JP2003094367A - Robot hand with tip visual sense - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a hand capable of autonomously correcting position of a work grip and gripping a work without receiving control of a robot controller. SOLUTION: A work imaging means 201 is provided on the hand 103 and an image of a gripped work is picked up. A work position and shape detecting means 202 calculates data of a position and shape of the work from picked up image data of the work. A gripping position data generating means 203 generates gripping position data by the hand 103 from the calculated data of the position and shape. A hand gripping position control means 204 controls movement of fingers 206 to a position of gripping the work on the basis of the gripping position data. Each means is provided in the hand 103, and the hand 103 autonomously grips the work at an optimum gripping position.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a robot hand used for assembling and disassembling products, and more particularly, to a robot hand with a visual tip capable of autonomously grasping a work accurately.

[0002]

2. Description of the Related Art A robot hand is used for assembling and disassembling products on a production line or the like. As this robot hand, there is a visual robot hand having a function of visually detecting a product at the hand of the robot. The robot hand having such a visual input function has various forms depending on transmission / reception of visual information and control.

A first conventional robot hand with a visual sense (disclosed in Japanese Patent Laid-Open No. 6-206186) adds a handling position and tool axis offset information to the position data of a work obtained from a visual sensor to accurately measure the robot. It realizes various handling work.

The second conventional robot hand with vision (disclosed in Japanese Patent No. 2679614) relates to the handling of a self-propelled robot from the image obtained by the image pickup means attached to the hand portion, and the position and posture of the robot. Is configured to correct.

FIG. 13 is a block diagram of a control system in a third conventional robot hand with vision (disclosed in Japanese Patent No. 2767417). The image data captured by the camera 1301 attached to the tip of the arm of the robot body 1300 is used in the vision system 1 for performing processing calculation.
The image data is output to 302, the work target is recognized by the vision system 1302, and the image data of the reference point is processed. The robot control unit 1303 corrects and controls the position of the end effector 1304 provided at the arm tip of the robot body 1300 using the processing result.

[0006]

However, in the above-mentioned conventional robot having the visual input means, the visual information of the work is sent to the robot controller outside the hand and processed. Therefore, in order to correct the position of the hand, communication between the hand and the robot controller such as a signal for position correction and correction data is required each time the workpiece is gripped, which complicates the control and limits the communication speed. There was a problem that the tact time could not be improved.

Further, the robot program also needs a description of a processing sequence corresponding to the position correction, and this processing sequence is executed every time the workpiece is gripped, which imposes a control burden.

Further, in recent years, it has become possible to perform teaching work of a robot only to a robot controller without actually operating the robot by off-line teaching or the like by computer simulation. However, in reality, the positioning accuracy cannot be accurately obtained due to the machine difference between the robots and the deflection of the robot body. Therefore, it is necessary to actually operate the hand to correct the fine teaching on the robot controller side. This work hinders the introduction of a large number of robots into high-mix low-volume production.

In order to solve the above-mentioned problems of the prior art, the present invention provides a robot hand with a visual tip capable of autonomously correcting the position of the work grip by the hand and gripping the work without the control of the robot controller. The purpose is to provide.

[0010]

[Means for Solving the Problems]
In order to achieve the object, a robot hand with a visual tip according to the invention of claim 1 is a work imaging means for obtaining image data representing a shape and a position of a work grasped by the robot hand, and the work imaging means images the work. Based on the image data of the work, the work position shape detection means for calculating the position and shape data of the work, and for grasping the work by the robot hand based on the position and shape data calculated by the work position shape detection means The robot hand is provided with a grip position data generation unit that generates grip position data and a hand grip position control unit that changes the grip position of the robot hand based on the grip position data generated by the grip position data generation unit. It is characterized in that the work can be gripped at the optimum gripping position.

According to the first aspect of the present invention, since it is possible to find the gripping position of the work in the robot hand and grip the work, precise control is not received only by the opening / closing operation command from the robot controller. Since the work can be gripped autonomously, the takt time can be improved and the teaching work for gripping can be reduced.

Further, in the robot hand with a hand tip according to a second aspect of the invention, in the invention according to the first aspect, the work position shape detecting means accumulates data relating to the shape of the work and the gripping position. An internal database is provided, and the gripping position of the work is obtained based on the calculated work shape data by referring to the internal database.

According to the second aspect of the present invention, various work pieces can be gripped by obtaining work shape data using the internal database.

Further, a robot hand with a visual tip according to a third aspect of the present invention is the external database according to the second aspect of the invention, which is provided outside the robot hand and stores data relating to the shape and gripping position of the work. When,
The work position / shape detecting means is provided with a communication means capable of transmitting / receiving data to / from the external database, and when the work position shape detecting means does not match the shape data of the corresponding work in the internal database, ,
It is characterized in that a gripping position corresponding to the corresponding work shape data is obtained from the external database.

