JP2016178441A - Digital filter, mass measurement device, and robot system incorporating mass measurement device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a digital filter capable of performing filtering at optimum timing matched with a motion of a robot, a mass measurement device using the digital filter, and a robot system in which the measurement device is incorporated.SOLUTION: A linear phase filter is used as a digital filter for performing filtering on quantized time series data from a sensor for detecting force or acceleration applied to an article whose speed is varying, the linear phase filter taking, as a phase delay, time from a point of time at which an extreme value emerges in the time series data during speed variation to a point of time immediately before elimination of the speed variation. The digital filter is constituted by programmable processing means for performing arithmetic processing on the quantized time series data and performs, when the time has been detected from time series data stored in storage means for storing time series data before filtering, filtering using a linear phase filter having a phase delay corresponding to the detected time.SELECTED DRAWING: Figure 12

Description

  The present invention relates to a digital filter that is optimal for measuring the mass of an article while the speed of the article is changing, a mass measuring device using the digital filter, and a robot system incorporating the mass measuring device.

  In spring balances and electronic balances, the mass is measured while the article is stationary in order to eliminate the influence of acceleration other than gravitational acceleration. However, recently, the opportunity to move an article with a robot hand has increased, and accordingly, there is a demand for detecting the mass of the article when it is lifted and using it for post-processing.

  In order to meet this demand, the present applicant has developed a series of mass measuring apparatuses shown in the following patent documents. This mass measuring device detects a force sensor that detects a force acting on a moving article, a robot hand that holds the article, a robot arm that moves the article while accelerating, and an acceleration that acts on the moving article. The mass of the article is measured from the output of these sensors while the speed of the article changes.

JP2013-079931A JP 2013-174503 A JP 2013-174570 A JP 2013-185846 A JP 2013-185847 A JP 2013-185848 A JP 2013-195200 A

  When incorporating this mass measuring device into a robot, it is necessary to link the movement of the robot and the mass measurement by the mass measuring device. For example, when a robot lifts an article and tries to sort the article according to the mass at that time, the robot needs to receive the measured mass from the mass measuring device before deciding where to sort the lifted article. .

  However, since the output of the force sensor or the acceleration sensor is output through a digital filter, the output has a phase delay, which results in a delay in calculating the mass of the article. Therefore, it is necessary to match the timing of mass calculation with the movement of the robot.

  The present invention is intended to solve the problems peculiar to the mass measuring device incorporated in such a robot, and a digital filter capable of filtering at an optimal timing matched to the movement of the robot, and a mass measuring device using the digital filter An object of the present invention is to provide a robot system in which the measuring device is incorporated.

  In the first invention, as a digital filter for filtering quantized time series data from a sensor that detects a force or acceleration acting on an article whose speed is changing, a final filter appearing in the time series data during the speed change is used. It is characterized by using a linear phase filter whose phase delay is the time from when the extreme value is taken until immediately before the speed change disappears.

  FIG. 1 shows an output waveform f1 of the force sensor when the robot lifts an article, and an output waveform f2 after the quantized data of the output waveform f1 is filtered by a linear phase filter. As shown in FIG. 1, the output waveform f2 of the linear phase filter has a phase delay (X time) with respect to the output waveform f1 before filtering. The output of the acceleration sensor is also not shown here because it has the same characteristics with only different amplitudes. The movement of the robot at this time includes an acceleration process A in which an object is grasped and lifted while accelerating, and a deceleration process B in which the robot stops while decelerating after reaching a predetermined speed.

  In a robot performing such an operation, in order to smoothly move the robot to the next operation, it is necessary to obtain the mass of the article by time t2 when the robot enters the next operation.

The mass of the article can be obtained by detecting the force and acceleration acting on the article being accelerated / decelerated. For example, when the mass of the article is m, the mass m is obtained from the following calculation formula.
m = S {(Fm / Fa)-(Fmz / Faz)}
Here, S is a span coefficient. Fm is the output of the force sensor when moving an article of mass m. Fa is the output of the acceleration sensor at that time. Fmz is the output of the force sensor when the robot is moved with no load. Faz is the output of the acceleration sensor at that time. The span coefficient S and the output ratio (Fmz / Faz) of each sensor at no load are values that are set and registered in the initial setting. Therefore, if the output ratio (Fm / Fa) of each sensor when the article is lifted is obtained, the mass m of the article can be obtained.

  The timing for obtaining the output ratio is to obtain the mass of the article by using each sensor output at the same timing as long as the article is accelerating or decelerating, unless high measurement accuracy is required. Can do.

  However, since the force sensor and the acceleration sensor are different sensors, and the A / D converters that quantize their outputs are also different, the phase characteristics of the sensor outputs are slightly different. Therefore, when high measurement accuracy is required, the mass of the article is obtained using the maximum and minimum values after filtering, which are most unlikely to be affected by the difference. At this time, if the mass of the article is obtained based on the deviation between the maximum value and the minimum value of each output, the measurement accuracy can be further improved by eliminating the influence due to the fluctuation of the zero point of each output.

