Machine vision system for mega-casting inspection

Fast and reliable mega-casting inspection

Mega-casting is revolutionizing automotive production, but quality control of the large aluminum castings remains a manual, error-prone, and time-consuming process.

The complex geometry and the glossy aluminum surface of the mega cast make visual inspection challenging. Defects as small as 0,1 mm can be concealed in deep “pockets” of the cast that are difficult to inspect with the human eye.

JLI vision has developed a unique automated inspection system for surface defect detection and dimensional measurement. The mega-cast inspection system uses advanced machine vision based on 2D, 3D, and AI models, capable of inspecting a mega-cast in 80 seconds.

Reliable and robust

An advanced AI-based vision system detects even the smallest defects.

Automated inspection

A two-step process enables a fully automated inspection process.

Faster production cycle

The inspection process is reduced from hours to 80 seconds.

How mega-casting inspection works

To complete a thorough inspection in a short period of time, the system uses a two-step process that combines 2D and 3D scanning with multiple robot arms.

First, the cast is scanned with multi-purpose 2D sensor heads that carry multiple cameras and lighting systems.

Using an advanced AI model, this step detects potential defects. Subsequently, the mega-cast is scanned with 3D sensors, both laser line and area scan, and AI is used to classify defect types.

The key to achieving robustness in a complex inspection process like this is to employ multiple approaches for each inspection type and run several specialized inspection processes simultaneously.

By utilizing the 3D sensors for dimensional measurements as well, we can perform this inspection task in the same session as the surface inspection.

This way, we can add dimensional measurement early in the production process, instead of waiting for the more time-consuming inspection with a 3D measuring machine at the end of the production line.

System specifications

Automated inspection of numerous defect types: coldflow, porosity, flashes, misruns, cracks, etc.
Detects defects from 0,1-1 mm
Dimensional measurement
Inspection cycle: 80 seconds
Up to 5 robot arms
Up to 9 cameras per robot arm.
Multiple lighting systems
Combination of 2D and 3D vision
AI-based classification