JP2015213139A - Positioning device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To accurately position a moving body at a target position without stopping it.SOLUTION: In a positioning device, a body control part includes: a prediction processing part which calculates, on the basis of an input/output signal U, a position Xof a target before by a delay cycle R associated with a photographic image of a camera; a correction processing part which corrects the position Xbased on an image processing signal Y; a switch which switches and outputs an output value of the correction processing part and the prediction processing part as position information before R cycle in accordance with image processing signal Ytiming; a processing part which estimates a present position X of the target from the position information before the R cycle by application of delay compensation processing; and a command output part which outputs a demand signal to a movement mechanism based on the present position X.

Description

  The present invention relates to a positioning device that positions a moving body at a target position using an imaging device.

  In an industrial machine such as an electronic apparatus manufacturing apparatus or various processing apparatuses, the positioning accuracy of a moving body constituting the apparatus greatly affects the quality of products. For example, a component mounting apparatus (chip mounter) that produces an electronic substrate by mounting an electronic component (such as a resistor or an IC chip) at a predetermined position on a printed circuit board is known. The component mounting apparatus includes, for example, a mounting head that can be moved to an arbitrary position in a horizontal plane by a linear motor mechanism, and a suction nozzle provided in the mounting head. The component supplied from the electronic component feeder is sucked by the suction nozzle, and the suction nozzle is moved and lowered at the target mounting location, and then the electronic component is mounted on the printed board by releasing the component suction. Further, the component mounting apparatus includes a positioning controller that executes a process of determining a drive current command to be given to the linear motor from the position of the mounting head acquired by the encoder at a predetermined period.

  As described above, in the component mounting apparatus, the positioning accuracy of the mounting head that is the moving body greatly affects the quality of the electronic board that is the product. Similarly, the time from when the electronic component is picked up by the suction nozzle to when it moves to the target mounting location greatly affects the productivity. In recent years, miniaturization of electronic components has been progressing rapidly, and there has been a demand for higher speed as well as improvement in positional accuracy of component mounting. However, the printed circuit board has different target mounting locations for each individual due to variations in manufacturing processes, deformation, and the like. That is, when positioning is performed depending on the design value of the printed circuit board, there may be a deviation between the actual electronic component placement location and the target mounting location. In addition, when the device itself is distorted, thermally expanded, or the like, the encoder value may be shifted. This also becomes a cause of displacement. As electronic components to be mounted are miniaturized, the influence caused by such various misalignment factors is increasing.

  On the other hand, a technique is proposed in which the relative displacement of the suction component relative to the target mounting location is calculated based on the image signal of the imaging device, and the suction nozzle is moved in accordance with the calculated displacement to correct the component position. (See Patent Documents 1 and 2).

JP-A-7-115296 US Patent Application Publication 2001 / 0055069A1 JP 2001-325005 A

  Usually, it takes a predetermined time to take an image with a camera and calculate a positional shift based on the image, and this is a delay time. If the mounting head moves within the delay time, the relative relationship between the target mounting position and the mounting head changes during that time. For this reason, in the techniques of Patent Documents 1 and 2, if the mounting head is not stationary at least from the start of shooting until the amount of displacement is calculated based on the image, an appropriate correction operation can be executed from the shot image. Can not. In other words, it is difficult to speed up the positioning control because the operation is started after the mounting head is stopped and the photographing is started, and when the positional deviation correction is executed, the mounting head is stopped again and the photographing is started again.

  In addition, since the camera requires a predetermined exposure time to obtain a photographed image, the camera photography cycle is usually longer than the cycle in which the positioning controller gives a drive current command to the linear motor. Therefore, even if the mounting head is moved during image capturing and image processing, there is a problem that the calculation of the drive current command becomes undefined in a cycle in which no image signal is input. Although there is a technique (see Patent Document 3) for estimating a long-cycle signal as short-cycle information using a short-cycle signal, since the parameters used for the estimation are fixed values, errors in the model used for the estimation Greatly affects the estimation accuracy.

  The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a positioning device that can accurately position a moving body at a target position without stopping the moving body.

  In order to achieve the above object, the present invention is characterized in that the position of the target mounting location of the system is obtained by a position estimation type procedure and a state reconstruction procedure.

  In addition, the present invention includes a processing unit, an imaging unit that images a sample, and a moving body that moves relative to the sample, and the processing unit includes (1) a predetermined input and a signal from the imaging unit. Performing a position estimation type correction procedure on at least one of the images; (2) performing a predetermined process on the result of the correction procedure; and (3) using the result of the predetermined process. It is one feature to obtain the position of.

  According to the present invention, the moving body can be accurately positioned at the target position without being stationary.

1 is a perspective view illustrating an overall configuration of a configuration example of a positioning device according to Embodiment 1. FIG. FIG. 2 is a block diagram illustrating a configuration example of a control device according to the first embodiment. It is a figure which shows typically operation | movement which adsorb | sucks an electronic component with a mounting head, and arrange | positions it in a target position. It is a figure which shows typically the operation | movement which adsorb | sucks an electronic component with a mounting head, and arrange | positions it in a target position. It is a figure which shows typically operation | movement which adsorb | sucks an electronic component with a mounting head, and arrange | positions it in a target position. It is a figure showing an example of a picked-up image. FIG. 3 is a chart illustrating an example of timing of each process of a camera, an image processing unit, and a main body control unit in Embodiment 1. It is a block diagram which shows the structural example of the position estimator in Example 1. FIG. 3 is a block diagram illustrating a configuration example of a positioning control system in Embodiment 1. FIG. FIG. 10 is a diagram illustrating a conceptual configuration of a positioning device according to a second embodiment. FIG. 10 is a chart illustrating an example of the timing of each process of a camera, an image processing unit, and a main body control unit in Embodiment 3. FIG. 10 is a chart illustrating an example of timing of each process of a camera, an image processing unit, and a main body control unit in another example of the third embodiment. It is a block diagram which shows the structural example of the position estimator in Example 4. FIG. It is a figure explaining the expression of a code.

  Embodiments of the present invention will be described below with reference to the drawings. In all drawings, the same portions are denoted by the same reference numerals in principle, and redundant description is omitted.

