Robot Axis Command Inference for Accurate Trajectory Reproduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for improving robot trajectory accuracy, such as learning control methods, face limitations due to noise in sensor data and the need for frequent recalibration with instruments like laser trackers, which are cumbersome and troublesome to set up.
Innovation Solution
A controller and machine learning device that directly measure a robot's operation trajectory as position data, using machine learning to infer command data for each axis based on target trajectory data, incorporating an axis angle conversion unit, state observation unit, label data acquisition unit, and learning unit to associate axis angle changes with command data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If learning control is repeated based on sensor data (acceleration, angular speed), then trajectory accuracy can be improved, but measurement precision deteriorates due to noise in sensor data requiring integration calculations
Solution Approach 1:
The patent replaces sensor-based indirect measurement (acceleration/angular speed sensors requiring integration) with direct position measurement using a position measurement instrument. This substitution eliminates the need for integration calculations and avoids accumulation of errors from noisy sensor data, directly providing accurate position information for trajectory learning.
Solution Approach 2:
The patent introduces a position measurement instrument as an intermediary device between the robot and the measurement process. This intermediary directly measures the position of the robot's end effector or key points, providing clean position data without the noise and integration requirements of traditional sensor-based methods.
2Measurement precision
If laser trackers are used to directly measure robot trajectory position data, then measurement precision improves, but device complexity increases due to installation requirements
Solution Approach 1:
The patent creates a virtual model (copy) of the robot's target trajectory through machine learning. The position measurement instrument measures the actual trajectory once, and the learned model reproduces the target trajectory without requiring repeated physical measurements or reinstallation of measurement instruments for different operations.
Solution Approach 2:
The robot system performs self-learning by automatically acquiring position data from the measurement instrument and using machine learning to build its own trajectory reproduction capability. This eliminates the need for external experts to repeatedly install and configure laser trackers for different robot operations.
3Manufacturing precision
If machine learning is performed with measured trajectory data and command data, then trajectory accuracy improves, but loss of time increases due to data processing and learning computation
Solution Approach 1:
The patent extracts only the essential position data from the measured trajectory using a position measurement instrument, rather than processing continuous sensor data streams. This extraction of key position information reduces the volume of data requiring machine learning processing, thereby reducing computation time while maintaining learning accuracy.
Data Source
AI summary
A machine learning device is provided in a versatile controller capable of inferring command data to be issued to each axis of a robot. The device includes an axis angle conversion unit calculating, from the trajectory data, an amount of change of an axis angle of an axis of the robot, a state observation unit observing axis angle data relating to the amount of change of the axis angle of the axis of the robot as a state variable representing a current state of an environment, a label data acquisition unit acquiring axis angle command data relating to command data for the axis of the robot as label data, and a learning unit learning the amount of change of the axis angle of the axis of the robot and the command data for the axis in association with each other by using the state variable and the label data.