For example, assuming that the maximum output after filtering in FIG. 1 is a, the minimum output is b, and the zero point (reference value) is c, the maximum value Fmax = ac−minimum value Fmin = bc
Therefore, the deviation Fm between the maximum and minimum values is
Fm = (ac)-(bc) = ab
It becomes. In the right side of this equation, the term of the zero point c is eliminated, so that the influence of the changing zero point can be eliminated. The same applies to the acceleration sensor.

  Since the maximum value and the minimum value of the output waveform f2 appear later than the output waveform f1 before filtering, as shown in FIG. 1, the time t1 when the output waveform f2 after filtering becomes the minimum value is If the mass appears using the minimum value b at the time t1 if it appears immediately before the time t2 when the operation starts, the mass measurement with high accuracy can be performed without delaying the movement of the robot. In FIG. 1, the phase delay at that time is the time T from the time t0 at which the final extreme value of the time-series data is obtained to the time t2 immediately before the speed change disappears.

Therefore, in the first invention, as a digital filter for filtering the quantized time-series data, the time from the time when the final extreme value appearing in the time-series data during the speed change to the time immediately before the speed change disappears is set as the phase delay. Use a linear phase filter. Specifically, the time T from the time t0 when the time series data in FIG. 1 starts to decrease to the increase until the time t2 when the speed change ends is defined as a phase delay.
FIG. 2 shows a comparison between the output waveform f2 after filtering by the phase delay linear phase filter and the waveform f1 before filtering. When this filter is used connected to the output side of the force sensor and the acceleration sensor, the output waveform f2 after filtering is matched immediately before the time t2 when the robot enters the next operation after matching the phase delay between the two. Since the mass of the article can be obtained from the deviation between the maximum value and the minimum value (in the force sensor, the deviation between a and b in FIG. 1), it can be made in time for the movement of the robot.

  1 and 2, the time-series data when lifting an article has been described. However, when the robot moves in the opposite direction, for example, when the upper article is grabbed and lowered, the output waveform of each sensor is in the reverse direction. Therefore, at that time, the time from the time when the time-series data takes the extreme value at which the time series data changes from the increase to the decrease until immediately before the speed change disappears becomes the phase delay time.

  The first invention is effective when the acceleration / deceleration operation of the robot is constant. However, when the acceleration / deceleration operation of the robot changes, it is necessary to adjust the phase delay of the linear phase filter accordingly. .

  The second invention responds to such a demand, and the digital filter includes a programmable processing means, and the processing means has storage means for storing the time-series data before filtering, and stores it therein. When the time from the time when the final extreme value is taken to the time immediately before the speed change disappears is detected from the time-series data, a linear phase filter having a phase lag corresponding to the detection time is set in the digital filter and set. The time-series data to be quantized thereafter is filtered by a linear phase filter.

  Since the time-series data stored in the storage means is quantized data before filtering, as shown in FIGS. 1 and 2, the waveform changes minutely including vibration components. Therefore, in order to smooth this, the sensor output is repeatedly input and stored under the same conditions, and time series data averaged a plurality of times at the same time is stored in the storage means. Since the time-series data stored in this way has no phase delay, the time from the time when the above extreme value is obtained to the time immediately before the speed change is detected is detected. That is, the final extreme value of the deceleration process B appears at the time when the sensor output changes from decrease to increase (t0 in FIG. 1), and the time when the speed change disappears is the time when the sensor output returns to zero (t2 in FIG. 1). Therefore, when the time t0 at which the final extreme value is obtained and the speed change end time t2 are detected from the change in the stored time series data and the time difference (T = t2−t0) between them is obtained, the phase of the linear phase filter is obtained. A delay time T is obtained. Therefore, the processing means sets the linear phase filter having the phase delay time T as a digital filter, and filters the time-series data to be quantized thereafter with the linear phase filter.

Since the linear phase filter having a phase delay of T time is filtered using time series data of 2T time, the filter order (number of taps) in that case is
2T / sampling period
It becomes. Since the sampling period in this case is determined by the clock of the A / D converter to be used, the number of taps is determined according to the obtained phase delay T, and the phase delay is T by sequentially filtering the time-series data of the number of taps. A time linear phase filter is constructed.

In the second invention, the number of taps is changed corresponding to the phase delay time, but there are cases where the filter coefficient is also desired to be changed according to the number of taps.
The next third invention responds to such a demand, and the processing means includes a table for storing a plurality of linear phase filters having different phase delays. When the processing means detects the time, the time is calculated. A linear phase filter having a phase lag closest to that time within a range not exceeding is read from the table and set as a digital filter, and the set linear phase filter is used to filter time-series data that is subsequently quantized. Features.