Applicable Device The application target of the present invention is a positioning device that determines the position of the moving body using a captured image. For example, the present invention can be applied to a rail travel type transfer device and a manipulator type transfer device, but the device mode is not particularly limited, and any device that moves a moving body to a target position can be applied. In addition to the movable element of the positioning device itself, the movable body includes an article conveyed by the movable element. Mounting parts (chips, etc.) for component mounting devices (chip mounters, etc.), semiconductor wafers for semiconductor wafer supply devices, specimens for sample transport devices, tape reels, etc. for self-replenishing feeders, laser processing machines If so, a laser injection unit or workpiece, and if it is a machining center, a cutting tool or workpiece is an example of a moving body.

  Hereinafter, a case where the present invention is applied to a component mounting apparatus will be described as an example.

1. Positioning Device FIG. 1 is a perspective view showing the overall configuration of a configuration example of a positioning device according to Embodiment 1 of the present invention. The positioning device shown in the figure is a component mounting device that arranges electronic components and the like at desired positions such as a printed circuit board, and is schematically shown for convenience of explanation. Further, the X axis and the Y axis shown in the figure are orthogonal to each other in the horizontal plane, and the Z axis is orthogonal to the X axis and the Y axis and extends in the vertical direction.

  A component mounting apparatus 100 shown in FIG. 1 includes a mounting head 10, an X-axis beam mechanism 11, a Y-axis beam mechanism 12, a Y-axis guide mechanism 13, a pedestal 30 (only four legs are shown in the figure), a camera 40, a printed board transport mechanism 60, and a control device 90.

  The Y-axis beam mechanism 12 and the Y-axis guide mechanism 13 make a pair and extend in parallel to each other in the Y-axis direction. Both ends of the Y-axis beam mechanism 12 and the Y-axis guide mechanism 13 are supported by the legs of the gantry 30, respectively. The Y-axis beam mechanism 12 includes a Y-axis linear motor stator 123 and a Y-axis linear scale 124. The Y-axis linear motor stator 123 and the Y-axis linear scale 124 are provided on the side surface of the Y-axis beam mechanism 12 so as to extend in the Y-axis direction. On the other hand, the Y-axis guide mechanism 13 is provided with a Y-axis linear guide 131 extending in the Y-axis direction. A slide unit (not shown) is attached to the Y-axis linear guide 131. The slide unit moves along the Y-axis linear guide 131.

  One end of the X-axis beam mechanism 11 is supported by the Y-axis beam mechanism 12 via a support portion, and the other end is supported by the slide unit of the Y-axis guide mechanism 13. The X-axis beam mechanism 11 includes a Y-axis linear motor movable element 121, a Y-axis position encoder head part 122, an X-axis linear motor stator 113, and an X-axis linear scale 114. The Y-axis linear motor movable element 121 is built in a support portion for the Y-axis beam mechanism 12 and faces the Y-axis linear motor stator 123. The Y-axis position encoder head part 122 is provided at the lower end of the support part for the Y-axis beam mechanism 12 and faces the Y-axis linear scale 124. The X-axis linear motor stator 113 and the X-axis linear scale 114 are provided on the side surface and the upper surface of the X-axis beam mechanism 11 so as to extend in the X-axis direction. With such a configuration, the X-axis beam mechanism 11 is driven by the electromagnetic force induced between the Y-axis linear motor movable element 121 and the Y-axis linear motor stator 123, and the Y-axis beam mechanism 12 and the Y-axis It moves along the guide mechanism 13 in the Y-axis direction. The Y-axis position encoder head unit 122 moves together with the X-axis beam mechanism 11, reads the scale provided on the Y-axis linear scale 124, and converts it into an electrical signal.

  The mounting head 10 is supported by the X-axis beam mechanism 11 via a support portion. The mounting head 10 includes at least one suction nozzle 20, an X-axis linear motor movable element 111, and an X-axis position encoder head unit 112. The suction nozzle 20 vacuum-sucks electronic components at the lower end. The X-axis linear motor movable element 111 is built in the support portion of the mounting head 10 and faces the X-axis linear motor stator 113. The X-axis position encoder head part 112 is provided at the upper end of the mounting head 10 and faces the X-axis linear scale 114. The mounting head 10 has a mechanism for moving the suction nozzle 20 in and out. With such a configuration, the mounting head 10 is driven by the electromagnetic force induced between the X-axis linear motor movable element 111 and the X-axis linear motor stator 113, and the X-axis direction along the X-axis beam mechanism 11. Move to. The X-axis position encoder head unit 112 moves together with the mounting head 10, reads the scale provided on the X-axis linear scale 114, and converts it into an electrical signal.

  The printed circuit board transport mechanism 60 transports the printed circuit board 61 and puts the printed circuit board 61 in and out of the component mounting apparatus 100. At least one target mounting portion 50 having a predetermined shape is formed on the upper surface of the printed circuit board 61. The target mounting location 50 is a target position for mounting an electronic component.

  At least one camera 40 is attached to the mounting head 10 described above in a posture in which the photographing optical axis is directed downward in the Z-axis direction. This camera 40 includes the XY coordinates of the suction nozzle 20 of the mounting head 10 in the visual field region, and when a specific position (origin O described later) in the photographed image is lowered to the position of the suction nozzle 20 (the height of the printed circuit board 61). The tip position of the suction nozzle 20). It is not always necessary for the suction nozzle 20 itself to be included in the visual field region during normal times.

  The control device 90 is electrically connected to elements such as the X-axis linear motor movable element 111, the X-axis position encoder head part 112, the Y-axis linear motor movable element 121, the Y-axis position encoder head part 122, and the camera 40 by wire or wirelessly. It is connected. In addition to the absolute position information of the mounting head 10 in the X-axis direction and Y-axis direction read by the Y-axis position encoder head unit 122 and the X-axis position encoder head unit 112, respectively, is input to the control device 90. Image data of the camera 40 (image data based on a predetermined format) is output to an image processing unit 940 (described later in FIG. 2) of the control device 90. Next, the control device 90 will be described.

  In this embodiment, the suction nozzle 20 or the electronic component sucked by the suction nozzle corresponds to the “moving body” described above, and the mounting head 10, the X-axis beam mechanism 11, the Y-axis beam mechanism 12, and the Y-axis guide mechanism. The drive mechanism which consists of 13 comprises the moving mechanism which moves a moving body. Further, position detection signals output from the X-axis position encoder head unit 112 and the Y-axis position encoder head unit 122, and drive current commands output to the X-axis linear motor mover 111 and the Y-axis linear motor mover 121 are output. This corresponds to an input / output signal exchanged with the moving mechanism for positioning control.

