Robot Learning Control for End Effector Vibration Suppression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing robot systems face challenges in accurately and easily reducing vibrations in end effectors due to resonance frequencies being within the control band of the learning controller, leading to phase delays and gain increases, which can cause vibration divergence.
Innovation Solution
A robot system with a learning control unit that includes a power spectrum calculating unit, comparison unit, and learning correction amount updating unit, which adjusts the phase and gain of the learning correction amount based on power spectrum comparisons to ensure the power spectrum at the current learning cycle is less than the preceding cycle, thereby correcting operation commands to reduce vibrations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the speed of the robot operation is increased to shorten tact time, then productivity is improved, but vibrations are generated in the end effector due to insufficient rigidity
Solution Approach 1:
The patent implements learning control that uses vibration sensors to detect end effector vibrations and feeds this information back to the control device. The control device calculates learning correction amounts based on detected vibrations and applies corrective commands to suppress vibrations, enabling high-speed operation without excessive vibration.
Solution Approach 2:
The system performs learning control by executing operation programs multiple times to preliminarily identify vibration characteristics and generate correction amounts. This preliminary learning phase enables the system to proactively compensate for vibrations before they become problematic during normal high-speed operation.
2Device complexity
If a learning controller is designed in advance based on frequency response data for a representative end effector, then device complexity is reduced, but vibration elimination accuracy deteriorates when the actual end effector has low rigidity and resonance frequency within the control band
Solution Approach 1:
The patent implements adaptive learning control that dynamically adjusts the control parameters based on actual vibration measurements from the specific end effector. Instead of using fixed parameters designed for a representative end effector, the system learns and adapts to the actual resonance characteristics of each end effector, enabling accurate vibration elimination across different rigidity levels.
Solution Approach 2:
The system changes control parameters (phase and gain of learning correction amounts) based on detected vibration characteristics. By monitoring power spectra and identifying resonance frequencies, the system automatically adjusts correction parameters to match the actual end effector properties, thereby maintaining vibration elimination accuracy without redesigning the controller.
3Manufacturing precision
If the phase is delayed and gain is increased in the learning controller to handle low rigidity end effectors, then vibration elimination capability is improved, but vibration divergence occurs
Solution Approach 1:
The system continuously monitors vibration levels through power spectrum analysis and uses this feedback to adjust learning correction parameters. When vibrations are detected to be increasing, the system can reduce gain or adjust phase to prevent divergence, maintaining stability while eliminating vibrations.
Solution Approach 2:
The patent applies learning correction amounts that are tailored to the actual vibration levels detected, rather than applying maximum correction. By using partial correction based on measured vibrations, the system avoids excessive gain that could cause divergence while still effectively reducing vibrations.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach allows for accurate and efficient vibration reduction in end effectors, even when resonance frequencies are within the control band, without the need for frequent redesign of the learning control unit, thus simplifying and cost-effectively addressing vibration issues.
Implementation Method 1
a power spectrum calculating unit configured to calculate a power spectrum by Fourier transforming vibration data detected by the sensor for each learning cycle by the learning control unit
Data Source
AI summary
A robot system includes: a learning control unit configured to perform learning for calculating a learning correction amount for bringing a position of a control target portion toward a target position; a robot control unit configured to control the operation of the robot mechanism unit; a power spectrum calculating unit configured to calculate a power spectrum of a vibration data of the control target portion; a comparison unit configured to compare each power spectrum between at the time of the current learning and at the time of the immediately preceding learning; and a learning correction amount updating unit configured to adjust at least one of a phase and a gain of the learning correction amount used at the time of the current learning to set the adjusted learning correction amount as a new learning correction amount used at the time of next learning.