  FIG. 3 shows an example of a table storing a plurality of linear phase filters having different phase delays. In this figure, the filter time is a time corresponding to 2T described above. Therefore, each linear phase filter has a phase delay of 1/2 of the filter time set for each linear phase filter. In addition, an optimum filter coefficient corresponding to each filter time is set for each filter.

  If the delay time T detected by the processing means is, for example, 52 msec, the processing means reads the nearest 100 msec filter from the table within a range not exceeding 104 msec which is doubled, and the read filter Thus, a linear phase filter is formed. In this case, the phase delay is 50 msec.

  Here, the filter having the closest phase delay time (50 msec) within the range not exceeding the detected delay time (52 msec) is selected as shown in FIG. 2 as the extreme value b of the output waveform f2 after filtering. This is to make the calculation of the mass in time for the movement of the robot by making the point of time at which the robot appears slightly earlier than the point of time t2 immediately before the robot moves to the next motion.

  In the inventions so far, the phase delay time is detected from the change in the time series data before filtering. However, when multiple extreme values appear in the sensor output, it may be difficult to automate the time detection. For example, FIG. 4 shows a case where two peaks of the output waveform appear in the acceleration / deceleration period (A + B). In this case, in the deceleration process B, the time series data changes from decreasing to increasing. Since the time point appears at the point d and the point f, and the time point when the increase changes to the decrease appears at the point e, the phase delay time differs depending on the extreme value at which the point is taken.

Accordingly, the fourth invention is a digital filter that filters the quantized time series data from the sensor that detects the force or acceleration acting on the article whose speed is changing. Possible processing means, the processing means storing the time-series data before filtering, a table storing a plurality of linear phase filters having different phase delays, and any one of the tables stored in the table Specifying means for specifying a linear phase filter; and display means for displaying the time-series data before filtering stored in the storage means,
From the time-series data displayed on the display means by the operator, the time from the time when the extreme value appearing in the time-series data is taken to the time immediately before the speed change disappears, and the time within the range not exceeding the calculated time is the most. When a linear phase filter having a near phase delay is designated by the designation means, the processing means reads the designated linear phase filter from the table, sets the read linear phase filter to the digital filter, and sets the set linear phase filter. It is characterized by filtering time-series data that is subsequently quantized by a filter.

  The storage means stores averaged data of time-series data that is repeatedly input and is read out and displayed on the display means. The waveform fx in FIG. 5 is an example of averaged time-series data that is read from the storage means and displayed.

  The operator looks at this output waveform fx and takes the time (T = t3−t1) from time t1 at which the minimum value d of the deceleration process B is shown to time t3 when the speed change almost disappears, for example, 172 msec. Then, the filter having the closest filter time is searched from the table of FIG. 3 within a range not exceeding the 344 msec filter time that is doubled. In this case, since the filter time 340 msec is the nearest filter time, when a filter having the time is designated by the designation means, the processing means reads the designated filter from the table, and then performs quantization with the filter. Filter the time series data. As a result, the filtered waveform fx2 becomes a signal delayed by 170 msec from the waveform fx before filtering, as indicated by the one-dot chain line in FIG. 5, and the mass of the article is determined from the deviation between the maximum and minimum values of the waveform fx2. Ask for.

  In the above example, the time point t1 at which the minimum extreme value appears and the time point t3 at which the speed change almost disappears are selected as the phase delay time T. Instead, the time point t2 at which the speed change disappears is selected. Thus, the phase delay time (T = t2-t1) may be used. Alternatively, the time t5 when the next extreme value appears and the time t2 when the speed change disappears may be selected, and the phase delay time TX may be calculated from the time difference (TX = t2−t5).

  The fifth invention is characterized in that the time-series data stored in the storage means is data obtained by averaging the time-series data repeatedly input. As described above, fine vibration components are included in the data obtained by quantizing each sensor output. This vibration component appears when the article is lifted, due to the vibration of the contents or the vibration of the robot hand that lifts the article. Are repeatedly stored, and an average of those time series data is stored. Thereby, the time-series data before filtering from which fine vibration components are removed can be stored.

  When detecting the phase delay time of the filter using the stored time series data, any time series data of the force sensor or the acceleration sensor may be used. However, in the case of the force sensor, the robot hand holding the article The output of the force sensor varies considerably depending on the characteristics of the sensor, its tare, and whether the article is solid or fluid. On the other hand, since the output of the acceleration sensor can eliminate such influence, it is better to use the quantized data of the acceleration sensor when detecting the phase delay time of the filter.

A sixth invention is a mass measuring apparatus for measuring the mass of an article from the force and acceleration acting on the article whose speed is changing,
A force sensor for detecting a force acting on the article;
An acceleration sensor for detecting an acceleration acting on the article;
First and second A / D converters for quantizing each sensor output;
First and second digital filters for filtering the outputs of these A / D converters, respectively.
The first and second digital filters are configured by the above-described digital filters having the same phase characteristics.

  By configuring the digital filters connected to the force sensor and acceleration sensor with linear phase filters with the same phase lag, high-precision mass measurement is possible by eliminating the phase difference of the quantized time-series data of each sensor. become.