  The X-axis beam mechanism 11 or the Y-axis beam mechanism 12 moves the mounting head 10 or the X-axis beam mechanism 11 by a linear motor, but the first embodiment is not limited to this. For example, the mounting head 10 may be moved using a ball screw mechanism or the like. The mounting head 10 is an example of a moving body that moves relative to a target position.

2. Control Device FIG. 2 is a block diagram showing a configuration example of the control device 90.

  The control device 90 includes a main body control unit 930 and an image processing unit 940.

  The main body control unit 930 includes the X-axis linear motor movable element 111 and the Y-axis based on input / output signals of the X-axis position encoder head unit 112 and the Y-axis position encoder head unit 122 and image processing signals (described later) of the image processing unit 940. The control of the linear motor movable element 121 is repeated in a cycle period, and the function of bringing the position of the suction nozzle 20 closer to the target (in this example, the target mounting position 50 of the printed board 61 shown in FIG. 4) is assumed. The main body control unit 930 includes a CPU 901, a ROM (Read Only Memory) 902, and a RAM (Random Access Memory) 903, and controls the entire component mounting apparatus 100 based on a program stored in the ROM 902 or the RAM 903. Run. The main body control unit 930 acquires the current position on the X axis of the mounting head 10 detected by the X axis position encoder head unit 112 and gives a drive current command to the X axis linear motor movable element 111. Similarly, the main body control unit 930 acquires the current position on the Y axis of the X beam mechanism 11 detected by the Y axis position encoder head unit 122 and gives a drive current command to the Y axis linear motor movable element 121. By such processing of the main body control unit 930, the mounting head 10 moves to a predetermined position. In the following description, the position control of the mounting head 10 by the main body control unit 930 is referred to as “positioning control processing”.

  In addition, although the main body control unit 930 shows a configuration in which a single CPU 901, ROM 902, and RAM 903 are built in, the configuration is naturally not limited to this configuration. For example, a configuration including a plurality of at least one of the CPU 901, the ROM 902, and the RAM 903 may be employed. Thereby, it is possible to execute arithmetic processing necessary for operation control of each component at high speed.

The image processing unit 940
An imaging trigger for instructing the image acquisition timing is given to the camera 40, and a captured image of the camera 40 is acquired. Then, the image processing unit 940 performs predetermined image processing on the captured image, calculates an image processing signal (described later) based on the observation position of the target mounting position 50 in the captured image, and transmits the image processing signal to the main body control unit 930. .

3. Basic Operation FIGS. 3A to 3C are diagrams schematically showing an operation in which an electronic component is attracted by a mounting head and placed at a target position (target mounting location 50 on a printed circuit board). The directions indicated by the Y axis and the Z axis in these drawings coincide with the directions indicated by the Y axis and the Z axis in FIG. 3A to 3C, the central axis of the suction nozzle 20 of the mounting head 10 is C, the optical axis of the camera 40 is L, and the field of view of the camera 40 is F.

First, FIG. 3A shows an example of a state in which the mounting head 10 is moving. As shown in FIG. 3A, when the target mounting location 50 is not in the visual field region F of the camera 40, the main body control unit 930 uses the Y-axis linear information based on the position information on the Y-axis acquired by the Y-axis position encoder head unit 122. and output to the motor armature 121 is moved toward the mounting head 10 which has adsorbed the electronic component 62 to the target point y _t. In parallel with this, the image processing unit 940 transmits an imaging trigger to the camera 40 to start imaging.

Here, the target point _{y t} is the estimated position information of the target mounting position 50 in the operating range of the mounting head 10 that is set on the basis of the design data of the printed circuit board 61 is previously stored in the ROM902 or RAM 903. That is, the main control unit 930, component placement location 50 starts moving mounting head 10 as present in a given target point y _t. However, if the error factors of the distortion or the like during processing of the printed circuit board 61 is present, there is a case where the target mounting portion 50 as shown in FIG. 3A does not match the target location y _t. If there is a deviation in the position of the target mounting position 50 and the target position y _t, the mounting head 10 by placing the target location y _t, the electronic component 62 is arranged at a position shifted from the target mounting portion 50 on the printed circuit board 61 End up.

  As the mounting head 10 continues to move, the target mounting location 50 enters the visual field region F of the camera 40 as shown in FIG. When the component placement region 50 is captured as the visual field region F, the main body control unit 930 starts execution of movement correction control using the observation position of the target mounting location 50 calculated by the image processing of the image processing unit 940.

  Here, FIG. 4 is a diagram illustrating an example of a captured image corresponding to the state of FIG. 3B.

  In the captured image 400 illustrated in FIG. 4, the intersection point of the central axis C of the suction nozzle 20 and the printed board 61 and the target mounting location 50 are captured. Here, the coordinate system in the captured image 400 is defined as a pixel coordinate system in which the intersection point is the origin O (0, 0) and the size of the pixel 401 of the captured image 400 is a unit. In this pixel coordinate system, the directions of the X axis and the Y axis indicated by arrows in the drawing coincide with the directions indicated by the X axis and the Y axis in FIG.

  An example of processing performed on the captured image 400 by the image processing unit 940 is as follows. First, the image processing unit 940 calculates the coordinates (px, py) of the center point p of the target mounting location 50 in the pixel coordinate system from the captured image. Then, by multiplying the coordinates of the obtained point p by a predetermined coefficient, a deviation vector (Δcamx, Δcamy) between the origin O (the center position of the suction nozzle 20) and the center position of the target mounting location 50 is calculated. Then, the calculated deviation vector (Δcamx, Δcamy) is output to the main body control unit 930 as an image processing signal. The image processing signal is a signal based on position information, and is not limited to such a deviation vector, but may be, for example, position information itself.

  The main body control unit 930 controls the position of the mounting head 10 by executing predetermined processing by a position estimator 91 (described later) based on the deviation vector (Δcamx, Δcamy) calculated by the image processing unit 940. Specifically, the positioning control process is repeated until the magnitude of the deviation vector becomes zero. In the following description, such positioning control processing based on the image processing signal of the image processing unit 940 is referred to as “visual servo control processing”. As a result of the visual servo control process, the center axis C of the suction nozzle 20 substantially coincides with the center of the target mounting location 50 as shown in FIG. 3C.

4). Control Cycle FIG. 5 is a chart showing an example of timing in the visual servo control process of each process of the camera 40, the image processing unit 940, and the main body control unit 930.

