Computer Vision Inspection for Autonomous Substrate Handling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing substrate processing systems lack real-time, in-situ inspection and control capabilities, leading to inefficiencies and suboptimal operation of processing chambers and components.
Innovation Solution
An integrated computer vision system that captures images of substrates and components within substrate processing systems, using machine learning models to analyze these images for feature recognition and extraction, enabling real-time monitoring, predictive maintenance, and automated control of processing operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional metrology systems are used for substrate inspection, then the system structure remains simple, but real-time inspection capability and diagnostic accuracy are insufficient
Solution Approach 1:
The patent replaces traditional mechanical metrology systems with a computer vision system using cameras and machine learning algorithms. The camera captures images of substrates and components, while ML models perform real-time analysis for defect detection, substrate classification, and component condition assessment, achieving superior diagnostic accuracy without mechanical contact
Solution Approach 2:
The patent introduces an image processing and machine learning intermediary layer between the camera and the control system. This intermediary processes captured images through ML models to extract meaningful information about substrate conditions, defects, and component status, enabling accurate real-time diagnostics while maintaining modular system architecture
2Productivity
If manual intervention is used for substrate processing monitoring, then the system remains simple to operate, but throughput and productivity are reduced
Solution Approach 1:
The patent implements a self-monitoring system where the computer vision and machine learning autonomously inspect substrates, detect defects, classify substrate types, and assess component conditions without human intervention. The system automatically generates alerts and recommendations, enabling continuous operation and maximizing throughput while reducing manual labor requirements
Solution Approach 2:
The patent establishes a closed-loop feedback system where the camera continuously captures images, ML models analyze them in real-time, and the system automatically adjusts processing parameters or alerts operators based on detected conditions. This continuous feedback enables real-time quality control and maintains high productivity without compromising operational simplicity
3Manufacturing precision
If real-time in-situ inspection is implemented, then process quality and yield improve, but the device complexity and cost increase
Solution Approach 1:
The patent designs a multi-functional computer vision system where a single camera and ML platform perform multiple inspection tasks including substrate defect detection, substrate type classification, component condition assessment, and process parameter monitoring. This universal approach achieves high manufacturing precision across multiple functions while avoiding the complexity of multiple separate inspection systems
Solution Approach 2:
The patent uses optical copying through camera imaging to create digital representations of substrates and components for analysis. Instead of physical measurement or direct manipulation, the system captures optical images and processes them through ML models, enabling non-contact real-time inspection that maintains manufacturing precision while minimizing system complexity
4Reliability
If traditional inspection methods are used, then the system requires minimal infrastructure, but real-time monitoring and predictive maintenance capabilities are lacking
Solution Approach 1:
The patent implements predictive maintenance by continuously monitoring component conditions through image analysis and ML models. The system detects early signs of wear, damage, or degradation in robots, pads, and other components, enabling proactive maintenance scheduling before failures occur. This preliminary action maintains high reliability while ensuring real-time data is captured and analyzed
Data Source
AI summary
A substrate processing system comprises an edge computing device including processor that executes instructions stored in a memory to process an image or video captured by camera(s) of at least one of a substrate and a component of the substrate processing system. The component is associated with a robot transporting the substrate between processing chambers of the substrate processing system or between the substrate processing system and a second substrate processing system. The cameras are located along a travel path of the substrate. The instructions configure the processor to transmit first data from the image to a remote server via a network and to receive second data from the remote server via the network in response to transmitting the first data to the remote server. The instructions configure the processor to operate the substrate processing system according to the second data in an automated or autonomous manner.