JPS6073326A - Calibration of sensor - Google Patents

Abstract

PURPOSE:To perform the calibration of a sensor simply and accurately, by adjusting the amplification factor of each detection gauge by successively adding constant external force to a position, where the max. moment of the indivisual detection gauge on a base plate is added, to converge the amplification factor of each detection gauge to a predetermined value. CONSTITUTION:Wt. providing positions P1-P4, where the max. moment is added to each of indivisual detection gauges Sa-Sd, are set to the outer frame 22 of a cross spring 13 supported through a support rod 21. After input resistance is regulated to perform the setting of the initial values of the detection gauges Sa -Sd, a wt. W is provided to the position P1 to adjust the amplification factor of Sa so as to make output equal to the wt. of the wt. W. The wt. W is successively transferred to P2-P4 to similarly adjust the amplification factors of Sb-Sd. By this method, the amplification factors of the detection gauges Sa-Sd are converged to predetermined values and calibration is performed simply and accurately.

Description

[Detailed Description of the Invention] (+> Technical Field of the Invention The present invention relates to a calibration method for a hinge,
In particular, it is effective for a sensor in which a plurality of detection gauges are attached to different parts of an elastically deformable substrate, and components of a plurality of dimensions of the envelope h are detected by the output from each detection gauge. Concerning improvements in grinning methods.

(2) Technology costs In recent years, industrial robots have been introduced onto product assembly lines in order to automate the assembly of various products. For example, as shown in figure v51, drag 81S4Δ1
Let us take as an example the case where the multi-axis robots 1 to R perform the fitting work of inserting the 7fi silk fsu8tS month 2 into the hole 3.
12 At the end of the day, the robot l-R holds the hand 5 that grips the part 1 of the assembly 4 at the tip of the farm 4. -U +3
After gripping the assembly member 1 with the hand 5, a predetermined feeding operation is performed to insert the assembly (=J4 member 11) into the hole 3 of the member 2. In case C, the hand 5 is rigidly fixed to the arm 4, and the above-mentioned phase A ζJ part 4A1 and the silk A' part tJ
The robot I
This results in the problem that the finishing precision of -R must be significantly improved.

In order to solve this problem, a mechanical compliance 1if is established between the arm 4 and the hand 5.
t horizontal C't! : nQ ke,) 7-mu 4 against han 1~
5 is spelled as elastically displaceable is Jij.
1, this combination plane 7 is connected to the connecting plate 10 connected to the arm 4, as shown in FIG. 2, for example, and this 3! I! Two sets of vertical substrates integrally formed on the connecting plate 10 and movable in mutually perpendicular X and Y directions, parallel plate springs 11 and 12, and parallel plate springs Ii and +2 hands 5. The hand 5 is connected to the hand 5 through a connecting rod 14 erected at the center of the cross spring 13.
It is designed to be connected and fixed to. Therefore, the hand 5 is moved in the X direction, Y direction, Z direction, and X direction with respect to the arm 4.
This allows for phase displacement around the axis and around the Y axis.

Since the compliance mechanism C, which is like a double series, is elastically displaced, C-only C is also utilized by detecting the displacement of a predetermined portion. That is, the fin 4S is connected to the cross spring 1 as shown in FIG. 2, for example.
Detection gauges Sa, Sb, Sc, Sd, Se such as Hesen gauges are attached to each month of 3 and parallel plate springs 11 and 12.
, SF is attached, and +3 (Puru Kabeku 1
It is designed to detect each component of ~le.

To explain the history in detail, as shown in Fig. 2, the moments acting on the detection gauges Sa to Sd are Ma, respectively.
MI+, Mc, Md Toshi, detection cage Se
, 3[acts on 4[-ment to Me.

If M[, then the following equation holds approximately.

Ma=a+F/-IIVIV --(1) fvlb-8
21-7-1-MX ・・・・・・・・・Old・・・ (
2) MC-#33F7 My ・＋＋＋ Old ・＋＋＋・・
++ (3) Ivlcl =a4Fz -Mx...
... Old ... (4) Me -= +1 FV
・・・・・・・・・・・・River (5) Ml=rnFx
...... Old... (6) (1, 1, Fx
, Fy, Fz are the forces in the X, Y, and 7 directions that are applied to the tip of hand 5, and Mx and My are the moments in the X and Y directions that are applied to the tip of hand 5, m.