The main body control unit 930 repeatedly executes the positioning control process at a predetermined cycle. Each processing timing will be described with the execution period τ _u of this positioning control process as one cycle unit. In the following description, when the cycle n (n = 0, 1, 2,...) Is described, it means the nth cycle from the start of the visual servo process.

  In FIG. 5, the camera 40 receives a shooting trigger from the image processing unit 940 at the start of cycle 0, opens the shutter provided in the camera 40 for the time from the start of cycle 0 to the end of cycle 5, and performs exposure (exposure). 1) Do it. This time is set as the exposure time 1, and an image taken during this time is set as the first image. Next, the image processing unit 940 performs image processing (image processing 1) on the first image during the time from the start of cycle 6 to the end of cycle 9. Then, an image processing signal obtained from the first image is input to the main body control unit 930 at the start of the cycle 10.

  After the exposure 1 is completed, the camera 40 receives a shooting trigger from the image processing unit 940, opens the shutter, and performs the next exposure (exposure 2). In this example, the period from the start of cycle 6 to the end of cycle 11 is exposure time 2 for capturing the second image. The image processing unit 940 performs image processing (image processing 2) on the second image during the time from the start of the cycle 12 to the end of the cycle 15. Then, an image processing signal obtained from the second image is input to the main body control unit 930 at the start of the cycle 16.

  Thus, in this embodiment, when the number of cycles for each exposure time is N, the image processing signals (Δcamx, Δcamy) from the image processing unit 940 are intermittently input to the main body control unit 930 every N cycles. Is done. Specifically, in the range shown in FIG. 5, an image processing signal is input to the main body control unit 930 in cycles 10 and 16, and no image processing signal is input to the main body control unit 930 in other cycles. The image processing signal of the image 1 captured by the exposure 1 from the start of the cycle 0 to the end of the cycle 5 is delayed and becomes the cycle 10 and is input to the main body control unit 930. For example, in the state shown in FIG. 3B, the mounting head 10 continues to move even during exposure by a drive current command from the main body control device 930. Therefore, since the image processing signal is not input to the main body control device 930 in each cycle and the image processing signal is input with a delay, it is determined that the central axis C of the suction nozzle 20 matches the center of the target mounting location 50. Otherwise, the suction nozzle 20 may pass through the target mounting location 50, or the position of the mounting head 10 may be moved oscillatingly by repeating such passing. In the present embodiment, as a countermeasure against such a problem, the main body control unit 930 is provided with a position estimator 91 (see FIG. 6).

5. Position Estimator FIG. 6 is a block diagram showing an example of the configuration of the position estimator 91. The position estimator 91 determines the position X _k of the target (target mounting position 50) at the current time from the input / output signal U _k and the image processing signal Y _k input to the cycle k, where the current time is the cycle _k . Output the estimated value. In this embodiment, the input / output signal U _k includes the input / output signal of the main body control unit 930 in the cycle k (the position detection signal of the XY encoder head units 112 and 122 or the drive current command to the XY axis linear motor movable elements 111 and 121). ). Although this embodiment illustrates the position of the target the X _k, there in the internal state quantity X _k which can be expressed by the equation of state, which will be described later, such as the distance between the target object and the central axis C of the suction nozzle 20 (Equation 1) That's fine.

  Here, with reference to FIG. 5, the exposure time is N cycles (in this example, N is an even number), and the time required for image processing by the image processing unit 940 is M cycles. A natural number R = (N / 2 + M). Define R. The natural number R corresponds to the number of delay cycles until an image processing signal calculated from an image captured by the camera 40 is input to the main body control unit 930.

  The position estimator 91 includes an R cycle control command storage buffer 911, a filter 912, and a state reconstructor 913. Next, the R cycle control command storage buffer 911, the filter 912, and the state reconstructor 913 will be described sequentially.

(1) R cycle control command storage buffer 911
The R cycle control command storage buffer 911 is a shift register that stores the values of R input / output signals U for the past R cycles. That is, the input / output signals U _{k−R to} U _k−1 input to the position estimator 91 between the previous cycle and the previous cycle with respect to the current cycle k are stored, and the stored input outputting any signal from the output signal _{U k-R ~U k-1} . (R-1) The input / output signals U _{k−R + 1 to} U _k−1 from the previous cycle to the previous cycle are output to the state reconstructor 913, and the input / output signal U _k−R before the R cycle is a filter. It is output to 912.

(2) Filter 912
Based on the input / output signal U _kR before the R cycle of the current cycle k and the image processing signal Y _k input at the current cycle k, the filter 912 performs the position X of the target mounting location 50 before the R cycle. Estimate the value of _k-R . In this embodiment, a case where a Kalman filter is used as the estimation type filter 912 will be described as an example. The filter 912 includes a prediction processing unit 9121, a correction processing unit 9122, and a switch 9123.

-Prediction processing unit 9121
The prediction processing unit 9121 calculates the predicted value of the position X _k-R of the target mounting location 50 before the R cycle of the current cycle k in the cycle period (τ _u period) based on the input / output signal U _k . Specifically, based on the state equation (Formula 1), the position X _k- of the target mounting location 50 before the R cycle using U _k-R and X _k-R-1 as shown in (Formula 2). Predict _R. X _k−R−1 is the estimated position of the target mounting location 50 calculated by the filter 912 before (R + 1) cycles of the current cycle k.

In the state equation (Equation 1), A _d is a matrix indicating a linear model of time transition of the component placement apparatus 100, B _d is a linear operator for the input / output signal U _k , w _k is a zero mean and process error covariance matrix Q Process noise according to _w . In (Expression 2), the error is ignored.

In addition, the prediction processing unit 9121 also performs prediction calculation of the error covariance matrix M _k-R at this time transition, as in (Equation 3). The predicted value of the error covariance matrix M _k-R is output to the correction processing unit 9122.

In (Expression 3), P _k−R−1 is an estimation error covariance matrix one cycle before the position estimator 91.

Correction processing unit 9122
The correction processing unit 9122 has a function of correcting the predicted value of the position X _k-R of the target mounting location 50 before the R cycle based on the image processing signal Y _{k in} the input cycle of the image processing signal Y _k . Specifically, the correction processing unit 9122 uses the observation equation (Equation 4) to predict the image processing signal Y _k in cycle k and the error covariance matrix M _k-R obtained by the prediction processing unit 9121. Based on the above, the position X _k-R of the target mounting location 50 before the R cycle of the cycle k is corrected and output.

