JP2022090389A - Robot hand and operation control method of the same - Google Patents

Abstract

To provide a robot hand that is controlled in accordance with a shape of an article to move the article while gripping the article, and an operation control method of the same.SOLUTION: A robot hand is arranged with a plurality of pressure sensors in a matrix shape on a surface of a flexible gripping part deformable following a surface of an article, and estimates a shape of the surface of the article from a pressure distribution pattern on the surface contacting the article. A control method of the robot hand not only estimates the shape of the surface of the article, but also measures change with time of the pressure distribution pattern, and feedbacks the measured change to control movements of the gripping part.SELECTED DRAWING: Figure 4

Description

The present invention relates to a robot hand for gripping and moving an article and a motion control method, and more particularly to a robot hand and a motion control method for gripping and moving the article by giving control according to the shape of the article.

There is known an industrial robot that controls a robot hand based on information obtained by photographing an article with a camera and performing image processing, and performs a picking operation of grasping and moving the article. In the method using the image pickup with such a camera, the position of the article can be recognized, but the article can be identified only by the appearance. Therefore, for example, the box containing the contents and the empty box cannot be distinguished. Therefore, if the article is known in advance and it is not possible to determine how much gripping force it should be gripped, the article may fail to be gripped and be dropped or crushed and destroyed. Therefore, an attempt has been made to incorporate a pressure sensor into the robot hand to measure the slippage and the center of gravity of the gripped article to ensure that the picking operation is performed.

For example, in Patent Document 1, a tactile sensor composed of a plurality of sensors such as a pressure sensor is arranged on a finger surface having a flexible structure, and a normal contact force F _n acting on a fingertip at the start of gripping an article to be gripped is initially used. A robot hand that measures the pressure center position C ₀ and controls the picking operation is disclosed. When the article is lifted, the flexible structure deforms according to the external force (gravity) F acting on the article, so that the pressure center of gravity position C moves, but the difference between C and C ₀ , that is, the amount of movement S at the center of gravity position Based on this, the occurrence of slip is predicted, and when slip occurs, it is controlled to increase the gripping force of the finger to ensure that the picking operation is performed.

Further, also in Patent Document 2, a pressure sensor that is deformed by an external load to cause stress distribution is provided, the pressure center position is calculated from the pressure value detected by the pressure sensor, and the pressure center position is changed with time. A robot hand that detects slippage is disclosed. Here, it has a multi-layered structure in which two detection units that detect the pressure of the pressure sensor are arranged so as to sandwich a viscoelastic body that is deformed by an external load, and complicated control such as stable gripping and dexterous operation is performed. The grip information necessary for performing the above is obtained.

Japanese Unexamined Patent Publication No. 2006-297542 Japanese Unexamined Patent Publication No. 2009-34742

When a spherical article is lifted and moved by pinching it with two fingers, the gripping force greatly differs depending on the position and direction of the contact surface between the fingertip and the article. If you are a human, you can unknowingly control the force of your fingertips and change the positional relationship between your fingertips and the contact surface of the article relative to make it easier to grip the article, depending on the shape of the article. The gripping motion is controlled. Therefore, even in a robot hand incorporating a pressure sensor as described above, if the gripping force or the like can be adjusted according to the shape of the article, the picking operation can be more stabilized.

The present invention has been made in view of the above-mentioned situation, and an object thereof is a robot hand and an operation control method for gripping and moving an article by giving control according to the shape of the article. Is to provide.

The present invention is a robot hand that gives control according to the shape of an article to grip and move the article, and has a matrix shape on the surface of a grip portion having flexibility that can follow the surface of the article and deform. A plurality of pressure sensors are arranged in the article, and the surface shape of the article is estimated from the pressure distribution pattern at the contact surface with the article.

According to such a feature, the surface shape of the article can be estimated by comparing with the pressure distribution pattern corresponding to the known surface shape, and the ease of gripping can be determined from the surface shape and the gripping force can be adjusted. , The picking operation can be stabilized.

In the above-described invention, the pressure sensor may be provided so as to measure the pressure pressed against the surface of the grip portion by the pressure from the article. According to such a feature, the pressure distribution pattern can be obtained with high accuracy and the picking operation can be stabilized.