When such a mass measuring apparatus is incorporated in a robot, it is necessary to match the movement of the robot and the measuring operation of the mass measuring apparatus.
The seventh invention responds to such a demand, and is incorporated in a robot that moves an article. While communicating with the robot, the mass of the article is measured from the force and acceleration acting on the article whose speed is changing. A mass measuring device for transmitting to the robot,
A force sensor for detecting a force acting on the article;
An acceleration sensor for detecting an acceleration acting on the article;
First and second A / D converters that respectively quantize the outputs of the sensors;
A table storing a plurality of linear phase filters having different phase delays;
First and second digital filters for respectively filtering the first and second A / D converter outputs;
A controller for determining the mass of the article based on the output of these digital filters, and transmitting the determined mass to the robot,
When the control unit receives the synchronization signal including the acceleration / deceleration time of the article from the robot, the phase delay closest to the phase delay corresponding to the time is within a range not exceeding 1/4 of the acceleration / deceleration time. The linear phase filter is selected from the table, and the selected linear phase filter is set as the first and second digital filters.

  For example, when the robot lifts an article and tries to sort it based on the mass of the article at that time, a synchronization signal is output at the timing when the robot lifts the article. When the control unit of the mass measuring apparatus receives it, the sensor output is input from that time to obtain the mass of the article, and the obtained mass is transmitted to the robot. Thus, the mass of the article lifted by the robot can be measured before moving to the next sorting operation, so that the next sorting operation can be made in time.

  As a synchronization signal transmitted from the robot to the mass measuring device, for example, a signal notifying the start of movement of the article as shown in FIG. 6A or a signal notifying the acceleration / deceleration time of the article as shown in FIG. There is. Furthermore, there is a signal for informing the start and end of the acceleration / deceleration time as shown in FIG.

  Since the maximum value and the minimum value of the force sensor output during acceleration / deceleration basically appear in the acceleration process A and deceleration process B of the article as shown in FIG. And the signal of FIG. 6C is preferable. Therefore, when the signals of FIG. 6B and FIG. 6C are transmitted from the robot and the robot accelerates and decelerates at a constant movement, it does not exceed 1/4 of the ON time of the signal. Then, a linear phase filter having a phase delay closest to the phase delay corresponding to the time is selected from the table, and the selected linear phase filter is set as the first and second digital filters.

Since the movement of the robot can be freely changed by a program, for example, as shown in FIG. 7, the robot outputs a synchronization signal f3 including an acceleration / deceleration time (A + B) to the mass measuring apparatus at the same time. Is lifted. In this case, when the output of the force sensor or the acceleration sensor changes while drawing a waveform f4 close to a sine wave as shown in FIG. 7, it is within a range not exceeding 1/4 of the ON time of the synchronization signal f3. The linear phase filter having the phase lag closest to the phase lag corresponding to the time is selected from the table. By doing so, the time t4 at which the minimum value of the filtered waveform f5 appears can be set immediately before the time t2 at which the sensor output f4 returns to almost zero, and as described above, the maximum of the output waveform f5 after filtering is maximized. The mass can be determined using the deviation between the value and the minimum value.
The output waveform f4 in FIG. 7 is expressed as an average of a plurality of quantized time series data from the acceleration sensor.

  In an eighth aspect of the invention, when the control unit receives a synchronization signal from the robot, the control unit obtains the mass of the article based on a maximum value or a minimum value of each digital filter output or a deviation between the maximum value and the minimum value. And

  Since there is a slight phase difference between the outputs of the force sensor and the acceleration sensor, when the high measurement accuracy is required, the maximum value and the minimum value after filtering that minimize the influence of the phase difference are used. The maximum value appears when the filtered waveform f5 changes from increase to decrease (time h in FIG. 7), and the minimum value appears when the waveform f5 changes from decrease to increase (time i in FIG. 7). When the control unit receives the synchronization signal from the robot, the control unit detects the maximum value or the minimum value from the output waveform after filtering, and if they can be detected, obtains the mass of the article using them. Alternatively, the mass of the article is obtained based on the detected deviation between the maximum value and the minimum value.

A ninth invention is a robot system including the above-described mass measuring device and a robot in which the mass measuring device is incorporated,
A robot hand that holds an article;
A robot arm for moving the robot hand;
A controller for driving and controlling the robot hand and the robot arm;
A robot with
The mass measuring device includes the force sensor provided between the robot hand and the robot arm, and the acceleration sensor provided on the robot hand.
As a result, the robot hand can grasp the article and the robot arm can lift and move it. At that time, the force acting on the article is detected by the force sensor, and at the same time, the acceleration acting on the article is detected by the acceleration sensor. Detected. The control unit of the mass measuring device obtains the mass of the article based on the output of each sensor, and returns the obtained mass to the robot. Based on this, the robot sorts and sorts the articles.