According to the third aspect of the present invention, when there is no match with the shape data of the held work in the internal database, the external database is queried to program for a new type of work. Gripping is possible without the need for changes or intentional changes to the database.

Further, the robot hand with a hand tip according to the invention of claim 4 is the robot hand according to claim 3, wherein the work position shape detecting means obtains a work as a result of inquiring from an external database. Shape data,
It is characterized in that it is additionally stored in the internal database, and thereafter the shape data of the corresponding work is obtained using the internal database.

According to the invention described in claim 4, by adding a new type of work shape data from the external database to the internal database, the gripping process can be executed in the hand thereafter without inquiring the work externally. , Tact time can be improved by minimizing communication with the external database.

According to a fifth aspect of the present invention, there is provided a robot hand with a visual tip according to the first aspect, wherein the gripping position data generating means is the position and shape calculated by the work position shape detecting means. The grip position for each shape of the work is calculated based on the data of 1.

According to the invention described in claim 5, a characteristic shape such as a circle, a parallel shape, or a bar shape of the work is recognized from the shape data calculated by imaging the work, and the internal gripping is performed corresponding to this shape. It is possible to grasp various gripping positions such as external gripping and grip various works in an optimum state.

According to the sixth aspect of the present invention, in the robot hand with a fingertip visualizing device according to the first aspect of the present invention, the work imaging means comprises a two-dimensional CCD sensor or a two-dimensional CMOS sensor. It is characterized by

According to the sixth aspect of the invention, since the work is imaged by using the two-dimensional CCD or the two-dimensional CMOS sensor, the work imaging means can be downsized.

According to a seventh aspect of the present invention, there is provided a robot hand with a visual tip in the first aspect of the invention, wherein the work imaging means has at least two two-dimensional CCD sensors or two-dimensional CMOS sensors. And is for imaging the workpiece from different angles,
The work position / shape detecting means is characterized in that the shape, position, posture, and the like of the work are three-dimensionally obtained based on two or more image data obtained by the work imaging means.

According to the invention described in claim 7, the two sensors image the work in three dimensions,
The shape, position, and posture of the work can be obtained three-dimensionally, the shape of the work can be recognized regardless of the type of the work, and the work piece can be stably gripped in the optimum posture.

Further, a robot hand with a fingertip visual sense according to the invention of claim 8 is, in the invention of claim 6, provided with a pattern illuminating means for illuminating a workpiece with pattern, and the workpiece position / shape detecting means: It is characterized in that the work shape, position, and orientation are three-dimensionally obtained based on the image data of the work imaging means obtained by the pattern illumination.

According to the invention described in claim 8, it is possible to three-dimensionally determine the shape and position / orientation of the work by using the spatial coding method, the moire method, or the like by the pattern illumination, and more stable. Then, the work can be recognized, and the work can be gripped in an optimum posture.

According to the ninth aspect of the present invention, in the robot hand with a fingertip vision according to the first aspect of the invention, a laser range finder is used as the work imaging means, and the work position shape detection is performed. The means is characterized in that the shape, position, and orientation of the work are three-dimensionally determined based on the image data of the work imaging means.

According to the invention described in claim 9, the shape, position and posture of the work can be three-dimensionally determined by the laser range finder, and the work can be recognized more stably and gripped in the optimum posture. it can.

Further, the robot hand with a visual tip according to the invention of claim 10 is the robot hand according to claim 1, wherein the shape of the work calculated by the work shape position detecting means or the work based on the shape is calculated. When storing the work on the work to be assembled and storing the type, based on the image data of the work to be assembled and the shape of the work stored in the storage, the work on the work to be assembled And a work assembling means for determining an optimum work assembling position and assembling the work.

According to the tenth aspect of the present invention, by storing the shape or type of the gripped work, when the work is assembled to the assembly target work,
Since the assembly position can be determined from the shape of the workpiece to be assembled, the teaching load applied during assembly can be reduced.

According to the eleventh aspect of the present invention, the robot hand with a fingertip visual inspection according to the first aspect of the present invention is provided with a communication means capable of transmitting and receiving data to and from the robot controller. The generation means is
When it is detected that the generated grip position data is out of the operation range that can be gripped by the robot hand,
It is characterized in that grip position data indicating that it is out of the operation range is output to the robot controller via the communication means, and the position correction is executed by the robot controller.

According to the eleventh aspect of the present invention, when it is determined that the work is out of the operation range in which the robot hand can hold, a wide range of position correction is performed in order to execute the position correction by the robot controller. Will be able to.

According to the twelfth aspect of the present invention, the hand-visualized robot hand according to the second aspect is provided with a communication means capable of transmitting and receiving data to and from the robot controller, and the work position shape. The detection means is
It is characterized in that the work shape corresponding to the gripped work is obtained from the internal database, and the work shape or the kind of work corresponding to the work shape is output to the robot controller via the communication means.

According to the twelfth aspect of the present invention, since the type of the gripped work is output to the robot controller, even if a large number of mixed works are sequentially assembled, the gripped work is attached at the assembly position of the next process. It is possible to improve the efficiency of assembling such as moving to.

