. Industrial robots are used after the tip position of an article such as a tool or part is determined with high accuracy. However, the link portion may undergo thermal expansion during operation due to the influence of its own heat generation, environmental heat, and the like. In this case, a positional deviation occurs at the positioned article tip. Therefore, a technique for correcting the positional deviation of the article tip is disclosed.

  For example, a technique for sensing a temperature and calculating a correction amount is disclosed (for example, see Patent Document 1). Further, a technique for calibrating each time a positional shift occurs is disclosed (for example, see Patent Document 2).

JP-A 63-144980 JP-A-5-261686

  However, it is expensive to provide a temperature sensor for an industrial robot. In addition, it is troublesome to try to calibrate each time a positional deviation occurs.

  In one aspect, an object of the present invention is to provide a correction device, a correction method, and a correction program that can correct a positional shift with a simple process.

  In one aspect, the correction device includes a sensor that detects a predetermined position of the arm robot in a predetermined posture, or a predetermined position of an article that the arm robot holds in the predetermined posture, and the arm robot takes the predetermined posture. Using the time-series data of the positions detected by the sensors, an estimation unit that estimates the amount of positional deviation in the predetermined posture, and the positional deviation of the arm robot is corrected based on the amount of positional deviation estimated by the estimation unit. And a correction unit.

  The positional deviation can be corrected with a simple process.

(A) is a figure which illustrates the whole structure of a robot system, (b) is a functional block diagram of a calculating part, (c) is a figure which illustrates the flow of a process in a robot system. It is a figure which illustrates an arm robot. It is a figure which illustrates the time series data stored in the calculating part. (A)-(c) is a figure which illustrates operation | movement of an arm robot. It is a flowchart which illustrates the process which a robot system performs. (A) is a block diagram for demonstrating the hardware constitutions of a calculating part, (b) is a figure which illustrates about the working system concerning a modification.

  Hereinafter, embodiments will be described with reference to the drawings.

  FIG. 1A is a diagram illustrating an overall configuration of the robot system 100. As illustrated in FIG. 1A, the robot system 100 includes an arm robot 10, a control unit 20, a sensor 30, a calculation unit 40, and the like. FIG. 1B is a functional block diagram of the calculation unit 40. As illustrated in FIG. 1B, the calculation unit 40 functions as a storage unit 41, an estimation unit 42, and a correction value calculation unit 43.

  The arm robot 10 is a robot having at least one link. In this embodiment, as an example, the arm robot 10 is a robot in which a plurality of links are connected by one or more joints. The control unit 20 controls the posture and operation of the arm robot 10. The sensor 30 is a sensor that detects a predetermined position of the arm robot 10 in a predetermined posture or a predetermined position of the article 50 held by the arm robot 10 in the predetermined posture, for example, a camera. The article 50 is a tool, a part, or the like. In the present embodiment, as an example, the sensor 30 detects the tip position of the article 50 that the arm robot 10 grips at the tip. The calculation unit 40 performs a calculation for correcting the positional deviation of the arm robot.

  FIG. 1C is a diagram illustrating a process flow in the robot system 100. As illustrated in FIG. 1C, the control unit 20 controls the operation of the arm robot 10 by transmitting control data to the arm robot 10. The sensor 30 detects the tip position of the article 50 held by the arm robot 10 in a predetermined posture as a sensor value. For example, a Time of Flight method, a stereo vision method, or the like can be used. The detected sensor value is transmitted to the calculation unit 40.

  The storage unit 41 stores the received sensor value. The estimation unit 42 estimates the positional deviation amount of the arm robot 10 based on the stored sensor value. The correction value calculation unit 43 calculates a correction value for suppressing the positional deviation based on the estimated positional deviation amount, and transmits the correction value to the control unit 20. The control unit 20 transmits control data for correcting the positional deviation of the arm robot 10 to the arm robot 10 based on the received correction value.

  FIG. 2 is a diagram illustrating the arm robot 10. As illustrated in FIG. 2, the arm robot 10 according to the present embodiment is an articulated robot having a structure in which a plurality of links 11 are connected by joints 12. The arm robot 10 is fixed to a fixed portion (base) such as a ceiling, a floor, or a wall. The link is a non-deformable part that constitutes the arm robot 10. The link length of the link connected to the base is the length from the base to the joint. The link length of the link connected between the joints is the length from the joint to the joint. The link length of the tip link is the length from the joint to the tip. When the tip link holds the article 50, the link length of the tip link may be from a joint to a predetermined position (for example, the tip) of the article 50.

