Backdrop & Context
Automotive manufacturing demands unwavering precision. For TI Fluid Systems, a global supplier of brake, fuel, and fluid delivery systems, even a 1 mm hole in steel coils can cause safety-critical failures. Based in Liège, Belgium, the company faced ongoing challenges with its previous camera-based inspection system. The system struggled with calibration, generated numerous false positives, and sometimes missed actual defects, leading to potential downstream production issues. Operating at a blistering speed of 95 meters per minute, these steel strips required real-time defect detection without bottlenecks.
Who’s Involved?
To solve this precision conundrum, TI Fluid Systems turned to InDeep-solutions, a Belgian company renowned for its expertise in industrial process control and image-based inspection systems. Collaborating with them was phil-vision GmbH, a German firm that specializes in machine and computer vision technologies. Teledyne DALSA, a Canadian imaging solutions leader, provided the central piece of hardware: the AxCIS mono contact image sensor, originally engineered for printing and battery inspection.
Arnaud Philippart, CEO of InDeep-solutions, led the project alongside experts from phil-vision who contributed vital knowledge on camera selection, feasibility testing, and driver support. The cross-border technical collaboration proved pivotal.
What’s at Stake?
Defects in steel coils can compromise brake and fuel pipe systems, posing major safety hazards in vehicles. Missing a single hole could result in an entire batch of unusable material or, worse, defective components reaching the market. The cost implications range from halted production lines to expensive recalls and tarnished reputations. Furthermore, automotive clients demand 100% precision, leaving no room for manual error or visual inspection fatigue.
From an environmental standpoint, false positives lead to unnecessary material waste, increasing carbon footprint (CO₂ emissions) and operational inefficiencies. The new solution needed to be robust, precise, automated, and compliant with rigorous industry standards.
Current Development or Announcement
The final solution integrates Teledyne DALSA’s AxCIS contact image sensor inside a compact housing mounted above the fast-moving steel strip. Below, engineers installed a ribbon LED light to backlight the strip, dramatically enhancing contrast and hole visibility. The sensor runs at a high acquisition frequency of 10kHz and achieves a resolution of 0.15 mm in travel direction and 0.022 mm across width. It processes in 4,000-line batches and responds within 0.4 seconds, ensuring no delay in high-speed manufacturing.
The system uses Teledyne’s Xtium2 CameraLink HS frame grabber, which simplifies installation by reducing cabling and alignment needs. It operates autonomously via software created by InDeep, which starts automatically on boot and flags defects in real time with a red square overlay. Images are archived for six months, facilitating traceability and audits. The system was calibrated once during commissioning and now runs with minimal human intervention.
Reaction from Public or Experts
According to Philippart, “Although the AxCIS contact image sensor was initially made for clean print lines, its ultra-high resolution and frequency made it ideal for small-hole detection on steel.” Experts from phil-vision called the project a benchmark example of industrial repurposing, where a technology designed for one industry elegantly solves challenges in another.
Manufacturing analysts applauded the partnership’s foresight in using contact image sensors, as opposed to conventional overhead cameras. They highlighted the reduced CPU load, seamless integration, and plug-and-play usability of the new system. Industry stakeholders believe such sensor-based inspection systems will become the gold standard, especially as automotive materials grow lighter, thinner, and more complex.
Comparison with Past Events or Global Trends
Historically, the automotive sector has leaned on traditional camera-based vision systems, but these often demand frequent recalibration, manual checks, and suffer from reduced accuracy at high speeds. Globally, industries are shifting toward sensor fusion, integrating lighting, sensors, and smart software, to automate inspection.
In Asia and the U.S., major steel manufacturers are exploring AI-powered defect detection using neural networks. However, these often require GPU-intensive systems. In contrast, InDeep’s solution uses edge computing principles, analyzing images locally and in real time, thus reducing energy use and maintenance costs.
The transition toward Industry 4.0 has triggered growing interest in modular, scalable systems like the one implemented at TI Fluid Systems. It exemplifies the convergence of mechanical engineering with vision science.
Future Implications & What to Watch For
With the TI Fluid System installation deemed a success, the company has already ordered a second AxCIS sensor as a strategic spare. Philippart has indicated ambitions to replicate the model in other factories and industries, particularly where thin, fast-moving materials require micro-defect inspection.
There’s potential for this tech to evolve further with AI integration, enabling pattern recognition, predictive maintenance, and remote diagnostics. As automotive OEMs demand zero-defect supply chains, suppliers like TI Fluid are setting benchmarks for others to emulate.
This successful implementation might inspire similar upgrades in industries like packaging, electronics, and aerospace, where defect detection must be both granular and instantaneous.
Key Takeaways:
• TI Fluid Systems faced accuracy issues with an older camera-based steel strip inspection system.
• InDeep-solutions and phil-vision collaborated to install Teledyne DALSA’s AxCIS contact image sensor.
• The system operates at 95 meters per minute and detects holes as small as 1 mm with near-perfect accuracy.
• A LED ribbon backlight enhances contrast, allowing clear image capture even at high speeds.
• Custom software auto-launches, flags defects in red, and archives images for six months.
• The project demonstrates how cross-industry innovation and sensor repurposing can solve precision manufacturing challenges.