[0034]

BEST MODE FOR CARRYING OUT THE INVENTION Preferred embodiments of the robot hand with fingertips according to the present invention will be described in detail below with reference to the accompanying drawings.

FIG. 1 is a diagram showing an overall configuration of a robot hand with a fingertip visual guide applied to an embodiment of the present invention. The robot 100 has a plurality of arms 101 and a plurality of joints 102.
The arm 103 is provided with a hand 103 for gripping a work.

The robot controller 110 includes the arm 1
01 The operation of the entire robot 100 including the hand 103 is controlled and controlled so that the work can be gripped by the hand 103 provided at the tip. This robot controller 110 is different from the hand 103 in the tip of the arm 101 (hand 103).
After positioning, the gripping signal described later is sent to start the gripping operation.

FIG. 2 is a block diagram showing the configuration of the hand 103. The hand 103 has a plurality of fingers 206, for example, a pair of fingers 206 as shown in the figure, at the tip thereof, and the stage 20
The workpiece can be gripped by sliding the pair of fingers 206 so as to change the distance between them.
The work to be gripped and used by the hand 103 part is a work imaging means 20 having an imaging function such as a lens and CCD.
1 is output as image data.

The work position / shape detecting means 202 performs image processing on this image data to extract a work part, and detects the shape and position of the work. The grip position data generation means 203 calculates the handling position of the work based on the detected shape and position of the work. (hand)
The gripping position control unit 204 controls the driving so that the pair of fingers 206 grips the work based on the calculated handling position. This hand gripping position control means 2
04 is, for example, a stepping motor or a servo motor,
Alternatively, the positioning of the finger 206 is controlled by an air pressure control method using an electropneumatic regulator.

The hand 103 has a deviation correction function for accurately gripping the work, and is configured to detect the shape and position of the work and grip the work at an optimum position. Therefore, the hand 103 does not need to send and receive the image data regarding the grip of the work and the information such as the work shape and the position with the robot controller 110.

Hand 103 and robot controller 11
0 is connected via the communication means 205, and the hand 103
Starts a gripping operation by receiving a gripping preparation signal, a gripping signal, etc. from the robot controller 110, and also sends a reply signal to the gripping preparation signal, a gripping signal, a work presence / absence signal, etc. to the robot controller 110. Then, it operates in cooperation with the robot controller 110.

FIG. 3 is a flow chart showing the control contents of the work holding by the hand 103. In the figure, the control content for gripping the work without performing the processing for receiving the grip preparation signal from the robot controller 110 is described. First, the hand 103 includes the robot controller 1
It waits for the grip signal from 10 (step S301).
When the grip signal is received (step S301: Yes),
The work image forming means 201 converts the work into image data and takes it into the hand 103 (step S302).

Next, this image data is subjected to image processing by the work position shape detecting means 202 to extract the work part, and the shape and position of the work are recognized (step S30).
3). Then, based on the recognized shape and position of the work, the grip position data generation unit 203 detects the amount of deviation between the current position of the work and the current position of the finger 206 (step S304), and based on this amount of deviation. A hand gripping position required for handling is calculated (step S3).
05).

Then, based on the calculated handling position, the hand gripping position control means 204 causes the finger 2
Drive control of 06 to grip the work (step S3
06). While the workpiece is being gripped, a gripping signal (gripping completion signal) is output to the robot controller 110 (step S307).

As described above, the hand 103 operates only by waiting for the input of the gripping signal from the robot controller 110, and only after sending the gripping completion signal to the robot controller 110 after gripping, the work can be accurately gripped. . As a result, the hand 103 can autonomously grip the work at the optimum gripping position even if the accuracy of teaching to the robot controller 110 is low.

The hand 103 is the robot controller 1
When the above-mentioned gripping signal is received from 10, the shape of the work may not be discriminated due to the shape of the work to be gripped, the shape of the hand (finger 206), or the like.

FIG. 4 is a flow chart showing another control content of the work gripping by the hand corresponding to this case. In this case, the grip position is detected in advance at the input timing of the grip preparation signal transmitted from the robot controller 110. The hand 103 first waits for a grip preparation signal from the robot controller 110 (step S401). At this time, the finger 206 is controlled so as to be at a position offset from the gripping position of the work in the normal direction of the upper surface of the gripping work.

When the grip preparation signal is received (step S401: Yes), the work image forming means 201 converts the work into image data and takes it into the hand 103 (step S402).

Next, this image data is subjected to image processing by the work position / shape detecting means 202 to extract a work part, and the shape and position of the work are recognized (step S40).
3). After that, based on the recognized shape and position of the work, the gripping position data generation unit 203 detects a deviation amount between the current position of the work and the current position of the finger 206 (step S404), and based on this deviation amount. A hand gripping position required for handling is calculated (step S4).
05).