The invention described above may be characterized in that the predetermined information is encoded and the surface of the article is conspicuously provided, the unevenness is measured as the pressure distribution pattern, and the concavo-convex is back-coded to obtain the predetermined information. According to such a feature, the article information can be given to the article in advance and the gripping force or the like can be adjusted in response to this information, and the picking operation can be stabilized.

The invention described above may be characterized in that the time change of the pressure distribution pattern is measured and feedback control is performed to the movement of the grip portion. Further, the pressure distribution pattern may be characterized in that the relative position between the grip portion and the article is given together. According to such a feature, the picking operation can be stabilized by performing relative movement between the grip portion and the article by adjusting the gripping force or the like.

The invention described above is characterized in that a plurality of grip portions that can be separately driven and controlled are included, the article is sandwiched between the grip portions, and the article is gripped, and the surface shape of the article is estimated from the plurality of pressure distribution patterns. May be. According to such a feature, the relative movement between the grip portion and the article can be stabilized to stabilize the picking operation.

Further, the present invention is a method for controlling a robot hand that grips and moves the article by giving control according to the shape of the article, and has the flexibility to follow the surface of the article and deform. A plurality of pressure sensors are arranged in a matrix on the surface of the article, the surface shape of the article is estimated from the pressure distribution pattern at the contact surface with the article, and the time change of the pressure distribution pattern is measured to measure the grip portion. It is characterized by feedback control on the movement of the robot.

According to such a feature, the surface shape of the article can be estimated by comparing with the pressure distribution pattern corresponding to the known surface shape, and the ease of gripping can be determined from the surface shape and the gripping force can be adjusted. , The picking operation can be stabilized.

In the above-described invention, the pressure distribution pattern may be characterized in that the relative position between the grip portion and the article is given together. According to such a feature, the picking operation can be stabilized by performing relative movement between the grip portion and the article by adjusting the gripping force or the like.

The invention described above is characterized in that a plurality of grip portions that can be separately driven and controlled are included, the article is sandwiched between the grip portions, and the article is gripped, and the surface shape of the article is estimated from the plurality of pressure distribution patterns. May be. According to such a feature, the relative movement between the grip portion and the article can be stabilized to stabilize the picking operation.

It is an external photograph which attached the robot hand as an Example by this invention to a robot arm. It is a side view (partial sectional view) of the main part of a robot hand. It is an external photograph of the gripping part and the pressure detecting part used for a robot hand. (A) A side view of the grasped spherical article and (b) its pressure distribution pattern. (A) A side view of the gripped cubic article and (b) its pressure distribution pattern. (A) A side view of the grasped columnar article and (b) its pressure distribution pattern. (A) A side view of a robot hand holding a spherical article, (b) a diagram showing a pressure distribution pattern of one pressure detection unit, and (c) a diagram showing a pressure distribution pattern of the other pressure detection unit. It is a side view which shows an example of the control of a robot hand. (A) A side view of an article and a grip portion to which an article identification label is attached, (b) a front view of an article identification label, and (c) a view of a pressure distribution pattern. (A) A front view of another article identification label, (b) a side view thereof, and (c) a pressure distribution pattern.

Hereinafter, a robot hand and a control method thereof, which are one embodiment of the present invention, will be described with reference to FIGS. 1 to 10.

As shown in FIG. 1, the robot hand 10 includes a plurality of grip portions 1 that can be separately driven and controlled, and is sandwiched between them so that an article can be gripped. Here, the robot arm 20 is used. It is attached to the tip.

As shown in FIG. 2, the grip portion 1 has a pressure detecting portion 2 arranged on the surface thereof to form a part of the robot hand 10. The grip portion 1 has flexibility and can be deformed following the surface of the gripped article, and for example, a soft finger for a robot hand can be used. Further, the pressure detection unit 2 may be covered with the protective film 3. When a flexible strip such as a flexible substrate is used for the wiring of the pressure detection unit 2, this can also be used as the protective film 3.

Referring to FIG. 3 together, the pressure detecting unit 2 is attached and arranged on the contact portion 9 which is located on the tip end side of the grip portion 1 and easily comes into contact with the gripped article (see particularly FIG. 3A). Further, the pressure detection unit 2 is a flat plate-like body in which a plurality of pressure sensors are arranged in a matrix, and for example, a patch type pressure distribution sensor or the like can be used (see particularly FIG. 3 (b)).