  According to the present invention, it is possible to filter the output of each sensor at an optimal timing that matches the movement of the robot. In addition, since such a digital filter is used in the mass measuring device, it is possible to perform mass measurement with high accuracy even for an article being accelerated or decelerated. Furthermore, since a high-accuracy mass measuring device is incorporated in the robot, for example, even when a lifted article is sorted by its mass, a high-precision sorting operation can be performed. Therefore, the field of use of the robot can be further expanded.

An example of the output waveform of the force sensor when the article is lifted and the output waveform after filtering the output waveform. An example of the output waveform of the force sensor and the output waveform obtained by filtering the output waveform with a linear phase filter having a phase delay of T time. An example of a table in which a plurality of linear phase filters having different phase delays are stored. An example of the waveform before filtering of the force sensor output when an article is lifted. An example of a display screen when calculating the phase delay time from the displayed output waveform. An example of the synchronous signal transmitted to a mass measuring device from a robot. Explanatory drawing which calculates | requires the phase delay time T based on the synchronizing signal f3 from a robot. 1 is a schematic configuration diagram of a robot system according to an embodiment of the present invention. The structure schematic of other embodiment of a robot hand. The configuration block diagram of one embodiment of the 2nd invention. The configuration block diagram of one embodiment of the 3rd invention. The configuration block diagram of one embodiment of the 4th invention. The configuration block diagram of one embodiment of the 6th-the 9th invention.

  Hereinafter, an embodiment of the present invention will be described with reference to the drawings. The following embodiments do not limit the technical scope of the present invention.

(1) Basic Configuration of Robot System FIG. 8 is a schematic configuration diagram of a robot system 100 including the mass measuring device 10 according to an embodiment of the present invention and a robot 20 in which the mass measuring device 10 is incorporated. In this figure, the mass measuring apparatus 10 includes a force sensor 11 that detects a force acting on a moving article Q, and an acceleration sensor 12 that detects an acceleration acting on the article Q. On the other hand, the robot 20 includes a robot arm 21 and a robot hand 22 that holds the article Q.

  For example, a strain gauge type load cell is used as the force sensor 11. In the strain gauge type load cell, the free end side is displaced relative to the fixed end side by the load of the article Q, and thereby mechanical strain generated in the load cell is converted into an electric signal. The electric signal is output as a digital signal to a control unit 30 (described later) via an amplifier circuit (not shown), a first A / D converter 13a (described later), and a digital filter 14.

  As the acceleration sensor 12, for example, any one of a MEMS type small acceleration sensor and a general commercially available acceleration sensor is appropriately employed in addition to a strain gauge type load cell. The acceleration sensor 12 also includes an amplifier circuit (not shown), a second A / D converter 13b, and a digital filter 14, and outputs the digital signal to the control unit 30 described later. The acceleration sensor 12 and the robot hand 22 are attached to the free end side of the force sensor 11.

  10-13, the 1st A / D converter 13a and the 2nd A / D changer 13b convert the analog signal of each sensor 11 and 12 into a digital signal, and both the converters 13a and 13b include A synchronization signal is simultaneously input from a high-performance clock (not shown) so that the respective conversion operations are synchronized. Thereby, each A / D changer 13a, 13b quantizes each sensor 11, 12 output at the same timing, and outputs it to the digital filter 14.

  The digital filter 14 is connected to the output sides of the first and second A / D converters 13a and 13b, respectively, and filters the digital data output from the A / D converters 13a and 13b. In particular, the digital filter 14 is composed of a programmable DSP (digital signal processor) or microcomputer so as to be a linear phase filter capable of controlling the phase delay of the output. The digital filter 14 is connected to a later-described control unit 30 configured by a computer, and various linear phase filters having different phase delays are set in the digital filter 14 from the control unit 30. . Therefore, this control unit 30 is used as a processing means for setting various linear phase filters in the digital filter. The control unit (processing unit) 30 includes a storage unit 15, a detection unit 16, and a table 17.

  In particular, the storage means 15 stores averaged data of time-series data before filtering. Since the time-series data before filtering includes noise components due to vibrations of the robot arm 21 and the robot hand 22, the robot 20 lifts the article Q a plurality of times under the same conditions during the test operation, and before filtering. Quantized time-series data is stored, and an average of those time-series data is stored.

  The detection means 16 changes the speed from the time when the time series data changes from decrease to increase (for example, t0 in FIG. 1) or from the time when change increases to decrease from the change of the time series data stored in the storage means 15. Is detected until the time point (e.g., t2 in FIG. 1) ends. Specifically, the time series data is sequentially compared to detect the time when it becomes the final extreme point and the time when the speed change ends, the time difference (T = t2−t0) between the detected times is obtained, and the delay of the filter 14 Time T is detected.