In the observation equation (Equation 4), C _d is a matrix indicating the observation model of the component mounting apparatus 100, and v _k is noise associated with observation according to the zero mean and the observation error covariance matrix Q _v . As can be seen from the observation equation (Equation 4), as one of the features of this embodiment, the image processing signal Y _k is _obtained as a value obtained by observing the position X _k−R of the target mounting location 50 of the component mounting apparatus 100 before the R cycle. We are dealing.

  The processing executed by the correction processing unit 9122 includes processing of (Expression 5) to (Expression 7).

The process of (Formula 5) is a process for calculating the optimum Kalman gain L _k-R . In the process of (Equation 6), the predicted value of the position X _k-R of the target mounting location 50 before the R cycle obtained by the prediction processing unit 9121 is corrected using the Kalman gain L _k-R to correct the value before the R cycle. This is a process for obtaining an estimated value of the position X _k-R of the target mounting location 50. The process of (Expression 7) is a process of calculating an estimation error covariance matrix P _k-R of the estimated value of the position X _k-R of the target mounting location 50 before the R cycle.

Switch 9123
The switch 9123 has a function of switching the input source of the signal input to the state reconstructor 913 to one of the prediction processing unit 9121 and the correction processing unit 9122. The prediction processing unit 9121 executes processing every cycle. On the other hand, the correction processing unit 9122 executes the process only once every N cycles, and does not execute the process for the remaining (N−1) cycles. In this embodiment, as shown in FIG. 5, the image processing signal Y _k is input to the position estimator 91 of the main body control device 913 only once every N cycles. Therefore, in the filter 912, in the cycle in which the image processing signal Y _k is input, the position X _k−R corrected by the correction processing unit 9122 by switching the switch 9123 to the correction processing unit 9122 side is used as the other image processing signal Y. _{In a} cycle in which _k is not input, the switch 9123 is switched to the prediction processing unit 9121 side, and the position X _k-R that is not corrected is switched and output to the state reconstructor 913 as information on the estimated position of the target mounting location 50 before the R cycle. To do. That is, in a cycle in which the image processing signal Y _k is not input, the position X _k-R that is not corrected is output as a substitute value for the corrected position X _k-R . It is not calculated in the cycle of the image processing signal Y _k is not input estimated error covariance matrix P _k-R, the error covariance matrix M _k-R, which is calculated by the prediction processing unit 9121 is directly substituted into P _k-R Is done.

(3) State reconstructor 913
State reconstructor 913 functions as a processing unit for estimating the current position X _k of the target mounting position 50 from the position information before R cycle by performing delay compensation processing R cycles. The state reconstructor 913 includes R delay compensation processing units 9131. In the state reconstructor 913, the input / output signals U _{k−R + 1 to} U _k−1 from the (R−1) cycle to the previous cycle stored in the R cycle control command storage buffer 911 and the current cycle k are stored. by repeating R step delay compensation processing using each of the input and output signals U _k for the output value of the filter 912 (the output value of the switch), to estimate the position X _k of the target mounting position 50 in the current cycle k.

  Specific processing contents of each delay compensation processing unit 9131 are shown in (Equation 8).

That is, for a certain integer i, using the position X _kR + i-1 of the target mounting location 50 of the cycle (k−R + i−1) and the input / output signal U _{k−R + i} of the cycle (k−R + i), the cycle ( This is a calculation for calculating the position X _{k−R + i} of the target mounting location 50 of k−R + i). The state reconstructor 913 executes the delay compensation processing unit 9131 for i = {1,..., R−1, R}. By performing the calculation shown in (Formula 8) R times, the current state can be derived from the state before R cycles.

Command signal output unit of the main controller 930 (e.g., CPU 901), the calculated X-axis linear motor movable elements 111 and Y-axis linear motor movable drive current command based on the position X _k of the target mounting position 50 in the current cycle k Output to the child 121. When the deviation vector (Δcamx, Δcamy) becomes 0 (zero) by repeatedly executing the above positioning control processing every cycle, the main body control device 930 ends the visual servo processing.

  The position estimator 91 described above can be configured by hardware using a predetermined electronic circuit, but can also be embodied in the form of a program executed by, for example, the CPU 901 of the main body control unit 930.

6). Positioning Control System Next, an example of a positioning control system using the position estimator 91 will be described.
FIG. 7 shows an example of a positioning control system in the X-axis direction using the position estimator 91. The information transmission path is extracted and shown. Needless to say, the positioning control system in the Y-axis direction can be configured in the same manner.

The positioning control system in FIG. 7 includes an internal control system and an external control system. The internal control system includes an internal controller C _in 2011, a holder 2012 with a period τ _u , a control target P ₁ 2013 from the control input I _q to the encoder output E, a sampler 2014 with a period τ _u , and an adder 2015. External control system includes a mechanism characteristic 2025 of the external controller _C out 2021, the position estimator 91, a camera image processing unit 2022, the encoder output E up to the camera position _{x a.} Camera image processing unit 2022 outputs the image processing signal Δcam which is a difference between the position x _T street, camera position x _a and the target mounting portion 50 described with reference to FIG. This image processing signal Derutacam, with the N · tau _u sampler 2024 by a delay time element 2023 and the imaging of the R · tau _u by the image processing is outputted as an image processing signal Y.

In the internal control system, control (positioning control processing) for causing the encoder output E to follow the track Rr is executed. The control target P ₁ 2013 is the X-axis linear motor movable element 111, the X-axis position encoder head unit 112, and the mounting head 10, and the internal controller C _in 2011 is realized by, for example, a PID controller provided in the main body control unit 930 Is possible. As shown in FIG. 3A, when the target mounting location 50 is out of the visual field region F of the camera 40 and the image processing signal Y cannot be obtained, the switch 2002 is set by the trajectory generator 2001 according to a command from the CPU 901 (see FIG. 2). The encoder trajectory R _enc side that is the output is selected, and the trajectory Rr becomes the encoder trajectory R _enc . Encoder trajectory _{R enc} is moving trajectory to the target location _{y t} described in Figure 3A.

