Wire Harness Optical Inspection Using ML-Based Feature Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional Automated Optical Inspection (AOI) systems for products like wire harnesses are inflexible, require extensive manual programming, and struggle with adapting to changes in environment or product variations, leading to inefficient and inaccurate inspections.
Innovation Solution
A machine learning model is trained to detect acceptable and unacceptable features of wire harnesses using a custom-engineered optical system with a camera and LED ring light, enabling continuous learning and adaptable inspection through video-based AI that identifies regions and makes mathematical decisions based on probability calculations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional AOI systems use static dimensional tolerances and manual programming, then inspection coverage is comprehensive, but the system lacks adaptability to environmental changes and product variations
Solution Approach 1:
The patent applies dynamics by transitioning from static dimensional tolerances to dynamic, learned tolerances. The ML model continuously learns from training images to adapt to environmental changes and product variations, making the inspection system flexible rather than rigid. The system dynamically adjusts its understanding of acceptable variations based on learned patterns from diverse training data.
Solution Approach 2:
The patent replaces the mechanical programming system with an intelligent learning system. Instead of manually coding inspection rules and tolerances, the system uses machine learning models that automatically learn inspection criteria from training images, substituting computational mechanics with cognitive processing.
2Productivity
If conventional AOI systems require extensive manual programming, then inspection rules can be precisely defined, but inspection efficiency and speed are reduced
Solution Approach 1:
The patent applies preliminary action by performing the learning process in advance during an offline training phase. The ML model learns from extensive training images before actual inspection, so that during production inspection, no time-consuming programming or rule-setting is needed. The system is pre-adapted to various product variations, enabling rapid inspection without setup delays.
Solution Approach 2:
The patent implements self-service by enabling the system to automatically learn and define its own inspection criteria without human intervention. The ML model autonomously processes training images to establish acceptance criteria, eliminating the need for manual programming and reducing dependency on expert knowledge for system configuration.
3Measurement precision
If conventional AOI systems use wider static tolerances to accommodate part-to-part variation, then fewer products are rejected, but measurement precision and quality control are reduced
Solution Approach 1:
The patent applies local quality by enabling different precision levels for different features and contexts. The ML model learns feature-specific tolerances from training data, allowing tight tolerances for critical features and broader acceptance for non-critical variations. Each product feature is evaluated with locally optimized criteria rather than uniform global tolerances.
Solution Approach 2:
The patent implements parameter changes by dynamically adjusting acceptance criteria based on learned patterns. Instead of fixed dimensional tolerances, the system adapts its evaluation parameters based on the specific product being inspected and the variations learned during training, optimizing the balance between precision and acceptance rate for each inspection context.
4Ease of operation
If human inspectors perform manual inspection, then complex judgment calls can be made, but inspection is time-consuming and labor-intensive
Solution Approach 1:
The patent applies universality by creating a single automated system that performs multiple inspection functions simultaneously. The ML model evaluates multiple features and criteria in parallel, replacing multiple human inspectors or sequential inspection steps with one unified system that maintains comprehensive evaluation while dramatically increasing speed and consistency.
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
The system significantly improves inspection efficiency and accuracy by automatically distinguishing between acceptable and unacceptable features, reducing human error and the need for manual programming, and can be applied to mission-critical systems like high-voltage autonomous vehicles.
Implementation Method 1
a light (e.g., an LED ring light)
Data Source
AI summary
Methods and systems for inspecting a product, such as a wire harness, including product features for inspection. A camera of an inspection station may capture a product image. A machine learning (ML) model may detect one or more objects in the captured product image and provide, for each detected object, an identification of a class of the detected object and an identification of a region of the detected object in the captured product image. The class of the detected object may be either an acceptable product feature class or an unacceptable product feature class. The inspection station may display an enhanced product image that includes the captured product image to which the identification of the class of the detected object and the identification of the region of the detected object in the captured product image for each detected object have been added.