Then, a reply signal for replying to the grip preparation signal is sent to the robot controller 110 (step S406). Next, the robot controller 110 waits for a grip signal from the robot controller 110 (step S407). When the gripping signal is received (step S407: Yes), the hand gripping position control means 204 drives and controls the plurality of fingers 206 based on the calculated handling position to grip the work (step S408). While the workpiece is being gripped, a gripping signal (gripping completion signal) is output to the robot controller 110 (step S409).

As described above, the hand 103 can detect the gripping position in advance by using the gripping preparation signal transmitted from the robot controller 110, so that the shape of the workpiece to be gripped cannot be determined. However, after the hand 103 autonomously recognizes the shape of the work, the work can be gripped at the optimum gripping position.

FIG. 5 is a block diagram showing another configuration example of the hand 103. The work position shape detecting means 20 described above.
Reference numeral 2 has an internal database (DB) 501 in which work data representing the work shape and position for each work (for example, the center of gravity position, outer shape, circle, rectangle or reference image) is registered and stored in advance. It can be configured.
Every time the image data of the work is obtained from the work imaging unit 201, the internal database (DB) 501 is inquired of the work data, and the data regarding the gripping position of the work is obtained based on the work data corresponding to the recognized shape of the work. Then, the position of this work is output to the grip position data generation means 203. As a result, the hand 103
Even if the type of workpiece to be gripped is different, it is possible to grip at the optimum position.

Further, as shown in FIG. 5, an external database (DB) 502 may be provided outside the hand 103. Internal database (DB) 5 of hand 103
If 01 is searched and the work to be grasped does not match the data registered in the internal database (DB) 501, communication is performed with the external database (DB) 502 provided outside the hand 103. Means 503
It may be configured to make an inquiry via.

As a result, it becomes possible to obtain data relating to the shape and position of the unknown work. In addition, the robot 100 including the hand 103 described in the present embodiment.
When a work is added after the introduction, the new work can be grasped at the optimum position without the need for changing the configuration and programming of the control system (each of the above means) of the hand 103 and the database.

An external database (DB) 502
For Ethernet (R) and RS-232
Inquiries can be made via communication means 503 such as C and LAN. External database (DB) 50
2 is SQL (Structured Query La)
It is possible to use a server-client type database using a language such as "nguage". In the example shown in FIG. 5, the external database (DB) 502 is provided outside the robot controller 110. However, the external database (D) 502 is provided inside the robot controller 110.
B) 502 may be provided. In this case, the external database (DB) 502 can also be configured to send and receive data via the communication means 205.

Work data obtained by making a new inquiry to the external database (DB) 502 is the hand 103
May be added to the internal database (DB) 501. As a result, the work data relating to this new work can be held inside the hand 103, and thereafter, the work data relating to this new work can be queried and processed by the internal database (DB) 501 of the hand 103, and therefore the communication means 503. Since the communication via the terminal is unnecessary, the time required for the communication can be shortened and the control speed can be increased.

Further, only when a new work or a new hand is added, the internal database (D
B) It is also possible to construct 501. According to this configuration, even when the robot 100 is first introduced, the internal database (DB) 501 of the hand 103 can be taken out from the external database (DB) 502, and it is not necessary to specially make it inside the hand 103. The cost of introduction can be reduced.

Further, a database (DB) 501 inside the hand 103 and an external database (DB) 50.
It is possible to adopt a configuration in which the grip position is calculated by recognizing, based on the image data, whether the location of the workpiece to be gripped is circular or parallel based on the image data.

FIG. 6 is a flow chart showing the procedure for calculating the gripping position based on the recognition of the work shape. The image data picked up by the work imaging unit 201 is input to the work position shape detection unit 202 (step S601). The work position shape detection means 202 cuts out an image of the work part from the image data (step S602).

Next, it is recognized whether the cut-out work shape is circular or parallel (step S60).
3). If the recognition result is a circular shape (step S6)
(03: circular shape), and then it is determined whether the gripping position is outside or inside (step S604). Any one of the grip positions is set in advance. Not limited to this,
It can also be determined based on the obtained outer diameter and inner diameter of the work, the size of the finger 206 of the hand 103, and the amount of movement that can be grasped.

When the gripping position is outside (step S604: outside), the outer diameter of the work is calculated (step S605). Next, the gripping position data generating means 203
Determines the center position of the work based on the obtained outer diameter position of the work, and calculates the current center position of the work and the finger 2
A deviation amount of the current center position of 06 is detected (step S
606). A hand gripping position required for handling with each finger 206 is calculated based on this shift amount (step S607). Then, the calculated handling position is output to the hand gripping position control means 204 (step S
608), and the work is gripped by the fingers 206.

When the grip position is inside (step S604: inside), the inner diameter of the work is calculated (step S609). Next, the gripping position data generating means 203
Calculates the center position of the work based on the obtained inner diameter position of the work, and calculates the current center position of the work and the finger 2
The deviation amount of the current center position of 06 is detected (calculated) (step S610). Each finger 2 based on this deviation amount
A hand gripping position required for handling in 06 is calculated (step S611). Then, the calculated handling position is output to the hand gripping position control means 204 (step S608), and the finger 206 grips the work.