When the delay time T is detected in this way, the linear phase filter having a phase delay of T time filters using time series data of 2T time, and the filter order at that time is 2T / sampling period.
It becomes. Further, since the sampling period is determined by the clocks of the A / D converters 13a and 13b, the filter order is determined according to the obtained time T. When the filter order is determined in this way, the processing means 30 sets them in the digital filters 14 and 14 together with the predetermined filter order, thereby constituting a linear phase filter with a phase delay of T time. At that time, it goes without saying that a linear phase filter having the same phase delay is set for each digital filter 14.

  The table 17 in FIGS. 11 and 12 stores a plurality of linear phase filters having different phase delays. Specifically, as shown in FIG. 3, the filter numbers and the filter times are stored in association with each other. It is. Further, in each linear phase filter stored in the table 17, an optimum filter coefficient corresponding to each filter time is set.

  When this table 17 is incorporated, the detection means 16 selects a linear phase filter having a phase delay time closest to the detected time T from the table 17 within a range not exceeding the detected time T. For example, if the detection time T is 52 msec, the detection means 16 reads out from the table 18 a filter having a filter time of 100 msec that is closest to the filter within a range not exceeding 104 msec, which is doubled, and outputs it to the digital filter 14. , 14 is set.

  The digital filters 14 and 14 respectively convert the quantized data output from the first A / D converter 13a or the second A / D converter 13b based on the linear phase filter of the phase delay time set as described above. Filtering is performed, and the result is output to the control unit 30.

  Returning to FIG. 8, the robot arm 21 moves the robot hand 22 three-dimensionally, and one end of the force sensor 11 is fixed to the distal end portion 23 thereof. As this robot arm 21, for example, an articulated robot, a parallel link robot, or the like can be adopted.

  The robot hand 22 grips the article Q, and the robot hand 22 of FIG. 8 shows an example of a finger type driven by a motor. Instead, the article Q is negatively pressured as shown in FIG. An air adsorbing type that adsorbs and holds can be used. The finger type of FIG. 8 is suitable when the article Q is a solid material, and the air adsorption type of FIG. 9 is suitable when the shape is not constant, such as a bag-packed product.

  FIG. 9 shows a schematic configuration diagram of this air adsorption type. In this type, a bellows-like suction pad P made of silicon rubber is attached to an aluminum box B so that the suction pad P communicates with the inside of the box B. By holding the inside of the aluminum box B at a negative pressure, the articles Q are sucked and held by the suction pads P. In addition, since a plurality of articles Q may be simultaneously held by a large number of suction pads P, the number, shape, arrangement, etc. of the suction pads are appropriately changed according to the type of the articles Q. Therefore, a plurality of types of robot hands 22 are prepared according to the type of the article Q, and these are used properly according to the type of the article Q.

(2) Basic Configuration of Control System FIG. 12 shows a block diagram of the configuration of the mass measuring apparatus 10 incorporated in the robot 20, and FIG. 13 shows a configuration block of the control system of the robot system 100 including the robot 20 and the mass measuring apparatus 10. The figure is shown. In these drawings, the outputs of the force sensor 11 and the acceleration sensor 12 of the mass measuring apparatus 10 are controlled via the corresponding A / D converters 13a and 13b and digital filters (processing means) 14 and 14, respectively. Input to the unit 30. The control unit 30 is configured by a computer and realizes various functions by executing various programs stored in the built-in memory. Further, as shown in FIG. 13, the control unit 30 is provided with a communication means 31 for communicating with the robot 20, and is further connected with an operation display unit 32.

  The communication means 31 receives a synchronization signal from the communication means 25 on the robot 20 side, and returns the measured mass of the article to the communication means 25. The communication means 31 is connected to the communication means 25 via a serial or parallel communication cable. ing.

  The operation display unit 32 is configured by a touch panel, and displays operations stored in the storage unit 15 for quantizing time-series data of the sensors 11 and 12 by screen operation of the display unit and stored contents stored in the table 17. It is possible to operate. Further, a designation means 33 for designating a desired linear phase filter from the stored contents of the table 17 displayed on the display unit 32, and a display for displaying the averaged time series data before filtering stored in the storage means 15. Means 34. The operation display section is provided with a “span key”, a “zero key”, and a “filter setting key” described later.

The control part 30 calculates | requires the mass of the articles | goods from which the speed is changing from the following calculation formula by running a program.
m = S {(Fm / Fa)-(Fmz / Faz)}
Here, S is a span coefficient. Fm is the output of the force sensor 11 when moving an article of mass m, and Fa is the output of the acceleration sensor 12 at that time. Fmz is an output of the force sensor 11 when the robot hand 22 is moved in an unloaded state, and Faz is an output of the acceleration sensor 12 at that time.

The span coefficient S is a coefficient represented by the following relational expression.
S = ms / {(Fms / Fas)-(Fmz / Faz)}
Here, ms is the mass of the span weight, Fms is the output of the force sensor 11 when the robot hand 22 lifts the span weight, and Fas is the output of the acceleration sensor 12 at that time. Fmz and Faz are the output of the force sensor 11 when the robot hand 22 is moved with no load, and the output of the acceleration sensor 12 at that time, as described above.