"-1 is the distance between the center of the parallel leaf spring acting in the X and Y directions and the detection gauges Se and Sf, at, +2+83°a, l is the distance between the center of the cross spring 13 and the detection gauges sa to Stl (
·be. Further, the signs of the horns and T-men 1~ are positive in the direction of the arrow shown in the figure.

According to the above formulas (1) to (6), each component of Nonovec 1 to 1 is as follows.

Fx = Mf / m ・・・・・・・・・・・・・・・
... (7) FV = Me/r+ ...
・・・・・・・・・ (8 Fz = (Ma 1 M
tl +MCl-Md )/4 (at +a24-a
a + a4 )・・・・・・・・・・・・・・・・・・
(0)MX = (Mb - Md - "z (+2-
+4))/2 ・・・・・・・・・・・・・・・・・・
・ (10) fvlV = (Ma = Mc-FZ (
at-+3) )/2 ・・・・・・・・・・・・・・・
(11) In this way, using the outputs of the detection gauges 3a to S, it is possible to perform RiLt using the horns and moment I- in each direction at the tip of the hand 5, for example. A mechanism for correcting the tilting of the hand 5 is installed on the robot R side or the assembled member 2 side (2), and by feeding the signal from the sensor S to a varactor, the hand tilting is kept at almost zero for smooth mating operation. becomes possible.

By the way, the conversion equations (7) to (11) above are shown in FIG. 3, for example, as shown in FIG.
1' to R6', return resistances R1 to ↑R12, and other resistance elements of A can be easily constructed into an electric circuit. In this electric circuit, the mounting positions of the detection gauges S8 to S' are inaccurate and the detection gauge Sa
) Due to problems such as variations in the gauge factor of
Adjust the values of resistance Ri to R12, and calculate each detection 1 gauge S, I to =, t, S(
It becomes necessary to perform so-called sensor calibration to adjust the amplification factor of the sensor.

(1) Problems with conventional techniques
! , measure the output of each of these 114 detection gauges, and -
1) Using the data of γ-ta of j and the output 1 of the detection gauge, change the above formula (7) to (11) 1g
! From the formula, the amplification factor of each detection gauge was calculated to be 4 centimeters.

However, in such a conventional sensor kill replacement method, for example, the force F in the X-axis and Y-axis directions is
As shown in formulas (7) and (8), X and F'V are
Since it is detected by applying J to the output of one detection gauge, it is possible to perform sharp repression relatively easily. However, if we look at the force in the Z-axis direction, for example, the above equation (9) It is given as the sum of the output values of 1' detection gauges as shown in . For this reason, it is necessary to solve the simultaneous culm equation of the following equation (12) and determine the amplification factor of 3.1 for each detection gauge (J).

This IR, 16 pieces of data from M a i to Md1l are measured 81 times, and if the above equation (12) is solved, it will not be 4j, so it takes time to measure j'-t. In other words, if the amplification factor of each detection gauge was set in the same way as above, it would cause variations in resistance values and detection gauges. Detection errors etc. are complicatedly intertwined, and the iEll(t:1:17 replacement becomes difficult).

(4) The present invention according to Invention 1 is based on the viewpoint of
Its purpose is simple and simple.
01-cho ``Na1!Nri no I: t+ IJ expansion method l1ii (It's about being together.

(5) The 4t1 structure of the present invention is similar to the Ll tree structure of the present invention.
If we set in advance the external force application position that applies the maximum moment to each detection gauge, and then sequentially apply a known constant external force to each external force application position 1u, then in each case, C maximum heat 1- is applied. Adjust only the amplification factor of the detection gauge C, make the sensor output value match that corresponding to the external force mentioned above J: Sea urchin, repeat the above operation until the amplification factor of each detection gauge converges to the predetermined value 1 [Sensor Key I7 Replacement Method]

(6) Embodiments of the invention Hereinafter, the present invention will be described in detail based on embodiments with reference to the accompanying drawings.