As a result, when the article is gripped by the robot hand 10, the pressure detecting section 2 arranged in the contact portion 9 obtains a pressure distribution pattern corresponding to the surface shape of the article on the contact surface with the article. Such a pressure distribution pattern can be expressed by the position (coordinates) in the matrix of the pressure sensor that has detected the pressure. In addition, it is also preferable to measure the pressure pressed against the contact portion 9 by the pressure from the article. That is, the pressure intensity detected by each pressure sensor is added to obtain a pressure distribution pattern with higher accuracy.

For example, as shown in FIG. 4A, the robot hand 10 was made to grip the spherical article 4a, and the pressure detection unit 2 was brought into contact with a part of the surface of the article 4a. Then, as shown in FIG. 4B, the pressure detecting unit 2 detected a plurality of pressures having different intensities within a relatively narrow range.

Further, for example, as shown in FIG. 5A, the article 4b having a rectangular parallelepiped shape was grasped, and the side portion of the article 4b was brought into contact with the pressure detection unit 2. Then, as shown in FIG. 5B, the pressure detection unit 2 detected a plurality of pressures having a relatively strong constant intensity on a straight line.

Further, for example, as shown in FIG. 6A, the article 4c having the shape of a triangular prism was grasped, and the surface portion of the article 4c was brought into contact with the pressure detection unit 2. Then, as shown in FIG. 6 (b), the pressure detecting unit 2 detected a plurality of relatively weak and constant-strength pressures in the entire range of the quadrangle.

As described above, the pressure distribution pattern differs depending on the surface shape of the portion of the article in contact with the pressure detection unit 2. This makes it possible to obtain an estimate of the surface shape of the article. For example, the surface shape of an article can be estimated by comparing it with a pressure distribution pattern corresponding to a known surface shape. Then, the ease of gripping or the like can be determined from the estimated surface shape, and the gripping force or the like can be adjusted. Thereby, in the picking operation of gripping and moving the article, the picking operation can be stabilized by giving control according to the shape of the article. Further, by accumulating and learning about the obtained pressure distribution pattern and gripping state, it is possible to obtain a more stable picking operation.

As described above, since the robot hand 10 includes a plurality of gripping portions, a pressure detecting portion can be provided in each gripping portion. In that case, the surface shape of the gripped article can be estimated from a plurality of pressure detection patterns, and the result can be used.

For example, as shown in FIG. 7, when the spherical article 4a is gripped by the robot hand 10, the pressure detecting portion 2 of one gripping portion 1 has a pressure distribution pattern in which the pressure is detected near the center (FIG. 7). On the other hand, in the pressure detection unit 2'of the other grip portion 1', the pressure was detected in the vicinity of the lower end portion (FIG. 7 (c)). At this time, the spherical article 4a is gripped at the lower end of the grip portion 1'and is unstable as a gripped state by the robot hand 10. Information about the relative position between the grip portion 1'and the article 4a can be obtained together from the pressure distribution pattern.

Here, in the robot hand 10, it can be estimated from the pressure distribution pattern obtained by the pressure detecting units 2 and 2'as described above that the surface of the gripped article in contact is a convex curved surface. Further, when the shape of the article is limited in advance, it can be specified that the gripped article is spherical. In such a case, the robot hand 10 is controlled in order to eliminate the unstable gripping state described above.

Therefore, as shown in FIG. 8, the grip portion 1'is opened at the lower end of the grip portion 1'outward and the upper end is closed inward so that the spherical article 4a can be gripped at the center of the pressure detecting portion 2'. To move. Further, as the grip portion 1'moves, the lower end of the grip portion 1 is opened outward and the upper end is closed inward so that both sides of the spherical article 4a can be gripped as tangent surfaces. Move part 1. These movements may also include adjustment of gripping force.

In this way, it is possible to obtain information on the relative position between the gripped article and the gripped portion based on the pressure distribution pattern obtained by the pressure detecting portions 2 and 2'. Then, the robot hand 10 can be controlled to obtain a stable gripping state to perform relative movement between the gripping portion and the article, and the picking operation can be stabilized.

