Wafer Placement Detection Using AI Sub-Area Segmentation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current manual inspection methods for semiconductor wafers in wafer carriers are inefficient and prone to errors due to the dense packing of wafers, varying shapes, and potential oblique or damaged positions, leading to eye fatigue and increased risk of wafer collision during inspection.

Innovation Solution

An electronic device equipped with an object detecting apparatus that captures images of wafers in a container, inputs them into an artificial intelligence model, and determines the states of placement, including correct positioning, orientation, and integrity, by dividing the container into sub-areas and using computer vision or deep learning for image processing.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual inspection from six viewpoints is used, then comprehensive inspection coverage is achieved, but inspection time increases and productivity decreases

Engineering Contradiction:
Improveinspection coverageVSAvoidinspection throughput
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent uses image copying technology to capture multiple viewpoints of wafers simultaneously through a single camera system. By capturing images from different angles and processing them computationally, the system achieves comprehensive inspection coverage without requiring manual inspection from six separate viewpoints, thereby maintaining measurement precision while significantly improving productivity

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The patent replaces the manual mechanical inspection process with an automated optical inspection system. Instead of manually positioning and inspecting wafers from six viewpoints, the system uses automated image capture and computational processing to achieve the same inspection coverage, eliminating the time-consuming manual operations while maintaining comprehensive inspection capability

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If the carrier is rotated for inspection, then all viewpoints are accessible, but wafers may suffer collision damages

Engineering Contradiction:
Improveviewpoint coverageVSAvoidwafer collision damage
Core Design Contradiction:
Measurement precisionVSObject-affected harmful factors

Solution Approach 1:

The patent captures images from multiple viewpoints simultaneously without physically rotating the carrier. By using multiple camera angles and computational image processing, the system obtains comprehensive viewpoint coverage while the carrier remains stationary, eliminating the risk of wafer collision damage that would occur during rotation

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The patent performs preliminary image capture from multiple angles before any carrier movement occurs. By capturing all necessary inspection data in a single stationary position, the system eliminates the need for subsequent carrier rotation, thereby preventing wafer collision damage while maintaining complete viewpoint coverage

Inventive Principle:
Principle #10Preliminary action

3Quantity of substance

If wafers are densely placed in carrier, then space utilization is improved, but detection difficulty increases

Engineering Contradiction:
Improvewafer densityVSAvoidwafer position detection
Core Design Contradiction:
Quantity of substanceVSDifficulty of detecting and measuring

Solution Approach 1:

The patent divides the inspection task into segmented image processing. By capturing images from multiple viewpoints and processing each region separately through computational algorithms, the system can distinguish individual wafer positions even in densely packed carriers. The segmentation approach breaks down the complex detection problem into manageable regions, maintaining high wafer density while improving detection capability

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent transitions from two-dimensional planar inspection to three-dimensional multi-viewpoint inspection. By capturing images from multiple angles and combining them computationally, the system adds a dimensional perspective that enables clear differentiation of densely packed wafers, maintaining high space utilization while significantly improving position detection accuracy

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

4Measurement precision

If inspector works for extended period, then inspection thoroughness may improve, but eye fatigue increases and errors occur

Engineering Contradiction:
Improveinspection thoroughnessVSAvoidinspection accuracy
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The patent implements self-service inspection through automated optical systems and computational image processing. The system performs comprehensive and accurate inspection without human intervention, eliminating eye fatigue and its associated errors. The automated system maintains consistent inspection thoroughness and accuracy regardless of operation duration, resolving the contradiction between inspection quality and reliability

Inventive Principle:
Principle #25Self-service

Data Source

PatentUS12272140B2Method for detecting objects and their state, apparatus and storage medium applying method, and electronic device
Publication Date: 2025.04.08 HON HAI PRECISION INDUSTRY CO LTD
  • US12272140B2 patent drawing
  • US12272140B2 patent drawing
  • US12272140B2 patent drawing

AI summary

A method for detecting the presence and the correct or incorrect placement of target objects in a container or carrier divides the container space into areas where the states of placement of the target objects in the container can be recognized, to generate N number of sub-areas, N is a positive integer. An artificial intelligence model is obtained by training the same according to training images of objects, the training images being images of the N sub-areas. The images are input into the artificial intelligence model and the states of placement are determined. The disclosure also provides an electronic device and a non-transitory storage medium.