As illustrated in FIG. 2, the tip position p _{E, 0} of the arm robot 10 can be represented by the sum of vectors p _{i + 1, i} having the base as the origin and the link length and link direction. In the present embodiment, the tip position _{PE, 0} is the tip position of the article 50. i is the number of the link allocated from the base side. That is, the tip position p _{E, 0} can be expressed by the following formula (1).

When a temperature change occurs in the arm robot 10 in accordance with the heat generated by the arm robot 10 itself, environmental heat, or the like, the link length changes. For example, when the temperature change value is ΔT and the linear expansion coefficient of the i-th link is α _i , the i-th link vector p _{i + 1, i} changes as shown in the following equation (2).

Thus, when the temperature change occurs in the arm robot 10, the tip position is displaced. In this case, a problem may occur in the work performed by the arm robot 10. Therefore, it is preferable to correct the position shift of the arm robot 10. According to the above formula (2), if the expansion rate Δa _i (= α _i ΔT _i ) can be estimated, the positional deviation amount can be estimated. Therefore, in this embodiment, by estimating the elongation rate .DELTA.a _i, to correct the positional displacement of the arm robot 10.

  Specifically, the sensor 30 detects the tip position of the article 50 in a predetermined posture before fitting the article 50 to the workpiece 60. The sensor 30 detects the tip position of the article 50 every time the arm robot 10 takes the posture. The calculation unit 40 stores time-series data of the detected tip position of the article 50. FIG. 3 is a diagram illustrating time-series data stored in the calculation unit 40. As illustrated in FIG. 3, in the time series data, the tip position of the article 50 is associated with time (t_1, t_2,...).

Next, the computing unit 40 estimates the expansion rate Δa _i from the stored time series data and estimates the link length of each link. Next, the calculation unit 40 calculates a correction value for the tip position of the article 50 from the estimated link length. Next, the control unit 20 corrects the positional deviation by correcting the operation and posture of the arm robot 10 based on the calculated correction value. Details will be described below.

  FIG. 4A to FIG. 4C are diagrams illustrating the operation of the arm robot 10. As illustrated in FIG. 4A, the gripping portion provided at the tip of the arm robot 10 grips the article 50 (Operation 1). Next, as illustrated in FIG. 4B, the arm robot 10 moves the article 50 in front of the sensor 30 and takes a predetermined posture (operation 2). The sensor 30 detects the tip position of the article 50 after the arm robot stops at the predetermined posture. Next, as illustrated in FIG. 4C, the arm robot 10 engages the article 50 with the workpiece 60 (operation 3). By repeating FIG. 4A to FIG. 4C, time series data of the tip position of the article 50 can be acquired.

Here, the control unit 20 controls the arm robot 10 so that the posture of the arm robot 10 becomes the predetermined posture when the tip position p _{E, 0} of the article 50 is detected by the sensor 30. That is, when detecting the tip position _{p E, 0} of the article 50 by the sensor 30, tip position _{p E, 0} represented by the above formula (1) should not change. However, when a temperature change occurs in the arm robot 10, the tip position pE _{, 0} is shifted. The positional deviation amount Δp _{E, 0 in} this case can be expressed by the following formula (3).

The expansion rate Δa _i is a function of the time t, the heat generation time t _i in the i-th link, the coefficient A _i , and the heat transfer coefficient τ _i . That is, the elongation rate Δa _i is a function of a parameter related to heat generation. The elongation rate Δa _i can be expressed as the following formula (4). Since time t is an explanatory variable, the values to be estimated are the heat generation time t _i , the coefficient A _i , and the heat transfer coefficient τ _i . Note that since the expansion rate Δa _i does not change until heat is generated, Δa _i is zero at t <t _i .

The residual sum of squares E with respect to the time series data of the positional deviation amount Δp _{E, 0} can be expressed as the following equation (5). The heat generation time t _i , the coefficient A _i , and the heat transfer coefficient τ _i can be estimated by the nonlinear least square method for the residual sum of squares E. “N” represents the number of links. “N” is the number of time series data.