If the recognition result in step S603 is a parallel shape (step S603: parallel shape), it is next determined whether the gripping position is outside or inside this work (step S612).

When the grip position is on the outside (step S612: outside), the width between the outside of the work (grip width) is calculated (step S613). Next, the grip position data generation unit 203 obtains the center position of the work based on the obtained width of the work, and detects (calculates) the deviation amount between the current center position of the work and the current center position of the finger 206.
(Step S614). A hand gripping position required for handling with each finger 206 is calculated based on this shift amount (step S615). Then, the calculated handling position is output to the hand gripping position control means 204 (step S608), and the finger 206 grips the work.

If the gripping position is inside (step S612: inside), the width between the insides of the workpiece (grasping width) is calculated (step S616). Next, the grip position data generation unit 203 obtains the center position of the work based on the obtained width of the work, and detects (calculates) the deviation amount between the current center position of the work and the current center position of the finger 206.
Yes (step S617). A hand gripping position required for handling with each finger 206 is calculated based on this displacement amount (step S618). Then, the calculated handling position is output to the hand gripping position control means 204 (step S608), and the finger 206 grips the work.

FIG. 7 is a front view showing a specific example of the configuration of the hand 103, and FIG. 8 is a side view of the same. This hand is a form-changing hand 700, and has a plurality of (three in the illustrated example) fingers 206 having redundancy. A circular stage 701 is provided at the tip of the shape-changing hand 700, and the central portion of the stage 701 is provided with a base shaft 702.
As a, three arms 702 that are rotatable in the circumferential direction are provided. Each arm 702 has a finger 20
6, the fingers 206 are movable in the axial direction of the arm 702 (the expansion / contraction direction of the fingers 206) by the slide mechanism 703 of the arm 702.

By using the shape-changing hand 700 as described above, the three fingers 206 can be respectively rotated to change the mutual angles with respect to the center position and can be expanded / contracted. Any shaped work can be gripped.

As another example of the configuration of the hand 103, even a multi-fingered hand simulating a human hand can be corrected and grasped in the same manner. In the case where the robot 100 grasps only a shape in which the parallel portion can be grasped by two fingers, it is a center if a mechanism in which each of the two fingers can be independently driven. Since the position can be moved, a parallel work can be gripped.

FIG. 9 is a view for explaining the gripping position of the work by the above-mentioned shape change hand 700. As described above, each of the fingers 206 can be pivoted about the axis of the base 702a of the stage 701 by the arm 702, and can be independently driven in the radial direction of the stage 701 by the slide mechanism 703. ing.

Therefore, the gripping position data generating means 20
First, from the obtained image data, 3 is the center position of the hand 103, that is, the center positions of the three fingers 206 (black circles in the figure) P1, P2, P3, and the center positions of the work W (white circles in the figure). After each S is obtained, a virtual straight line L passing through the center position P1 of the reference finger 206 and the center position S of the work W is drawn.

Next, the work W intersecting the virtual straight line L
The reference gripping position F of the finger 206 with the periphery of the reference
Set to 1. With reference to this reference gripping position F1, the work W is divided into three parts every 120 degrees on the outer diameter (periphery), and the other two points on the circumference of the obtained work W are separated by the other fingers 206.
Grip positions F2 and F3.

The hand gripping position control means 204 rotates the arm 702 and linearly moves the slide mechanism 703 so that the fingers 206 are respectively positioned at the gripping positions F1 to F3 obtained. As a result, the hand 103 can stably grip the work W with the corresponding three fingers 206 at an angular position obtained by dividing the work W into three equal parts.

The work imaging means 201 is a two-dimensional C
A CD sensor and a two-dimensional CMOS sensor can be used to obtain two-dimensional image data of the work. Also, two-dimensional C
Equipped with two CD sensors or two-dimensional CMOS sensors,
It is also possible to adopt a configuration in which the work position shape detecting means 202 uses a stereo method to three-dimensionally obtain the shape, position, and orientation of the work.

Further, the hand 103 is provided with a pattern illuminating means (not shown), and based on the image data of the work obtained by pattern illuminating the work, the work position / shape detecting means 202 uses a space coding method, a moire method or the like. The shape, position, and orientation of the workpiece can be obtained three-dimensionally. Further, a laser range finder can be used as the work image forming means 201, and the work position shape detecting means 202 can three-dimensionally determine the shape, position, and orientation of the work.

In this way, the work imaging means 201
Obtaining the three-dimensional image data, the work position shape detecting means 202
By calculating the shape, position, and orientation of the work in three dimensions, the gripping point of the work having a more complicated shape can be detected. For example, the work shown in FIG. 10 and the work shown in FIG. 11 will be described as an example. Work 1001 shown in FIG.
Is a cylindrical convex portion 10 at the center of a square base 1002.
It is equipped with 03. A work 1101 shown in FIG. 11 is provided with a circular recess 1103 in the center of a square base 1102.

