Display Bonding Apparatus Using Deep Learning Alignment Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The challenge in manufacturing display apparatuses lies in minimizing alignment errors between materials during the bonding process, which affects the precision and quality of the final product.
Innovation Solution
An apparatus and method utilizing deep learning technology to sense image information of materials and calculate position and alignment information, with a controller that adjusts the bonding position through a series of correction models to minimize errors, incorporating a camera unit, stages for material support, and a deep learning unit for continuous improvement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If traditional bonding process is used, then manufacturing process is simple, but alignment error between materials is large
Solution Approach 1:
The patent implements a feedback mechanism where the deep learning unit continuously receives alignment error data from the image processor and updates the operator through deep learning. This closed-loop feedback system allows the bonding apparatus to automatically improve its alignment precision by learning from previous bonding results, directly addressing the technical contradiction between manufacturing precision and device complexity.
Solution Approach 2:
The patent replaces traditional mechanical alignment methods with an intelligent system combining image processing and deep learning. The image processor calculates position and alignment information from images, while the deep learning unit optimizes the operator, substituting complex mechanical adjustment mechanisms with software-based intelligent control that achieves higher precision without proportional increases in mechanical complexity.
2Manufacturing precision
If bonding process is repeated multiple times, then alignment error is reduced through learning, but manufacturing time increases
Solution Approach 1:
The patent performs preliminary action by pre-training the deep learning unit with calibration data before actual bonding operations. The calibration process establishes initial correction models that provide a head start, reducing the number of iterative bonding cycles needed to achieve target precision. This preliminary preparation minimizes the time penalty associated with repeated bonding processes.
Solution Approach 2:
The patent implements continuous learning during the bonding process, where the deep learning unit continuously updates the operator based on feedback from each bonding operation. This continuous improvement mechanism allows the system to progressively reduce alignment errors across multiple bonding cycles without requiring complete stops or manual interventions, maintaining productive action while achieving precision improvements.
Data Source
AI summary
An apparatus for manufacturing a display apparatus includes a controller configured to control a second stage, wherein the controller includes an image processor configured to calculate position information of a first material and a second material and alignment information of a bonded material based on image information sensed by a camera unit, an operator configured to calculate final bonding position based on the position information of the first material and the second material calculated by the image processor, a controller unit configured to move the second stage to the final bonding position calculated by the operator, and a deep learning unit configured to update the operator through deep learning based on the alignment information of the bonded material calculated by the image processor.