When the above nonlinear least squares method is executed in a naive manner, in order to determine the heat generation time t _i , coefficient A _i , and heat transfer coefficient τ _i (i = 0, 1, 2,..., N), A sufficient number of time series data is required. In some cases, it is necessary to correct misalignment before a sufficient number of time-series data is obtained. Therefore, the heat generation time t _i , the coefficient A _i , and the heat transfer coefficient τ _i are estimated according to the number of time-series data. The method may be divided into cases.

For example, when the number of time-series data is 1 to 2, all i may have A _i in common and τ _i may be infinite. Specifically, Δa _i = A may be set. In this case, the estimated value is limited only to the coefficient A. That is, one parameter needs to be estimated. Thereby, an estimation error can be suppressed even if a sufficient number of time-series data is not obtained.

Next, when the number of time-series data is 3 to 5, the heat generation time t _i , the coefficient A _i , and the heat transfer coefficient τ _i may be common to all _i . Although there are three estimated values, since it is assumed that each estimated value does not fluctuate, an estimation error can be suppressed.

Next, when the number of time-series data is 6 to 8, the heat generation time t _i , the coefficient A _i , and the heat transfer coefficient τ _i may be common for each of the even and odd of _i . Next, when the number of time-series data is 9 to 11, i may be divided into three modes, and the heat generation time t _i , the coefficient A _i , and the heat transfer coefficient τ _i may be shared in each mode. . Hereinafter, similarly, if the number of time-series data is 12 to 14, it may be divided into 4 and 3n to 3n + 2 may be divided into n. When the number of time-series data is 3n + 3 or more, the heat generation time t _i , the coefficient A _i , and the heat transfer coefficient τ _i (i = 0, 1, 2,..., N) may be estimated.

  Thus, the smaller the number of time-series data, the fewer parameters that need to be estimated, and the larger the number of time-series data, the more parameters that need to be estimated. In this case, parameter estimation errors can be suppressed even if sufficient time-series data is not stored.

Next, the time-series data is obtained every time the tip position pE _{, 0} of the article 50 is confirmed by the sensor 30, but the positional deviation needs to be corrected only when the positional deviation occurs. Therefore, it is preferable to detect misalignment. For example, as a method for defining the positional deviation, the initial positional data is used as a reference, and the subsequent time-series data whose positional deviation from the reference is equal to or greater than a threshold (for example, a predetermined number of times greater than the error of the sensor 30). It may be defined as data. As the predetermined multiple, for example, double can be used. Further, the positional deviation may be corrected when the positional deviation reaches a predetermined magnitude larger than the predetermined number of times. The predetermined size depends on the size of the arm robot 10 and the required accuracy, but is, for example, 0.1 mm. The above estimation may be executed according to the number of time-series data when a positional deviation occurs. After the correction, all the time series data may be deleted, and the initial position data may be set again as a reference, and the subsequent steps may be repeated.

  Next, when a monocular camera is used as the sensor 30, only position data projected on a two-dimensional plane can be obtained. On the other hand, if the direction of the misalignment is known in advance, the three-dimensional misalignment data can be restored from the two-dimensional misalignment data, and even the one-dimensional misalignment data can be restored. Is possible. Therefore, before the estimation calculation is performed, the misalignment data may be restored to three dimensions using the misalignment direction obtained from the previous estimation.

For example, if the time-series data number of 1 to 6 when misalignment occurs, the time series data number 1 to 5, because .DELTA.a _i are common to all i, position shift direction, the following formula ( It can be defined as parallel to p obtained in 6). “P” is a vector extending from the base of the arm robot 10 to the tip. Since the immediately preceding estimated value Δa _i is not necessary, even if the number of time-series data is 1, it can be applied.

If the time-series data number of 1 to 7 when misalignment occurs, be defined as parallel to Delta] p _{E, 0} obtained by the following formula is calculated from .DELTA.a _i obtained in the last estimation (7) it can.