It is impossible to judge whether these workpieces 1001 and 1101 have a convex shape or a concave shape at the center only by one-dimensionally capturing an image from the plane (upper surface) direction. However, it is possible to determine the unevenness by obtaining three-dimensional image data of the works 1001 and 1101 by taking an image from the upper surface and the side surface of the works 1001 and 1101.
As a result, it becomes possible to calculate a grip point that can be gripped by the hand 103. The work 1001 shown in FIG.
02 or the periphery of the convex portion 1003 can be gripped, and the work 1101 shown in FIG.
It becomes possible to determine that the peripheral edge of the concave portion 1103 or the peripheral edge of the concave portion 1103 can be grasped.

Further, the hand 103 described in the embodiment of the present invention can be configured to have a storage means (not shown) for storing the shape or type of the work to be held. The work position shape detection means 202 stores the shape of the gripped work or the type corresponding to the shape in the storage means.
Then, the work assembling means (not shown) is suitable for the shape of the work stored in the storage means based on the image data of the work to be assembled which is picked up by the work imaging means when assembling the gripped work. The assembly position can be determined and the assembly can be performed.

For example, the workpiece 12 to be assembled shown in FIG.
When assembling the work W gripped by the hand 103 with respect to 01, the shape, size, type, etc. of the gripped work W are stored, so that the work W is attached to the mounted work 1201. It is possible to automatically recognize the position and assemble. In the illustrated example, the work 1201 to be assembled has three mounting positions 1202, 1203, and 1204. However, based on the shape of the work W, the mounting position 1203 is suitable for the shape of the work W. Will be able to.

Further, the robot 100 described above has a structure capable of mutually performing data communication with the robot controller 110 via the communication means 205 (see FIG. 2).
The grip position data generation means 203 determines this when the calculated grip position data is at a position beyond the operating range of the hand 103 (finger 206) that can be controlled by the hand grip position control means 204. Whether or not the hand 103 is included in the operating range can be easily detected by using a preset threshold value. Then, the grip position data indicating that gripping is impossible may be output to the robot controller 110 via the communication unit 205.

As a result, when the allowable work position of the hand 103 exceeds the operation range, the grip position data is output to the robot controller 110. Accordingly, the hand 103 can request the robot controller 110 to perform correction control and cause the robot 100 to perform position correction when the large displacement occurs.
This makes it possible to deal with the case where the positional accuracy of the work is extremely poor.

As described above, in the present embodiment, the hand 10
Only when the correction cannot be performed in step 3, the grip position data is output to the robot controller 110 via the communication unit 205, so that the control load on the robot controller 110 can be reduced. Further, the request for correction control to the robot controller 110 when such a gripping impossible state is generated does not occur every time the workpiece is gripped. Therefore, as in the prior art, the robot controller 110
In comparison with a configuration in which a similar positioning request is made each time and communication control is performed, the effect on tact time is low,
High throughput can be obtained at low cost.

Further, the hand 10 is connected via the communication means 205 between the hand 103 and the robot controller 110.
The work shape obtained by the work position shape detecting means 202 of No. 3 can be output to the robot controller 110. If the hand 103 is configured to include the internal database (DB) 501, the type of work can be output based on the shape of the work.

As a result, the robot controller 11
With 0, the work held by the hand 103 can be discriminated, and the hand 103 can be moved to a different assembly position for each work. Therefore, even if a plurality of types of works are mixed, it is possible to sequentially control the movement of each work gripped by the hand 103 to the corresponding mounting position on the work 1201 to be mounted, thereby improving versatility. On the contrary, the robot controller 1
By sending information such as the shape and type of the work to be grasped from the hand 10 to the hand 103,
Allows the target work to be taken out selectively from many kinds of mixed work.

The method for gripping a work described in the present embodiment can be realized by executing a program prepared in advance on a computer such as a personal computer or a workstation. This program is recorded on a computer-readable recording medium such as a hard disk, a floppy (R) disk, a CD-ROM, an MO, or a DVD, and is executed by being read from the recording medium by the computer. The program can be distributed via the recording medium and a network such as the Internet.

[0084]

As described above, according to the invention described in claim 1, the work imaging means for obtaining the image data representing the shape and position of the work gripped by the robot hand, and the work imaging means are provided. A work position shape detecting means for calculating the position and shape data of the work based on the image data of the imaged work, and a robot hand grips the work based on the position and shape data calculated by the work position shape detecting means. And a hand gripping position control means for changing the gripping position of the robot hand based on the gripping position data generated by the gripping position data generating means. Since the robot can autonomously grip the work at the optimum gripping position, the work gripping position is calculated in the robot hand. Since it is possible to grasp the workpiece, it becomes possible to autonomously grasp the workpiece without receiving precise control from the robot controller via communication, etc., improving the takt time and reducing teaching work for grasping. There is an effect that can be.