As shown in FIG. 4, the sensor according to this embodiment has almost the same basic structure as that in FIG. 2, but it is different from that shown in FIG. The side surface of the hand is covered with a support plate 20, and a support 4 is provided in the center of the support plate.
The middle part of the cross spring 13 is supported via q=21, and C1
13. A λ-shaped outer frame 22 is integrally formed around this cross spring. Therefore, in the outer frame 22,
The individual detection gauges Sa to Sd are filled with famous people P-men 1~
4 is set in advance. Note that the outputs 1il' and f of the detection gauges 3a to 3Sf are input to the electric circuit as shown in FIG. 2 as in the conventional case (
There is.

Therefore, C1 In this example, C1 For example, force Fz in the Z-axis direction
When performing the key 17 replacement, follow the flow 17-1 to 17-1 shown in FIG. That is, 1. First, install Lonza in a predetermined position as shown in the figure, and 1. Manpower resistance R1 as shown in FIG.
') to R/l'
a to S (output 1 of 1 + i'JMa7ri to A of Md)] (
Adjust Pill-III. This stage (“Ma71
Initial value of J potato M [1 (1 "( will be set. After this, a weight W of known weight @ peg installation position P1: Q flT
l-! Jru. Konori? , adjust the amplification factor of the detection gauge 3a on which the maximum moment 1- acts (specifically, adjust the return resistance R3 in Fig. 2)
L'i to be equal to Toyo (WLJ).

Next, 1.1 11i1i W is purchased from 1be Pl to 1
] Transfer to 2 - (set "σ". At this time, dB in Fig. 2) %l Ring resistance R / l by CI4 l"S J 1, Shodai 7-men 1~ is detected. Adjust the amplification factor of gauge sb, II + I L; <Sensor output Fz is weight W
J: 1, which is equal to the weight of , adjust the amplification factor of the detection gauge SC over which the maximum moment 1 ~ acts,
The value of the force output Fz is made equal to the weight of the joist W. Move the screw W to the screw installation position P4, adjust the return resistance R6, and adjust the amplification factor of the detection gauge 3d in the same way as described above. Repeatedly for the plow position P1 to P4
Isa, set the joss W 5 times for each joss installation (i'/jA []1 to 1) 4, adjust the character width ratio of each detection gauge 3a to 3 (I, and set the joss W When the head is placed at any position on the outer frame 22, the sensor output FZ becomes equal to the weight of the body W (if it is, at this stage, ``Kil against l =
, end the represion. In this case, the detection gauge Sa7! The amplification factor of +〒Sd is focused on a predetermined liQ.

In order to confirm the convergence of this sharp regeneration method, we performed the following simulation.

For example, (when a weight of 1.2 kg is placed at the peg installation position P1, M y...-7 (kg, / cm), al" 7 = 1 (ku/Cn1) C4 = 'tz =
tlz=-4,7='J Toshi Ma-'+ 7-=8, lvl b・--1,1-0= ,-1,to4()...
−117=Ci.

fvld −−1−0=−1 is calculated as 111. In this way, each detection gauge S51no! Ji S (Determine the + output as shown in Table 1, and set j11 width ratio Q according to
From 1 to 4, the unity shown in Table 2 can be obtained. According to Table 2, 6 self-inclusions are obtained, and 1-11! Reflation can be converged in 12 iterations. ) 1. Also, if we consider that -C±1% is enough for one person, it is understood that the predetermined melon will be picked up by Mr. Senri on the fifth time.

Now, the position of the detection gauge and the balance of the gauge factor: 1: From W, the output values of the detection gauges Sa to Sd are respectively predetermined Ma: 120% Mb: 82% Me: '95% Md: Table 3 shows the w-flux process of the amplification factor when it is 88%.

In this case, the virtual output value of the detection gauge is determined as shown in Table 4. According to Table 3, C
There is. In addition, experiment C was 10km with 10V Norsthal.
Reaching the resolution of g in 3 iterations is 11.