These span coefficients S and output ratios (Fmz / Faz) are set and registered by the following operations in the initial setting.
First, prepare a span weight of mass ms, and then press the “Span key”. Then, the robot hand 22 lifts and lowers the span weight of mass ms. Meanwhile, since Fms is output from the force sensor 11 and Fas is output from the acceleration sensor 12, the control unit 30 stores the output ratio (Fms / Fas) at that time.
Subsequently, when the “zero key” is pressed, the robot hand 22 moves up and down without holding anything, that is, in a no-load state. Meanwhile, since Fmz is output from the force sensor 11 and Faz is output from the acceleration sensor 12, the control unit 30 stores the output ratio (Fmz / Faz).

  Further, when the control unit 30 receives the synchronization signal including the acceleration / deceleration time of the article Q from the robot 20, the control unit 30 is closest to the phase delay corresponding to the time within a range not exceeding ¼ of the acceleration / deceleration time. A phase delay linear phase filter is selected from the table 17, and the selected linear phase filter is set in the first and second digital filters 14 and 14.

  On the other hand, the robot controller 24 is constituted by a computer, and is electrically connected to the drive motors of the robot arm 21 and the robot hand 22 so as to drive and control them. The controller 24 includes a communication unit 25 that communicates with the mass measuring device 10 and is connected to an operation unit (not shown) that starts and stops the robot.

  The robot controller 24 causes the robot arm 21 and the robot hand 22 to perform various operations by executing a built-in program. For example, when sorting it according to the mass of the article Q, the robot hand 22 goes to grab the article Q, and when it is grabbed, it is raised by the robot arm 21. The synchronization signal shown in FIG. 6 is transmitted from the communication means 25 to the control unit 30 immediately before the ascending process. When the control unit 30 receives the synchronization signal, it detects the maximum and minimum values of the filtered output waveform while sequentially inputting the outputs of the sensors 11 and 12, and based on the detected deviation between the maximum and minimum values. The mass of the article Q is obtained and the result is transmitted to the robot controller 24. The robot controller 24 determines and moves the sorting destination of the article Q based on the received mass.

Next, operations and operations when using the digital filter will be described.
First, the robot 20 and the mass measuring device are set to an initial mode, the robot 20 is put on standby for lifting an article, and the mass measuring device 10 is quantized to be output from the force sensor 11 or the acceleration sensor 12. Set the mode to capture the digital data as it is in the built-in storage area without filtering. When the robot 20 is started, the robot 20 outputs the synchronization signal and simultaneously lifts the article while accelerating the article. When receiving the synchronization signal from the robot 20, the control unit 30 of the mass measurement apparatus 10 sequentially takes in the quantized time series data of the force sensor 11 or the acceleration sensor 12. The data captured at that time is time-series data output during the ON time of the synchronization signal. This operation is repeated 10 to 20 times, for example, and time series data is stored in the built-in memory or the storage means 15 each time.

Next, when a “filter setting key” (not shown) is operated, the control unit 30 stores the time series data obtained by averaging the stored time series data in the storage unit 15, and then the detection unit 16 stores the time series data. A time T from when the series data takes the final extreme value to immediately before the speed change disappears is detected. Subsequently, the detection means 16 obtains the filter order (number of taps) defined by the detected time T from the following calculation formula.
Filter order = 2T / sampling period As a result, the filter order is determined, so that it is set in each of the digital filters 14 and 14 together with a predetermined filter coefficient. Then, each of the digital filters 14 and 14 becomes a linear phase filter having a phase delay time of T time, and filters the quantized data input thereafter.

  When the table 17 is provided, the detection unit 16 gives priority to the selection from the table 17. That is, the linear phase filter having the phase lag closest to the detected time T is read from the table 17 within a range not exceeding the detected time T, and the read linear phase filter is set in each of the digital filters 14 and 14. Thereby, each digital filter 14 and 14 becomes a linear phase filter which has the set phase delay time T, and filters the quantization data input after that.

  Further, in the initial mode, when the operation of taking the quantized time-series data of the force sensor 11 or the acceleration sensor 12 into the memory as it is is not performed, the acceleration / deceleration time of the synchronization signal transmitted from the robot 20 is ¼. A linear phase filter having a phase lag closest to the phase lag corresponding to that time is selected from the table 17 within a range not exceeding this time, and is set in each of the digital filters 14 and 14.

  When the above operation is completed, for confirmation, the robot system 100 is moved in the test mode, and an article having a known mass is actually lifted to measure the mass. If the result is sufficiently accurate, it can be determined that the phase delay time of the linear phase filter is appropriate, but if it is insufficient, high accuracy is achieved by the following operation.