According to a second aspect of the present invention, in the first aspect of the invention, the work position shape detecting means has an internal database for accumulating data relating to the shape and gripping position of the work, Since the grip position of the work is obtained by referring to the internal database based on the calculated work shape data, the autonomy by the hand is further improved, and various works can be gripped using the internal database.

According to the invention of claim 3, in the invention of claim 2, an external database provided outside the robot hand for accumulating data relating to the shape and gripping position of the work, and the external database. Communication means capable of transmitting and receiving data to and from the work position and shape detecting means, when there is no work matching the shape data of the corresponding work in the internal database for the work to be gripped, Since the grip position corresponding to the relevant work shape data is obtained from the database, the number of inquiries to the external database can be reduced as much as possible, and restrictions on communication etc. can be suppressed, and programming for new types of work is also possible. It is possible to grasp without requiring a change or an intentional change of the database.

Further, according to the invention described in claim 4, in the invention described in claim 3, the work position shape detecting means stores the work shape data obtained as a result of inquiring from an external database, Since it is additionally stored in the internal database and the shape data of the corresponding work is obtained using the internal database, if a new type of work shape data is added from the external database to the internal database, this work will be inquired to the outside. It is possible to perform the gripping process in the hand without the need for communication, minimize the communication with the external database, and improve the tact time.

According to the invention of claim 5, in the invention of claim 1, the gripping position data generating means is based on the position and shape data calculated by the work position shape detecting means. In order to calculate the gripping position for each shape of the work, from the shape data calculated by imaging the work,
Recognize characteristic shapes such as circular, parallel, and bar-shaped workpieces,
There is an effect that various gripping positions such as internal gripping and external gripping are determined according to this shape, and various works can be gripped in an optimum state.

According to a sixth aspect of the present invention, in the first aspect of the present invention, the work image forming means is composed of a two-dimensional CCD sensor or a two-dimensional CMOS sensor. This has the effect that the size can be reduced.

According to a seventh aspect of the present invention, in the first aspect of the present invention, the work imaging means is provided with at least two two-dimensional CCD sensors or two-dimensional CMOS sensors, each having a different angle. To image the work, and the work position shape detecting means three-dimensionally obtains the shape, position, posture, etc. of the work based on two or more image data by the work imaging means. Thus, the shape of the work can be recognized regardless of the type of the work, and it is possible to stably grip the work in an optimum posture.

According to the invention described in claim 8, in the invention described in claim 6, pattern illumination means for applying pattern illumination to the work is provided, and the work position shape detection means is obtained by the pattern illumination. Based on the image data of the work image forming means to be obtained, in order to three-dimensionally obtain the work shape, position, and posture, a space coding method,
There is an effect that the work can be recognized more stably by using the moire method or the like, and the work can be gripped in an optimum posture.

According to a ninth aspect of the invention, in the first aspect of the invention, a laser range finder is used as the work imaging means, and the work position / shape detecting means is the work. Since the shape, position and posture of the work are three-dimensionally determined based on the image data of the imaging means, the work can be recognized more stably, and the work can be gripped in the optimum posture.

According to the invention described in claim 10,
In the invention according to claim 1, storage means for storing the shape of the work calculated by the work shape position detection means, or the type of work based on the shape, and when assembling the work to the work to be assembled In addition, based on the image data of the work to be assembled and the shape of the work stored in the storage means, the optimum assembly position of the work on the work to be assembled is determined, and work assembly means for assembling is provided. Therefore, when the work is assembled to the work to be assembled, the assembly position can be determined from the shape of the work to be assembled, so that the teaching load applied during the assembly can be reduced.

According to the invention described in claim 11,
The invention according to claim 1, further comprising a communication unit capable of transmitting and receiving data to and from the robot controller, wherein the gripping position data generating unit has an operation range in which the generated gripping position data can be gripped by the robot hand. When it is detected that the position is outside the operating range, grip position data indicating that the position is outside the operating range is output to the robot controller via the communication unit and the position correction by the robot controller is executed. In cooperation with the correction, it is possible to perform a wider range of position correction.

According to the invention described in claim 12,
The invention according to claim 2, further comprising communication means capable of transmitting and receiving data to and from the robot controller, wherein the work position shape detecting means obtains a work shape corresponding to a gripped work from the internal database, Since the work shape or the kind of the work corresponding to the work shape is output to the robot controller via the communication means, even when a variety of mixed works are assembled in sequence, the gripped work can be attached to the assembly position in the next step. There is an effect that the efficiency of the assembling can be improved such as moving to.

[Brief description of drawings]

FIG. 1 is a diagram showing an overall configuration of a robot hand with a fingertip visual sense according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a configuration of a hand.

FIG. 3 is a flowchart showing the control content of the work gripping by the hand.

FIG. 4 is a flowchart showing another control content of gripping a work by a hand.