FIG. 5 is a flowchart illustrating a process executed by the robot system 100. As illustrated in FIG. 5, the control unit 20 controls the arm robot 10 so as to perform the operation 1 in FIG. 4A (step S1). Next, the control unit 20 controls the arm robot 10 to perform the operation 2 in FIG. 4B (step S2). Next, the calculating part 40 acquires the front-end | tip position pE _{, 0} according to said Formula (1) using the detection result of the sensor 30 (step S3).

Next, the computing unit 40 determines whether or not the acquisition of the tip position pE _{, 0} is the first time (step S4). When it is determined as “Yes” in Step S4, the calculation unit 40 sets the tip position pE _{, 0} as a reference value (Step S5). Next, the control unit 20 controls the arm robot 10 to perform the operation 3 in FIG. 4C (step S6). Next, the control unit 20 determines whether or not the predetermined number has looped (step S7). If “No” is determined in step S7, the process is executed again from step S1. If it is determined as “Yes” in step S <b> 7, the execution of the flowchart ends.

When it determines with "No" at step S4, the calculating part 40 determines whether the sensor 30 is a monocular camera (step S8). When it is determined as “Yes” in Step S8, the calculation unit 40 performs three-dimensionalization according to the acquired time-series data number (Step S9). After execution of step S9 or when it is determined “No” in step S8, the calculation unit 40 calculates the distance Δp _{E, 0} from the reference value of the tip position acquired in step S3 (step S10).

Next, the calculation unit 40 determines whether or not the distance Δp _{E, 0} calculated in step S10 is larger than a predetermined number of times (for example, twice) the measurement error of the sensor 30 (step S11). If it is determined “No” in step S11, step S6 is executed. If “Yes” is determined in step S11, it is determined whether or not the distance Δp _{E, 0} calculated in step S10 is greater than or equal to a predetermined size larger than the predetermined number of times (step S12). When it is determined as “No” in Step S12, the calculation unit 40 stores the distance Δp _{E, 0} calculated in Step S10 (Step S13). Thereafter, step S6 is executed. When it is determined as “Yes” in step S <b> 12, the calculation unit 40 calculates a correction value according to the number of stored time-series data. The control unit 20 corrects the positional deviation of the arm robot 10 according to the correction value (step S14). Note that after the execution of step S14, the calculation unit 40 performs initialization by deleting the stored time-series data. Thereafter, the execution of the flowchart ends.

  According to the present embodiment, in order to estimate the amount of positional deviation in a predetermined posture using time-series data of a predetermined position of an arm robot detected by a sensor or a predetermined position of a grasped article when a predetermined posture is taken. The positional deviation can be corrected by a simple process. Since it is not necessary to use a temperature sensor, cost can be reduced. Further, since it is not necessary to perform calibration every time a positional deviation occurs, it is not time-consuming. Usually, the position of the article is confirmed by a camera or the like in order to confirm the position of the gripped tool. Since the sensor value in this case can be used for the estimation of the positional deviation amount, it is not necessary to newly acquire a sensor value for application of the present embodiment.

  Further, the number of parameters that need to be estimated may be changed according to the number of time-series data. For example, the number of parameters that need to be estimated may be reduced as the number of time series data is smaller, and the number of parameters that need to be estimated may be increased as the number of time series data is larger. In this case, parameter estimation errors can be suppressed even if sufficient time-series data is not stored.

In addition, the capacity can be reduced by not using time series data in which the deviation between the tip position pE _{, 0} detected by the sensor 30 and the reference position is less than the threshold value. In addition, the amount of calculation can be reduced. When the deviation between the tip position pE _{, 0} detected by the sensor 30 and the reference position is equal to or greater than the threshold value, the amount of calculation can be reduced by correcting the positional deviation.

  In the above example, the sensor 30 detects the tip position of the article 50, but is not limited thereto. For example, another position of the article 50 may be detected, or any position of the link of the arm robot 10 may be detected. However, since the amount of displacement of the tip position of the arm robot 10 and the tip position of the article 50 increases, it is preferable to detect the tip position of the arm robot 10 and the tip position of the article 50. Moreover, in the state where each link is extended in a straight line, the expansion and contraction of each link is added without being canceled out, and the positional deviation becomes the largest. Therefore, it is more preferable to detect the tip position when the arm robot 10 extends each link in a straight line and the tip position of the article 50. From the viewpoint of length measurement theory (Abbe's principle), it is preferable to detect the tip position and the tip position of the article 50 in a state where each link is extended in a straight line.