Also, regarding the repulsion of the moment Mx around It will be done. In this case, I3 (, J, -f, set the known small joss W to the joss installation position P
'l buy. At this time, the maximum moment comes into play! J and the amplification factor of the detection gauge 3a (specifically, the force component F in the Z-axis direction
The amplification factor of z) is adjusted F b ' (the sensor force M
9 in key f! This is done by Du. Next, move the above-mentioned joss W to the joss installation position P2 1. At this time, 19i is shown in Figure 2.
Adjust the ring resistance [7] to adjust the amplification factor of the detection gauge sb on which the maximum moment is acting, and increase the sensor output M.
× is equal to ImW, J is equal to the acting heap, J
, move the sea urchin 1, and then move the weight W to the installation position [)3.1.
For this 11.1, please adjust the ring resistance R9.
Adjust the amplification factor of the detection gauge Sc (specifically, Z @ force direction force component [-7 amplification) on which the human being 1~ is acting, and set the sensor output Mx to Ol by 4T, Adjust the sea urchin. Furthermore, the 6th place of 'jrli' was established by Mr. W.
)4, in this case ll+7! Adjust the ring resistance [8] to maximize C.
Adjust the amplification factor of d, and add 11X to J by weight W so that the amplification factor of d becomes equal to the -ment.

Repeat this operation 5 times, and then! ! IiW outer frame 2
If the sensor output 1vlx reaches a predetermined value equal to the action 7-men 1- when the button is pressed, the calibration of the a3 C 7-ment Mx is completed at this stage. In this case, the detection gauge amplification factor is focused on a predetermined value.

Table 5 shows simulation results to show the convergence of the above-mentioned Ki-11 represion. According to Table 5, each amplification of the four Itashi Shimi 1C detection gauges! ? ! Q5. o6
．． G7 is shown to be Hayato1, in this case IJ
One virtual output of the detection gauges 3a to Sd [5 people 6].

In addition, the calibration of the seven-member l-M y around the Y axis is shown in Figures 6(a) and (b).
y-b means I, 'il A, on the hill (1 < f wa i'Muru's ζ-1 (2 (2, details of A < 2 will be omitted. In the experiment, 50 !+-CIll / 10V (1
) -7) It was confirmed that the Lt scale/l/Te5 u-cm resolution could be reached by four iterations.

Next, the second TI eggplant example shown in FIG. 7 will be explained.

The sensor S used in this example differs from the first embodiment in that positioning holes 25f and I11 are provided for positioning the weight W at the mounting positions P1 to 1)/l. The weight W has a positioning protrusion (Wi+) drilled into it. Therefore, when the calibration method of sensor 4 according to the present invention is carried out, the position of tilIW is fixed at a constant position for each operation. As a result, the installation error of the joist W is eliminated, and the amplification factor of each detection gauge is increased by 4.

The third embodiment will be explained based on FIGS. 8 and 9).

This embodiment differs from the 1-name embodiment in that it automatically adjusts the amplification factor of the valley detection gauge.
For example, there is the one shown in Fig. 8.The output vessel MHI to IMd of 7 detection gauges Sa to 3d
is inputted to the 8[) converter 27 via the 1G multiplex converter 26. The output from this A1) converter 27 is inputted to the input circuit 28. The microcontroller 28 controls the analog multiplier (+26) and the ΔD converter 27, respectively.The control system in this case is shown in FIG. 9, for example.
(according to). In FIG. 9, first, it is determined whether f+'Jq of Ken1 is correct or not, and then the initial value of each gauge output is determined to be, for example, X. I, XO2,
Microbe [ ] I? →
28. And valley detection gauges 3a to S
The initial value of the amplification factor of d is set to an arbitrary value, for example, 1, and the value of the amplification factor of the driver W is also set. After this, the spindle stage 1st row fffi P 17 to P4 -1 - weight distribution W
is placed, the amplification factor A of the valley detection gauges 3a to 3d
The tears of 1 to Δ4 are 1j. M'l' in this case
As C7j and 4, the following formula (13) is used.

A+ = (A2 X2-I A3 X3-A4 X4
4 (':) /X+A2 = (A+ X+ -A3
X3 +A4 X4-C) /X3 ・・・・・・・
...(13) A3-(△+Xl- to 2X21-A4 X
4-C) /x3△4=(-A+X+ 10△2X24-
A3 Is there a CI in this Danzan? Tree output! ifJ Cm (= A
+ X + -A 2 X 2Δ3X:l+△4X1)
Compare the small amount of actual weight W and if it is within a certain range, for example 5%, the final value is 51 J (transportation number △1 to A4
is stored in the microphone tree 28. For this reason, in this embodiment, C is automated when the key 17 reflexion Y1 operation is LI, and the microphone [lp]
The high-precision Kitribrage associated with the fi'i seagull of 28 lil (i!= becomes.Furthermore, it automatically restores Zhong W 1] Automatically refills to Bots 1 (if you have any inquiries) This enables a fully automated system for the work.