FIG. 5 is a block diagram showing another configuration example of a hand.

FIG. 6 is a flowchart showing a procedure of gripping position calculation based on recognition of a work shape.

FIG. 7 is a front view showing a specific configuration example of a hand.

FIG. 8 is a side view showing a specific configuration example of a hand.

FIG. 9 is a diagram for explaining a gripping position of a work by a form change hand as a specific configuration example of the hand.

FIG. 10 is a diagram showing an example of a work held by a hand.

FIG. 11 is a diagram showing another example of the work held by the hand.

FIG. 12 is a view for explaining the assembling of the work with respect to the work to be assembled.

FIG. 13 is a system configuration diagram showing a conventional robot hand with fingertip vision.

[Explanation of symbols]

100 robots 101 arm 102 joints 103 hands 110 Robot controller 201 Work imaging means 202 Work position shape detection means 203 gripping position data generating means 204 (hand) gripping position control means 205,503 Communication means 206 fingers 207 stage 501 internal database (DB) 502 External database (DB) 700 Hand to change shape 701 stage 702 arm 702a basic axis 703 slide mechanism 1001,1101, W work 1002, 1102 base 1003 convex part 1103 recess 1201 Work to be assembled 1202, 1203, 1204 Assembly position F1, F2, F3 Finger grip position Center position of P1, P2, P3 fingers Center position of S work L virtual straight line

Claims (12)

[Claims] 1. A work position imaging means for obtaining image data representing a shape and a position of a work held by a robot hand, and position and shape data of the work based on the work image data picked up by the work imaging means. A work position / shape detecting means for calculating; a gripping position data generating means for generating gripping position data for gripping a work by a robot hand based on the position and shape data calculated by the work position / shape detecting means; A hand gripping position control means for changing the gripping position of the robot hand based on the gripping position data generated by the position data generating means, and the robot hand is capable of autonomously gripping a workpiece at an optimum gripping position. A characteristic robot hand with a visual tip. 2. The work position shape detection means includes an internal database in which data regarding the shape and grip position of the work is accumulated, and based on the calculated work shape data, the work position gripping position is referenced with reference to the internal database. The robot hand with a fingertip visual sense according to claim 1, wherein the robot hand is obtained. 3. An external database provided outside the robot hand for accumulating data relating to the shape and gripping position of a work, and communication means capable of transmitting and receiving data to and from the external database. When the workpiece to be gripped does not match the shape data of the corresponding workpiece in the internal database, the grip position corresponding to the workpiece shape data is obtained from the external database. The robot hand with a hand tip according to claim 2. 4. The work position shape detection means additionally stores the work shape data obtained as a result of an inquiry to an external database in the internal database,
The robot hand with a visual point of the hand according to claim 3, wherein the shape data of the corresponding work is obtained by using the internal database thereafter. 5. The grip position data generation means calculates a grip position for each shape of the work based on the position and shape data calculated by the work position shape detection means. Robot hand with visual sense of hands. 6. The robot hand with a hand tip according to claim 1, wherein the workpiece imaging means is composed of a two-dimensional CCD sensor or a two-dimensional CMOS sensor. 7. The work imaging means includes at least two two-dimensional CCD sensors or two-dimensional CMOS sensors, and images the work from different angles, and the work position / shape detecting means includes the work. The robot hand with a fingertip according to claim 1, wherein the shape, position, posture, and the like of the work are three-dimensionally determined based on two or more image data obtained by the imaging means. 8. A pattern illuminating means for illuminating a work with pattern illumination is provided, wherein the work position shape detecting means determines the work shape, position, and orientation based on image data of the work imaging means obtained by the pattern illumination. The robot hand with a visual point of the hand according to claim 6, wherein the robot hand is obtained three-dimensionally. 9. A laser range finder is used for the work imaging means, and the work position shape detection means three-dimensionally determines the work shape, position, and orientation based on the image data of the work imaging means. The robot hand with a hand tip according to claim 1, characterized in that 10. Storage means for storing the shape of the work calculated by the work shape position detecting means, or a kind of work based on the shape, and an assembling work for assembling the work to the assembling work. Based on the image data of the work and the shape of the work stored in the storage means, a work assembling means for determining an optimum work mounting position on the work to be assembled and assembling is provided. The robot hand with a hand tip according to claim 1, which is characterized in that. 11. A communication unit capable of transmitting and receiving data to and from the robot controller, wherein the gripping position data generating unit is out of an operation range in which the generated gripping position data can be gripped by the robot hand. The grip position data indicating that the position is out of the operating range is output to the robot controller via the communication unit, and the position correction by the robot controller is executed when the position is detected. Robot hand with visual sense of hands. 12. A communication means capable of transmitting and receiving data to and from the robot controller, wherein the work position shape detecting means obtains a work shape corresponding to a gripped work from the internal database, and The robot hand with a hand tip according to claim 2, wherein the type of the work corresponding to the work shape is output to the robot controller via the communication means.