Robot system and method for controlling robot

Patent metadata

Field	Value
Publication number	US-11826902-B2
Application number	US-201816965649-A
Country	US
Kind code	B2
Filing date	Nov 27, 2018
Priority date	Feb 1, 2018
Publication date	Nov 28, 2023
Grant date	Nov 28, 2023

Abstract

A robot system and method for controlling a robot, wherein during learning, a detection unit detects, as waveform data for learning, the contact state when a socket is caused to contact a set position of the head of a bolt and is caused to rotate around the set position within a set movement range. A learning unit learns a plurality of detected sets of the waveform data for learning and writes the learning results to a determination unit. During practical operations, the determination unit recognizes the amount that the socket slips with respect to the bolt on the basis of actual waveform data indicating the change in the contact state when the socket is in contact with the head of the bolt and the written learning results.

First claim

What is claim is:

1. A robot system configured to, when a sensor has sensed a change in a contact state between an end effector of a robot and a workpiece, estimate a deviation amount of the end effector relative to the workpiece based on the change in the contact state sensed by the sensor and correct the deviation amount by controlling the robot and/or the end effector, the robot system comprising:
a control section configured to control the robot and/or the end effector;
a detecting section configured to detect data indicating change over time of the contact state when the contact state between the end effector and the workpiece has been changed by the control section;
a learning section configured to learn the data; and
a judging section configured to judge the contact state between the end effector and the workpiece during actual work performed on the workpiece by the end effector, using learning results of the data obtained by the learning section, wherein
the detecting section detects in advance, as learning data, a change in the contact state sensed by the sensor, for each of a plurality of set positions on a surface of the workpiece, when the end effector has been caused to contact the set position and a change in the contact state sensed by the sensor when the end effector has been caused to perform a feeling-around operation within a set operational range relative to the set position,
the learning section learns a plurality of pieces of the detected learning data and writes learning results of the plurality of pieces of learning data to the judging section,
the judging section, during the actual work, acquires the change in the contact state detected by the sensor when the end effector has contacted the surface of the workpiece and, based on the acquired change in the contact state and the written learning results, recognizes a deviation of the end effector relative to the workpiece, and
the control section estimates the deviation amount based on a judgment result of the judging section.

2. The robot system according to claim 1 , wherein
the detecting section, during the actual work, detects, as actual data, the change in the contact state sensed by the senor when the end effector has contacted an arbitrary position on the surface of the workpiece and the change in the contact state sensed by the sensor when the end effector has been caused to perform the feeling-around operation within an arbitrary operational range relative to the arbitrary position, and
the judging section recognizes the deviation of the end effector relative to the workpiece, based on the detected actual data and the learning results of the learning data that resembles the actual data and which is included in the written learning results.

3. The robot system according to claim 2 , wherein
the plurality of pieces of learning data have characteristics that differ according to a type of deviation of the end effector relative to the workpiece,
the learning section learns the characteristics that differ according to the type of deviation, for the plurality of pieces of learning data, and
the judging section, during the actual work, identifies the type of deviation corresponding to the actual data by judging the contact state indicated by the actual data, using the learning results, and recognizes the deviation of the end effector relative to the workpiece using the actual data and the learning result of the learning data corresponding to the identified type of deviation, which is included in the written learning results.

4. The robot system according to claim 1 , wherein
the end effector is caused to contact the workpiece by controlling the robot and/or the end effector with the control section or causing the end effector to move within a movable range using an external force movable mechanism, so that the contact state between the end effector and the workpiece does not exceed a threshold value.

5. The robot system according to claim 1 , wherein
the sensor is a touch sensor mounted in a vicinity of the end effector on the robot.

6. The robot system according to claim 5 , wherein
the touch sensor is mounted on the robot so as to be substantially coaxial with a central axis of the end effector, or is mounted on the robot in a state where an axis of the touch sensor is not coaxial with the central axis, and
if the axis of the touch sensor is not coaxial with the central axis, the end effector is mounted on the robot via the touch sensor.

7. The robot system according to claim 1 , wherein
the workpiece is a member assembled on a target object or a member tightened to the target object, and
the end effector is a tool configured to assemble the member on the target object or a tool configured to tighten the member to the target object.

8. A robot control method for, when a sensor has sensed a change in a contact state between an end effector of a robot and a workpiece, estimating a deviation amount of the end effector relative to the workpiece based on the change in the contact state sensed by the sensor and correcting the deviation amount by controlling the robot and/or the end effector, the robot control method comprising:
a first step of detecting, as learning data, with a detecting section, a change in the contact state sensed by the sensor for a set position on a surface of the workpiece when the end effector has been caused to contact the set position and a change in the contact state sensed by the sensor when the end effector has been caused to perform a feeling-around operation within a set operational range relative to the set position;
a second step of learning, with a learning section, the detected learning data;
a third step of, after the first step and the second step have been repeatedly performed in order, writing learning results of a plurality of pieces of the learning data obtained by the learning section to a judging section;
a fourth step of, during actual work performed on the workpiece by the end effector, acquiring, with the judging section, the change in the contact state detected by the sensor when the end effector has contacted the surface of the workpiece and, based on the acquired change in the contact state and the written learning results, recognizing a deviation of the end effector relative to the workpiece; and
a fifth step of estimating, with a control section, the deviation amount based on a judgment result of the judging section.

Assignees

• Honda Motor Co Ltd

Inventors

• Ishizuka Yasutaka
• Arimitsu Takeshi
• GODA Kasane
• Yoshida Yutaka
• Muraoka Kotaro

Classifications

• B25J13/084 Primary

  Tactile sensors (in general G01L5/16, G01L5/22) · CPC title

• B25J9/163

  learning, adaptive, model based, rule based expert control · CPC title

• B25J9/1687 Primary

  Assembly, peg and hole, palletising, straight line, weaving pattern movement · CPC title

• B25J15/0019

  End effectors other than grippers · CPC title

• B25J19/063

  working only upon contact with an outside object · CPC title