How a Robot System Inspects and Grades Objects for Flaws

This patent describes an automated system that uses cameras, special lighting, and deep learning AI to find and measure tiny flaws on an object's surface, then assigns a quality grade and sorts it.

What it covers

The system first uses an "image capture subsystem" to take pictures of an object, adjusting a "lighting subsystem" for optimal viewing conditions (Claim 1). An "image processing module" then employs "deep learning algorithms" to detect, segment, and classify surface defects, also measuring their length and width. For each identified defect, an "optical spot sensor subsystem" precisely measures its depth or height, providing a 3D understanding of the flaw (Claim 1). Finally, a "cosmetic grading module" determines the object's quality grade based on the number, type, and severity (length, width, and depth) of defects, and directs the item to the appropriate bin for shipment (Claim 3). For example, it could automatically inspect newly manufactured smartphone screens for scratches or dust particles.

What it doesn't cover

• —Does not cover systems that rely solely on human visual inspection for grading objects.
• —Does not cover systems that only detect 2D surface defects without measuring their 3D depth or protrusion (Claim 1).
• —Does not cover systems that grade objects without using deep learning algorithms for defect detection and classification (Claim 1, 2).
• —Does not cover systems that do not use an automated control system to align the object for 3D defect measurement (Claim 1).
• —Does not cover systems that do not recommend an optimal path of disposition based on the cosmetic grade (Claim 3).

The clever bit

The novelty lies in combining 2D image processing with deep learning for initial defect detection and classification, then precisely measuring the 3D depth of those specific defects using an "optical spot sensor" guided by the 2D analysis. This allows for a comprehensive and automated cosmetic grade.

Why it matters

This technology is important for quality control in manufacturing, especially for products where even small cosmetic flaws can impact value or customer satisfaction. It allows for consistent, objective, and high-speed inspection, reducing human error and labor costs. Industries producing electronics, automotive parts, or consumer goods can benefit from such automation to maintain high product standards.

Real-world examples

• 1.Automated inspection of smartphone displays for scratches
• 2.Quality control for painted car body panels
• 3.Sorting of pharmaceutical pills for surface imperfections
• 4.Grading of fruit and vegetables for blemishes
• 5.Inspection of semiconductor wafers for defects