Machine learning device and method, robot control device, robot vision system

Through machine learning devices and methods, the image data of the visual sensor is used to calculate the movement amount of the robotic arm and build a training data set, which solves the stability and efficiency of target position detection in the visual feedback system, and achieves high efficiency without pre-preparing marking points Target position adjustment.

CN110315505BActive Publication Date: 2024-11-26FANUC LTD
View PDF 14 Cites 0 Cited by

Patent Information

Application Number
CN201910237675.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Priority Date
2018-03-29
Filing Date
2019-03-27
Publication Date
2024-11-26
Estimated Expiration
2039-03-27

AI Technical Summary

Technical Problem

In robot systems that use visual feedback, target marks or feature points need to be prepared in advance, which leads to a high burden on the operator and unstable detection algorithms, affecting the efficient operation of the system.

Method used

Using machine learning devices and methods, the image of the workpiece is captured through visual sensors, the movement amount of the front end of the robot arm is calculated as the state variable, and the target image is used as the judgment data to construct a training data set, and the front end of the robot arm is learned to move from any position to The amount of movement of the target position reduces the burden on the operator and improves detection stability.

Benefits of technology

There is no need to prepare target marks or feature points in advance, which reduces the burden on the operator, and improves the stability and efficiency of target position detection through machine learning algorithms, allowing the robotic arm to automatically adjust to the target position.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN110315505B_ABST
    Figure CN110315505B_ABST
Patent Text Reader

Abstract

The present invention provides a machine learning device and method, a robot control device, and a robot vision system. The machine learning device has: an image of a workpiece photographed by a visual sensor and a movement amount of a front end of a robot arm from an arbitrary position calculated in a manner to make the image close to a target image as a state variable; a determination data acquisition unit that acquires the target image as determination data; and a learning unit that learns the movement amount for moving the front end of the robot arm or the workpiece from an arbitrary position to a target position according to a training data set composed of a combination of the state variable and the determination data, the target position being a position where the visual sensor and the workpiece are in a predetermined relative positional relationship, and the target image being an image of the workpiece photographed by the visual sensor when the front end of the robot arm or the workpiece is arranged at the target position.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Movement control method of robot

    JP1997076185A

  • Robot control device

    JP2003211381A

  • Control method and control device for mobile robot

    JP2003305676A

  • Position correction method of robot hand, robot hand, and robot

    JP2010188432A

  • Robot control method and robot

    CN101396829A