Thereafter, as shown in Figures 3B and 3C, when the target mounting portion 50 is included in the viewing area F of the camera 40, for example, the switch 2002 in response to a command CPU 901 (see FIG. 2) selects an image trajectory _{R X.} During this time, the position estimator 91 outputs the estimated position X of the target mounting location 50 using the image processing signal Y and the input / output signal U. External controller C _{out 2021} is executed to compute the image trajectory R _X output with the input estimated position X, which is the trajectory Rr of the internal control system. That is, the external controller C _out 2021 corresponds to a command output unit (for example, the CPU 901) that outputs a drive current command to the X-axis linear motor movable element 111. For example, a filter having a DC gain of 1, such as a low-pass filter, is used as the external controller C _out 2021 to converge the trajectory Rr to the estimated position X. By repeatedly executing this visual servo process, the encoder output E substantially coincides with the estimated position X, and the electronic component 62 can be mounted at the target mounting location 50.

The structure of the positioning control system shown in FIG. 7 is an example, and the embodiment is not limited. For example, the control input I _q and the trajectory Rr may be input to the internal controller C _in . Further, since the switch 2002 may finally perform the positioning of the suction nozzle 20 to the target mounting location 50 by visual servo processing, the timing for switching the switch 2002 is not limited to the time when the target mounting location 50 enters the visual field region F. The positioning control process in which the encoder trajectory _Renc is selected for a set time or distance after the target mounting location 50 enters the visual field region F of the camera 40 may be continued. Furthermore, the control system structure of the double loop as shown in FIG. 7 may be used, and different control systems may be formed depending on whether or not the visual field region F of the camera 40 includes the target mounting location 50.

7). As described above, according to the present embodiment, although the image processing signal Y _k can be obtained only every N cycles with the delay cycle R, the output of the correction processing unit 9122 and the prediction processing unit 9121 using the switch 9123. of by switching between alternative output can be every cycle outputs the estimated position X _k of the target mounting position 50 in the current cycle k based on output signals U _k and the image processing signals Y _k for the current cycle k. Thus, it is possible even during the delay cycle R continues to estimate the position X _k of the target mounting position 50. And since the position of the suction nozzle 20 can always be controlled based on the photographed image of the camera 40, the positioning accuracy is also improved. Therefore, the suction nozzle 20 can be positioned at the target mounting location 50 with high accuracy without being stationary.

In particular, in this embodiment, using a modification of the Kalman gain L _k-R when it is calculated from the image processing signal Y _k and the output signal U _k-R. The Kalman gain L _k-R changes from moment to moment so that the error becomes smaller as shown in (Formula 5). Then, by using this Kalman gain L _k-R as shown in (Expression 6), the estimation error can be corrected with time. Therefore, the reliability of estimation accuracy is extremely high.

Further, since the camera 40 is fixed to the suction nozzle 20, the origin O (0, 0) does not move in the captured image 400 even when the suction nozzle 20 is moving. On the other hand, since the target mounting location 50 moves toward the origin O (0, 0) in the captured image 400, the target mounting location 50 appears with subject blur corresponding to the exposure time N. In other words, from the time of input to the main body control device 930, the input image processing signal is accompanied by a delay time due to the exposure time N of the camera 40 and the image processing time M of the image processing device 940. It becomes. The natural number R represents the number of cycles corresponding to this delay time. Specifically, p (px, py) calculated by the image processing unit 940 is the center position of the image with subject blurring, and is the position at the intermediate point in the exposure time. Therefore, it is estimated from the point p (px, py) in the captured image 400 by considering the time obtained by adding half the exposure time N to the image processing time M as the delay cycle R (= N / 2 + M). it can provide highly relevant to the current position X _k of the target mounting position 50.

  Next, Example 2 will be described. The description of the same points as in the first embodiment will be omitted.

  FIG. 8 is a diagram illustrating a conceptual configuration of the positioning device according to the second embodiment. In the figure, the same parts as those in the first embodiment are denoted by the same reference numerals as those in the previous drawings.

  This figure shows that the gantry 30 that supports the X-axis beam mechanism 11 has resonance characteristics. The vibration due to the resonance characteristics of the gantry 30 is a relative vibration of the target mounting location 50 and the component mounting nozzle 20. Therefore, in order to position the suction nozzle 20 at the target mounting location 50, it is necessary to suppress the influence of the gantry vibration. However, the gantry vibration cannot be observed with the X-axis position encoder head 114. Therefore, it is necessary to estimate the vibration by the position estimator 91 based on the image processing signal Y and reflect it in the positioning control system.

Here, the position estimator 91 uses A _d and B _d shown in (Expression 1) and C _d shown in (Expression 4) as models of the component mounting apparatus 100. The model used in the position estimator 91 is generally a model of a driving system including the X-axis linear motor movable element 111, the X-axis position encoder head 114, and the mounting head 10. However, since the gantry vibration cannot be observed by the X-axis position encoder head 114 as described above, it cannot be modeled. Therefore, in this embodiment, in order to estimate the gantry vibration by the position estimator 91, it is conceivable to reflect the vibration mode of the gantry 30 in the model.

(Equation 9) is an example of transfer characteristics from the control input I _q to the image processing signal Δcam in the position estimator 91.

  s is a Laplace operator, m is a mass of the movable part, ω is a resonance angular frequency of the gantry vibration, ζ is a resonance damping constant of the gantry vibration, and K is a resonance residue of the gantry vibration. The first term on the right side is the characteristics of the drive system, and the second term is the vibration mode of the gantry 30. By using a model expressing the transfer characteristics in a discrete state space for correction processing, the position estimator 91 can perform estimation in consideration of gantry vibration, and the suction nozzle 20 can be positioned at the target mounting location 50.

Note that (Equation 9) is an example of transfer characteristics used in the model, and is not intended to limit the present embodiment. For example, the drive system characteristic of the first term on the right side may not be the characteristic from the control command I _q to the image processing signal Δcam, and may be, for example, the relationship between the speed and position with respect to the encoder output E. Although the X-axis is shown here, it goes without saying that the same applies to the Y-axis.

  Next, Example 3 will be described. The description of the same points as in the first embodiment will be omitted.

  In the first embodiment, the case where N indicating the number of cycles of the exposure time of the camera 40 in FIG. 5 is an even number has been described as an example. Therefore, R = (N / 2 + M) indicating the cycle number of the delay time is a natural number. However, the number of cycles of exposure time may be an odd number as shown in FIG. In this case, a Ceil function for rounding up is used, and the cycle number R of the delay time may be defined as R = (Ceil (N / 2) + M).

