360° Defect Detection for Starter Switches
Related Articles
Our Customers
Introduction
A key player in the automotive technology industry required a solution to identify defects in their starter switches. The specific area of concern was the caulking—a process where the metal parts get blended into the starter switches. This process could lead to defects such as cracks at the edges and knots (corners) of the switches, compromising the integrity and performance of the starters and potentially leading to product recalls and customer complaints.
Solution
To address these challenges, the company used the SwitchOn DeepInspect system. DeepInspect was integrated into the existing automation system using a PLC to communicate with the robot for picking and placing parts, as well as controlling a rotating motor.
System Hardware Specifications:
DeepInspect Pro Software
Basler acA2440-35uc Camera
Mitsubishi PLC
TMS industrial machine vision lights
Controller with i5 and Nvidia GPU
Integration with existing automation system
Results
Utilizing DeepInspect’s advanced vision technology, defects as small as 100-150 microns were detected.
The entire inspection process was automated with a robotic arm removing the rejected parts, reducing the need for 2 dedicated manpower.
Inspection completed in less than 11 seconds per part, ensuring high efficiency.
DeepInspect was able to identify the defects caused by the wear and tear of caulking tools. This enabled the company to address the issue promptly and reduce defects.
DeepInspect was able to seamlessly inspect two variants of starters, one with 4 knotches and the other with 6 knotches.
The system ensured that the starter switches met the required quality standards by detecting and rejecting defective parts.
Conclusion
The implementation of SwitchOn’s DeepInspect system at this leading automotive technology supplier significantly improved the defect detection process for starter switches. By automating the inspection process and utilizing advanced vision technology, the company was able to maintain high-quality standards, reduce manpower costs, and increase production efficiency. This case study highlights the effectiveness of integrating automated defect detection systems in the automotive manufacturing process.
Let's Discuss How We Can Transform Your Operations!
FAQ’s :
DeepInspect uses readily available industrial-grade hardware to ensure high repeatability and long lifespan. We provide a basic kit to help you get started, which includes a controller, camera, lights, and a PLC.
DeepInspect supports inspection speeds of up to 1000 parts per minute. The final speed depends on various factors, such as the number of cameras, lighting, and other line conditions. If you have a requirement above 1000 PPM, please contact us.
DeepInspect supports all GenICam-compliant USB 3.0 and GigE cameras, giving you the flexibility to work with the hardware you already have on your production line. Fully tested and compatible brands include Basler, IDS, Baumer, Omron Sentech, Allied Vision, FLIR, HIK Robot, and Lucid Vision. We also support Area Scan, Line Scan, and Thermal cameras, covering a wide range of inspection environments and product types.
DeepInspect is built specifically for defects that are random and unpredictable in nature, such as cracks, scratches, surface anomalies, missing components, seal failures, and label errors. Unlike rule-based systems that rely on a defect appearing at a fixed location or in a fixed pattern, DeepInspect uses AI models trained on your actual production data to recognize defects regardless of where they appear or how they look. This is exactly where AI inspection becomes critical: when a crack can appear anywhere on a part, or a scratch has no consistent shape or size, traditional vision systems fail. DeepInspect is trained to catch what others miss. The system has been successfully deployed across over 1,000 unique SKUs in automotive, pharma, electronics, and FMCG industries.
Curious to Know More?
Download the full case study and explore how our customer leveraged DeepInpsect to achieve zero defect lines