Robot controller, and robot control method

Abstract

The invention provides a robot controller having a function that simplifies learning and a robot control method. The robot controller includes: a learning section configured to carry out learning of detecting a deviation between a commanded trajectory representing a position of the robot generated according to the command values and an operation trajectory representing an actual position where therobot has moved, and generate a corrected program by adjusting the commanded trajectory; a saving section configured to save the corrected program; and a relearning section configured to carry out relearning on a relearning location, the relearning location being a part of the operation trajectory designated by an operator.

Description

technical field [0001] The present invention relates to a robot control device having a function of facilitating learning, and a robot control method. Background technique [0002] Generally, in a robot controlled by an operation program, a motor is controlled in accordance with command values ​​included in the operation program, and operates so as to draw a command trajectory taught by the operation program. At this time, the motor may not follow the command value due to factors such as backlash or friction of each joint of the robot, which may include errors in the actual robot position (actual trajectory) relative to the command trajectory. [0003] In order to eliminate this error, there is known a technique of detecting a deviation between the commanded trajectory and the actual trajectory, and repeatedly adjusting the command value to perform learning. For example, Japanese Patent Laid-Open No. 2015-221491 describes a robot control method, which includes: starting a u...

AI Technical Summary

Problems solved by technology

[0006] In general, since the result of learning is applied to the entire trajectory, there is a problem that it is difficult to perform learning by correcting only a part of the trajectory even when a part of the learned trajectory is not the trajectory desired by the operator

In addition, although there is a technology to automatically switch between the learning period and the non-learning period, the algorithm and calculation processing may become complicated.

Images