  That is, the robot system 100 is returned to the initial mode, and the time series data before filtering stored in the storage unit 15 is displayed on the display unit 34 of the operation display unit 32 by screen operation. If the displayed time-series data is, for example, an output waveform fx as shown in FIG. 5, the optimum phase delay time T (TX) is calculated from the waveform, and the range not exceeding the calculated time T (TX). The designating means 33 designates a linear phase filter having a phase delay closest to that time. Then, the designated linear phase filter is read from the table 17 and the read linear phase filter is set in each of the digital filters 14 and 14. As a result, each of the digital filters 14 and 14 becomes a linear phase filter having the set phase delay time T, and filters the quantized data input thereafter.

  As mentioned above, although one Embodiment of this invention was described, this invention is not limited to this, Other embodiment is also employable. For example, in FIGS. 10 to 12, the storage unit 15, the detection unit 16, the table 17, and the like are provided in the control unit 30 of the mass measuring apparatus 10, but these are provided in the digital filters 14 and 14 that are also configured by computers. You can also.

DESCRIPTION OF SYMBOLS 10 Mass measuring device 11 Force sensor 12 Acceleration sensor 13a 1st A / D converter 13b 2nd A / D converter 14 Digital filter 15 Memory | storage means 16 Detection means 17 Calculation means 18 Table 20 Robot 21 Robot arm 22 Robot hand 24 Robot controller 25 Communication means 30 Control unit (processing means)
31 Communication Unit 32 Operation Display Unit 33 Designating Unit 34 Display Unit 100 Robot System

Claims (9)

  As the digital filter that filters the quantized time series data from the sensor that detects the force or acceleration acting on the article whose speed is changing, the time when the final extreme value appearing in the time series data during the speed change is taken A digital filter using a linear phase filter having a phase lag from the time until the speed change disappears immediately before.   The digital filter includes programmable processing means, and the processing means has storage means for storing the time-series data before filtering, and when the time is detected from the time-series data stored therein, 2. The linear phase filter having a phase delay corresponding to a detection time is set in the digital filter, and time series data to be quantized thereafter is filtered by the set linear phase filter. Digital filter. The digital filter according to claim 2, wherein the processing means includes a table storing a plurality of linear phase filters having different phase delays, and when the processing means detects the time, the time does not exceed the time, A linear phase filter with a phase lag closest to the time is read from the table and set in the digital filter, and the set linear phase filter is used to filter time-series data that is subsequently quantized.
Digital filter. A digital filter for filtering quantized time series data from a sensor that detects a force or acceleration acting on an article of varying speed comprises programmable processing means, said processing means comprising said processing prior to filtering Storage means for storing time-series data, a table for storing a plurality of linear phase filters having different phase delays, a specification means for designating any one of the linear phase filters stored in the table, and stored in the storage means Display means for displaying the time-series data before filtering.
From the time-series data displayed on the display means by the operator, the time from the time when the extreme value appearing in the time-series data is taken to the time immediately before the speed change disappears, and the time within the range not exceeding the calculated time is the most. When a linear phase filter having a near phase lag is designated by the designation means, the processing means reads the designated linear phase filter from the table and sets it as the digital filter. A digital filter characterized by filtering time-series data to be converted.   5. The digital filter according to claim 2, wherein the time-series data stored in the storage means is data obtained by averaging time-series data repeatedly input. A mass measuring device that measures the mass of an article from the force and acceleration acting on the article whose speed is changing,
A force sensor for detecting a force acting on the article;
An acceleration sensor for detecting an acceleration acting on the article;
First and second A / D converters for quantizing each sensor output;
First and second digital filters for filtering the outputs of these A / D converters, respectively.
The mass measuring apparatus according to any one of claims 1 to 5, wherein the first and second digital filters have the same phase characteristics. A mass measuring device that is incorporated in a robot that moves an article, measures the mass of the article from the force and acceleration acting on the article whose speed is changing, and transmits it to the robot while communicating with the robot,
A force sensor for detecting a force acting on the article;
An acceleration sensor for detecting an acceleration acting on the article;
First and second A / D converters that respectively quantize the outputs of the sensors;
A table storing a plurality of linear phase filters having different phase delays;
First and second digital filters for respectively filtering the first and second A / D converter outputs;
A controller for determining the mass of the article based on the output of these digital filters, and transmitting the determined mass to the robot,
When the control unit receives the synchronization signal including the acceleration / deceleration time of the article from the robot, the phase delay closest to the phase delay corresponding to the time is within a range not exceeding 1/4 of the acceleration / deceleration time. The linear phase filter is selected from the table, and the selected linear phase filter is set as the first and second digital filters.   The control unit, when receiving the synchronization signal, obtains the mass of the article based on a maximum value or a minimum value of each digital filter output or a deviation between the maximum value and the minimum value. The mass measuring device described. A robot hand that holds an article;
A robot arm for moving the robot hand;
A controller for driving and controlling the robot hand and the robot arm;
A robot with
The mass measuring device according to claim 7 or 8,
The force sensor of the mass measuring device is provided between the robot hand and the robot arm, and the acceleration sensor of the mass measuring device is a robot system provided in the robot hand.

Images