In FIG. 5, the input period of the image processing signal Y _k to the main body control unit 930 is the exposure time N of the camera 40. This is because the image processing time M of the image processing unit 940 is shorter than the exposure time N of the camera 40. However, the image processing time M may be longer than the exposure time N as shown in FIG. Since the input cycle of the image processing signal Y _k is the longer of the exposure time N and the image processing time M, in this case, the image processing signal Y _k is input to the main body control unit 930 every M cycles. In this case, the timing for switching the switch 9123 in the filter 912 of the position estimator 91 shown in FIG. 6 may be set to once in M cycles. In other words, the processing by the correction processing unit 9122 is executed only once in M cycles, and the (M−1) cycles in between are not executed.

Next, Example 4 will be described.
The description of the same points as in the first embodiment will be omitted.

In the first embodiment, R delay compensation processes are performed using the input / output signals U _{k−R + 1 to} U _k−1 and the input / output signal U _k of the current cycle k stored in the R cycle control command storage buffer 911, respectively. In this embodiment, the delay compensation parameter derivation processing unit 914 repeats the delay compensation parameter derivation process by R steps to calculate the delay error correction parameter in advance. The difference is that a delay compensation process for R cycles is added to the position X _k-R using the calculated delay error correction parameter. The timing in the visual servo control process of each process of the camera 40, the image processing unit 940, and the main body control unit 930 is the same as that in the first embodiment shown in FIG.

  FIG. 11 is a block diagram illustrating an example of the configuration of the position estimator 91 according to the present embodiment. The position estimator 91 includes an R delay error correction parameter derivation processing unit 914, a prediction processing unit 915, an R cycle predicted value storage unit 916, a correction processing unit 917, a switch 918, a buffer 920, and an update processing unit 919. . Hereinafter, each of these elements will be described along the processing contents of the position estimator 91.

・ Prediction processing unit 915
Prediction processing unit 915, input and output signals _{U k} in cycle k, and based on an estimate of the position _{X k-1} of the preceding cycle a target mounting portion 50, the predicted value of _{X k} by a state equation (Equation 1) Output. Specifically, the calculation includes a calculation (Equation 10) for ignoring the error and predicting X _k−R and a calculation (Equation 11) for predicting the error covariance matrix M _k at that time. In (Expression 11), P _k−1 is an estimated error covariance matrix one cycle before an update processing unit 919 described later.

R cycle predicted value storage unit 916
The R cycle predicted value storage unit 916 stores R X _k predicted values (predicted values of X _{k−R to} X _k ) input from the prediction processing unit 915 during the past R cycles, and before the R cycle. The predicted value X _k−R is output to the correction processing unit 917.

Correction processing unit 917, switch 918, buffer 920
The correction processing unit 917 executes processing related to the correction of the predicted position X _k-R based on the image processing signal Y _{k in} the input cycle of the image processing signal Y _k . Specifically, the correction processing unit 917 calculates a delay observation residual ε _k−R indicating the relationship between the image processing signals Y _k and X _k−R using (Equation 12).

However, as in the first embodiment, in this embodiment, the image processing signal _Yk is input to the position estimator 91 once every N cycles. Therefore, the following processing is executed by the buffer 920 and the switch 918. First, the position estimator 91 switches the switch 918 to the correction processing unit 917 side in the cycle in which the image processing signal Y _k is input, and outputs the delay observation residual ε _k−R to the update processing unit 919 and the buffer 920. To remember. On the other hand, in a cycle in which the image processing signal Y _k is not input to the position estimator 91, the switch 918 is switched to the buffer 920 side, and the update processing unit 919 is replaced with the latest delay observation residual ε _k-R stored in the buffer 920. Output to.

  Thus, also in the present embodiment, the processing of the correction processing unit 917 is executed every N cycles, and the (N−1) cycles in between are not executed.

Update processing unit 919, delay error correction parameter derivation processing unit 914
In the update processing unit 919, and a predicted position X _k and the delay observed residual epsilon _k-R target mounting portion 50 in cycle k calculated by the prediction processing unit 915, the estimated position X _k of the target mounting position 50 in cycle k Is calculated.

In this embodiment, the predicted values of the image processing signals Y _k and X _k-R are not compared using the observation equation (Formula 6) as in the first embodiment, but as shown in (Formula 13). Obtained by multiplying the delayed observation residual ε _k−R by the delayed retrograde gain matrix G _k , the predicted position X _k of the target mounting location 50 in the cycle _k , and the observation model matrix C _d , a new pseudo image processing signal Y _k ^* _Is used to estimate Xk.

Specifically, the update processing unit 919 calculates a pseudo Kalman gain L _k ^* related to the pseudo image processing signal Y _k ^* (described later), and performs a prediction process using the pseudo Kalman gain L _k ^* as shown in (Equation 14). Our estimated position X _k by correcting the predicted value X _k-R obtained in part 915 is calculated, also the estimation error covariance matrix P _k of the estimated position X _k after the correction as shown in equation (15) Calculated. The symbols used in (Equation 1) to (Equation 14) are the first estimated value, the second estimated value, the third estimated value, the estimated position, the first predicted value, the first estimated value, as shown in FIG. It can also be expressed as a predicted value of 2.

Here, to describe the pseudo Kalman gain L _k ^* calculation process, first, to the unit matrix I of the linear model matrix A _d and the same size as described above, repeated R step treatment with delay error correction parameter deriving section 914 Thus, the delay error correction parameter Γ _[R] is calculated. The processing contents by the delay error correction parameter derivation processing unit 914 are as shown in (Expression 16).

As shown in (Expression 16), the delay error correction parameter Γ _[R] is obtained by multiplying the linear model matrix _Ad by R steps. Thus, in order to estimate the position X _k in the current cycle k, in addition to the model used in equation (1), varying the parameters when it is computed as the image processing signal Y _k from the input signal U _k (Pseudo By using Kalman gain L _k ^* ), highly accurate estimation is possible.

Next, based on the error covariance matrix M _k calculated by the prediction processing unit 915, the delay error correction parameter Γ _[R] , and the above-described observation model matrix C _d , the mutual covariance is expressed as (Equation 17). The matrix Ω _k is calculated. Similarly, based on the observation model matrix C _d , the error covariance matrix M _k , and the observation error covariance Q _v described above, the observation residual covariance matrix Ψ _k is calculated as in (Equation 18).