  In the above example, the sensor 30 functions as an example of a sensor that detects a predetermined position of the arm robot in a predetermined posture or a predetermined position of an article held by the arm robot in a predetermined posture. The estimation unit 42 functions as an example of an estimation unit that estimates a positional deviation amount in a predetermined posture using time-series data of positions detected by a sensor when the arm robot takes a predetermined posture. The correction value calculation unit 43 and the control unit 20 function as an example of a correction unit that corrects the positional deviation of the arm robot based on the positional deviation amount estimated by the estimation unit.

  FIG. 6A is a block diagram for explaining the hardware configuration of the arithmetic unit 40. Referring to FIG. 6A, the calculation unit 40 includes a CPU 101, a RAM 102, a storage device 103, a display device 104, and the like. A CPU (Central Processing Unit) 101 is a central processing unit.

  The CPU 101 includes one or more cores. A RAM (Random Access Memory) 102 is a volatile memory that temporarily stores programs executed by the CPU 101, data processed by the CPU 101, and the like. The storage device 103 is a nonvolatile storage device. As the storage device 103, for example, a ROM (Read Only Memory), a solid state drive (SSD) such as a flash memory, a hard disk driven by a hard disk drive, or the like can be used. The storage device 103 stores a correction program. The display device 104 is a liquid crystal display, an electroluminescence panel, or the like, and displays a determination result. In this embodiment, each unit of the calculation unit 40 is realized by executing a program, but hardware such as a dedicated circuit may be used.

(Modification 1)
FIG. 6B is a diagram illustrating a work system according to a modification. In each of the above examples, the calculation unit 40 acquires time series data from the sensor 30. On the other hand, the server 202 having the function of the calculation unit 40 may acquire data from the sensor 30 through the telecommunication line 201 such as the Internet. In this case, the calculation unit 40 transmits the calculated correction value to the control unit 20 via the telecommunication line 201.

  Although the embodiments of the present invention have been described in detail above, the present invention is not limited to such specific embodiments, and various modifications and changes can be made within the scope of the gist of the present invention described in the claims. It can be changed.

DESCRIPTION OF SYMBOLS 10 Arm robot 11 Link 12 Joint 20 Control part 30 Sensor 40 Calculation part 50 Article 60 Work 100 Robot system

Claims (8)

A sensor for detecting a predetermined position of the arm robot in a predetermined posture or a predetermined position of an article held by the arm robot in the predetermined posture;
An estimation unit that estimates a positional deviation amount in the predetermined posture using time-series data of a position detected by the sensor when the arm robot takes the predetermined posture;
A correction device, comprising: a correction unit that corrects a position shift of the arm robot based on a position shift amount estimated by the estimation unit.   The correction apparatus according to claim 1, wherein the estimation unit estimates the positional deviation amount by estimating a parameter related to heat generation and estimating an extension rate of the arm robot.   The correction device according to claim 1, wherein the estimation unit changes the number of parameters used for the estimation of the positional deviation amount according to the number of the time series data.   4. The estimation unit according to claim 1, wherein the estimation unit does not use time-series data in which a deviation between a reference position and a position detected by the sensor is less than a first threshold value when estimating the positional deviation amount. The correction apparatus as described in any one.   The correction unit corrects a position shift of the arm robot when a shift between a reference position and a position detected by the sensor becomes a second threshold value or more. The correction device according to one item.   The correction device according to claim 1, wherein the arm robot includes a plurality of links connected by one or more joints. Using a sensor, a predetermined position of the arm robot in a predetermined posture, or a predetermined position of an article held by the arm robot in the predetermined posture,
The estimation unit estimates the amount of positional deviation in the predetermined posture using time series data of the position detected by the sensor when the arm robot takes the predetermined posture,
A correction method, wherein the correction unit corrects the positional deviation of the arm robot based on the positional deviation amount estimated by the estimation unit. On the computer,
Processing for estimating the amount of displacement in the predetermined posture using time-series data of the predetermined position of the arm robot detected by a sensor when the arm robot takes a predetermined posture or the predetermined position of an article held by the arm robot When,
And a process of correcting the positional deviation of the arm robot based on the positional deviation amount estimated in the estimating process.

Images