Further, the time change of the obtained pressure distribution pattern may be measured, and feedback control may be performed on the movement of the grip portions 1 and 1'of the robot hand 10 based on the measurement. For example, the pressure distribution pattern (see FIG. 8B) gripping the spherical article 4a at the center of the pressure detection unit 2'changes with time, and the pressure distribution pattern gripped at the lower end (FIG. 8). 8 (c)). At this time, the grip portion 1'is moved so that it can be gripped at the center of the pressure detecting portion 2'in the same manner as described above. Further, when such a change in the pressure distribution pattern is obtained, it can be estimated that slippage has occurred between the pressure detection unit 2'and the contact surface of the gripped article. Adjustments such as increasing the gripping force may be made.

Further, as shown in FIG. 9A, predetermined information can be obtained from the unevenness provided on the surface of the article. Specifically, on the surface of the article 4d, an article identification label 5 is attached by encoding the shape and weight of the article 4d and other predetermined information for identifying the article 4d and giving the article 4d as an uneven arrangement on a plane. There is.

As shown in FIG. 9B, the article identification label 5 is a label expressing the above-mentioned unevenness by arranging the convex portions 6 within a predetermined range. The position of the convex portion 6 can be detected as a pressure distribution pattern by abutting the pressure detecting portion 2 of the robot hand 10 on the article identification label 5 (see FIG. 9C). In the robot hand 10, this pressure distribution pattern can be inversely coded to obtain predetermined information. Then, the picking operation can be stabilized by adjusting the gripping force according to the predetermined information.

Further, as shown in FIG. 10, as the article identification label 5', a convex portion 6'which is lower in height than the convex portion 6 may be provided. By doing so, the pressure detecting unit 2 detects a high pressure corresponding to the high convex portion 6 and detects a low pressure corresponding to the low convex portion 6', and thus further. The strength of the pressure can be added as information. That is, the positions and heights of the protrusions 6 and 6'can be detected as a pressure distribution pattern. Thereby, more predetermined information can be obtained from the article identification label 5'.

Although the examples according to the present invention and the modifications based on the present invention have been described above, the present invention is not necessarily limited to this, and those skilled in the art deviate from the gist of the present invention or the scope of the attached claims. Without doing so, various alternative and modified examples could be found.

1 Grip part 2 Pressure detection part 4a-4d Article 10 Robot hand

Claims (9)

A robot hand that gives control according to the shape of an article to grip and move the article.
A plurality of pressure sensors are arranged in a matrix on the surface of the grip portion having flexibility that can follow the surface of the article, and the surface shape of the article is estimated from the pressure distribution pattern at the contact surface with the article. A robot hand characterized by getting. The robot hand according to claim 1, wherein the pressure sensor is provided so as to measure a pressure pressed against the surface of the grip portion by a pressure from the article. The invention according to claim 1 or 2, wherein the predetermined information is encoded and the surface of the article is conspicuously provided, and the concavities and convexities are measured as the pressure distribution pattern and back-coded to obtain the predetermined information. Robot hand. The robot hand according to claim 1 or 2, wherein the time change of the pressure distribution pattern is measured and feedback control is performed to the movement of the grip portion. The robot hand according to claim 4, wherein the pressure distribution pattern also gives a relative position between the grip portion and the article. 4. The fourth aspect of the present invention includes a plurality of grip portions that can be separately driven and controlled, sandwiches and grips the article between them, and obtains an estimation of the surface shape of the article from the plurality of pressure distribution patterns. Or the robot hand according to 5. It is a control method of a robot hand that gives control according to the shape of an article to grip and move the article.
A plurality of pressure sensors are arranged in a matrix on the surface of the grip portion having flexibility that can follow the surface of the article, and the surface shape of the article is estimated from the pressure distribution pattern at the contact surface with the article. A method for controlling the operation of a robot hand, which comprises measuring a time change of the pressure distribution pattern and performing feedback control on the movement of the grip portion. The operation control method for a robot hand according to claim 7, wherein the pressure distribution pattern also gives a relative position between the grip portion and the article. 8. Claim 8 is characterized in that a plurality of grip portions that can be separately driven and controlled are included, and the article is sandwiched and gripped between them, and the surface shape of the article is estimated from the plurality of pressure distribution patterns. The robot hand motion control method described.

Images