Then, a pseudo Kalman gain L _k ^* is calculated from the mutual covariance matrix Ω _k , the observed residual covariance matrix Ψ _k , and the delayed retrograde gain matrix G _k based on (Equation 19).

As described above, the update processing unit 919, the current position X _k of the suction nozzle 20 from the position information before R cycle by performing delay compensation processing R cycles is estimated.

Also in the present embodiment, the current cycle based on the input / output signal U _k and the image processing signal Y _k of the current cycle k is switched by switching the output of the correction processing unit 917 and the alternative output of the buffer 920 using the switch 918. The estimated position Xk of the target mounting location 50 at _k can be output every cycle. Thus, it is possible even during the delay cycle R continues to estimate the position X _k of the target mounting position 50. And since the position of the suction nozzle 20 can always be controlled based on the photographed image of the camera 40, the positioning accuracy is also improved. Therefore, the suction nozzle 20 can be positioned at the target mounting location 50 with high accuracy without being stationary. As described above, the estimation accuracy can be further improved by using the time-varying pseudo-Kalman gain L _k ^* .

Furthermore, (Equation 17) to the pseudo Kalman gain _L ^{k *} calculated based on the equation (19) is carried out in the update processing unit 919, if a linear model matrix _{A d} is constant with respect to time, The delay error correction parameter Γ _[R] can be calculated in advance. That is, once the delay error correction parameter Γ _[R] is calculated once at a predetermined time such as when the component mounting apparatus 100 is initialized, it can be used for each cycle, so the amount of calculation per cycle of visual servo processing Can be reduced. Therefore, the control cycle of visual servo processing can be shortened, and control can be speeded up compared to the first embodiment.

  Needless to say, the position estimator 91 according to the present embodiment can be applied to the positioning control system of FIG. The positioning control system in FIG. 7 is an example as described in the first embodiment. For the model used for the position estimator 91 according to the present embodiment, the mechanism resonance model shown in FIG. 8 or (Equation 9) can be naturally used.

DESCRIPTION OF SYMBOLS 100 ... Component mounting apparatus 10 ... Mounting head 11 ... X-axis beam mechanism 111 ... X-axis linear motor movable element 112 ... X-axis position encoder head part 113 ... X-axis linear motor fixation Element 114 ... X-axis position encoder head 12 ... Y-axis beam mechanism 121 ... Y-axis linear motor movable element 122 ... Y-axis position encoder head part 123 ... Y-axis linear motor stator 124 ..Y axis encoder 13... Y axis guide mechanism 131... Y axis linear guide 20 .. suction nozzle 21 .. electronic component 22 .. target mounting location 30. 50 ... target mounting location 60 ... printed circuit board transport path 61 ... printed circuit board 62 ... electronic component 90 ... control device 901 ... CPU
902 ... ROM
903 ... RAM
940 ... Image processing unit 91 ... Position estimator 911 ... R cycle control command storage buffer 912 ... Filter 9121 ... Prediction processing unit 9122 ... Correction processing unit 9123 ... Switch 913 State reconstructor 9131 ... Delay compensation processing unit 914 ... Delay error compensation parameter derivation processing unit 915 ... Prediction processing unit 916 ... R cycle predicted value storage buffer 917 ... Correction processing unit 918 ... Switch 917 ... Buffer 919 ... Update processing unit 2001 ... Orbit generator 2002 ... Switch 2011 ... Internal controller 2012 ... Holder 2013 ... Mechanical characteristics 2014 ... Sampler 2015 ... Adder 2021 ... External controller 2022 ... Camera image processing unit 2023 ... Delay time element 2024 ... Sample La 2025 ... mechanism characteristics

Claims (9)

A moving mechanism for moving the moving body;
A camera provided in the moving mechanism;
An image processing unit that performs image processing on a captured image of the camera and calculates an image processing signal Y _k based on an observation position of the target in the captured image;
Main body control for bringing the position of the moving body closer to the target object by repeating the control of the moving mechanism in a cycle period based on the input / output signal U _k and the image processing signal Y _k exchanged with the moving mechanism for positioning control. With
The main body control unit
Due to the exposure time N of the camera and the image processing time M of the image processing unit, the position X _k-R of the target before the delay cycle R accompanying the captured image is determined based on the input / output signal U _k . A prediction processing unit that calculates the cycle period,
A correction processing unit that performs correction of the position X _k-R or a correction process that is related to the position X _k-R based on the image processing signal Y _{k in} the input cycle of the image processing signal Y _k ;
A switch that outputs the output value of the correction processing unit in a cycle in which the image processing signal Y _k is input, and a switch that outputs a substitute value as position information before the R cycle of the target in other cycles;
A processing unit for estimating a current position X _k from the position information before the R cycle by performing delay compensation processing R cycles the target,
Positioning device, characterized in that said a command output section instruction signal to the moving mechanism based on the current position X _k. The positioning device of claim 1,
The current value _Xk is estimated using a time-varying parameter calculated based on the input / output signal _Uk and the image processing signal _Yk . The positioning device of claim 1,
The positioning apparatus according to claim 1, wherein the input period of the image processing signal _Yk is the longer of the exposure time N and the image processing time M. The positioning device of claim 2,
A buffer for storing input / output signals for R cycles of the position detector;
The delay compensation processing for R cycles is performed by performing delay compensation processing using the input / output signals U _{k−R + 1 to} U _k from (R−1) cycles before the current cycle stored in the buffer to the current cycle. A positioning device characterized in that it is a process of repeating R steps for an output value. 5. The positioning apparatus according to claim 4, wherein the substitute value is the position _Xk-R before correction output from the prediction processing unit. The positioning device of claim 2,
A delay error correction parameter derivation processing unit for calculating a delay error correction parameter in advance by repeating the delay compensation parameter derivation process by R steps;
The delay compensation process for the R cycle is a process for delay compensating the output value of the switch using the delay error correction parameter calculated by the delay error correction parameter derivation processing unit. The positioning device of claim 6,
A buffer for storing the output value of the correction processing unit;
The positioning apparatus according to claim 1, wherein the substitute value is a latest output value of the correction processing unit stored in the buffer.   The positioning apparatus according to claim 2, wherein the correction process is executed using a vibration model of a gantry that supports the moving mechanism.   3. The positioning apparatus according to claim 2, wherein the delay cycle R is defined by (Ceil (N / 2) + M).

Images