In addition, soil 5 [shi each fruit/+ili Il'll smell-(,
Specific details of the sensor 17.
Of course, there are some things that can be changed. According to the represion method, the calibration work for Ron 4 is simplified, and moreover, it is possible to perform accurate calibration that is not affected by variations in pants-1=-1 and guji-no-1 when the detection gauge is installed in the same position. Cgiru.

[Brief explanation of the drawing]

Fig. 1 is an explanatory diagram showing an example of [] box 1~ equipped with a cage 1, Fig. 2 is an explanatory diagram showing a specific example of a hitch, and Fig. 3 is an explanatory diagram showing a specific example of the
4 is an explanatory diagram showing an example of the electric circuit of the sensor; FIG. 4 is an explanatory diagram showing a specific example of the cash register 1 used when using the sensor key replacement method according to the present invention; The figures relate to the present invention: Il: Figures 6(a) and 6(b) show an example of the v-replication method.
is X axis I7; 1 moment!・Fig. 7 shows the procedure for performing one-point repression. An explanatory diagram showing another example of a tree used when using the replication h method,
FIG. 8 shows an example of a specific circuit for automating the adjustment of the amplification of each detection gauge in an IWl using the sensor replacement method according to the present invention.
Fig. 9 shows the procedure of the Kikame/Rebage method using the circuit shown in Fig. 8. Company 1: '5 gauge sea 5C-1J-・4-1 8-1-8-2-Fz's kiripre engine simulation I) (weight w-2oog) 3: Fz's kiripre engine simulation Millesin II (weight W・2oog) -4: Search for output gauge Sap to γ lG4 -1.
2 4.92. , -0.85 -7.004-15=!
Kiribrejin's Similane (weight - 20
0g) (Momenoto M-7000g-C)n g(
5) g(f() g(7) 1 1 1 1 2 -750.594 4.06977 -91.32
43-75s, 734.11774 -82.7665
4 -759.873 4.11849 -92.78
915 -759.875 4.1185 -82.7
8946-758.875 4.1185 -92.7
8947 -759.875 4.1185 -92.
78948 -7511.875 4.1185 -8
2.78949 -7511.875 4.1185
-! 112.789410-759.875 4.11
85 -92.789411 -759.875 4.
1185 -92.789412 -758.875
'4.1185 -92.789413 -75L87
5 4.1185 -92°788414 -759.
875 4.1185 -92.789415 -75
9.875 4.1185 -! 32.78946:
Output gauge Ki Anao's provisional -111 Output value P Ma Mb
Mc Md 1 -9.6 -0.82 5.7 -0.882 -
1.2 -6.5Ei -0,955,2837・2
-0.82 ”7.8 -0.884 -1.2 4.
92 -0.85 -7.04th M Figure 2 Figure 2 Figure 7 Figure 8 Day 2

Claims (3)

[Claims] (1) For a sensor in which a plurality of detection gauges are attached to different parts of an elastically deformable substrate, and the components of a multidimensional curve are detected by the output from each detection gauge. , the amplification factor of each detection gauge is +, 1. 'I'll arrange it. 1: When performing I+replication, positions where external force is applied to each detection gauge at a maximum of 1 are set in advance on the board, and a known constant external force is sequentially applied to the external force application positions. At the same time, adjust the force of the amplification factor of the detection gauge, which applies a maximum of 7 mm each time the above external force is applied, so that the output of the sensor gauge matches the value corresponding to the above external force. sensor calibration method 1, characterized in that the above operations are repeated until the amplification factor of each detection gauge converges to a predetermined value of 1 iri. (2) The method of calibrating a sensor according to claim 1, characterized in that, when a known constant external force is applied to the external force application position of the substrate, the point of application of the external force is positioned. (3) The Ron 1 no calibration method according to claim 1, wherein the adjustment of the amplification factor of each detection gauge is controlled by a microprocessor.