Automation and AI-based wrinkle detection in robotic seat steaming become particularly challenging when different seat models with textile, vinyl, and leather covers are processed on the same production line.The methods used to remove wrinkles and the optimal control parameters for the steaming robot vary depending on the material – and this is where the real challenges for quality and process efficiency begin.
As a result, quality managers often report high rework rates, increasing process costs, and a strong dependence on experienced operators when optimizing AI-supported steaming processes. Although robots are already installed in many plants, they typically steam the entire seat surface regardless of whether wrinkles are actually present.
This, in turn, leads to unnecessary energy consumption, increased material wear, and avoidable process inefficiencies. With a fixed cycle time of only 53 seconds, there is virtually no time for manual corrections.
This solution builds on technologies developed within our VISION research project for AI-supported visual quality inspection in industrial production.
AI-based optimization of the seat steaming process in automotive manufacturing
The SCIIL VISION (VisuSteam) system combines camera-based AI wrinkle detection with direct integration into robot and production line control systems. It automatically identifies both the seat model and the cover material – whether leather, vinyl, or textile – and selects the optimal steaming parameters for each individual seat.
As a result, the system creates a self-learning closed-loop process that continuously refines robot control parameters and improves quality from one production cycle to the next.
The complete process consists of the following steps:
- Line control provides seat information, including the unique serial number, model, variant, and trim.
- Cameras capture high-quality images under controlled lighting conditions.
- AI detects and classifies wrinkles based on their position, direction, and severity.
- The system assigns ironing zones together with the required process parameters (pressure, priority, and number of steaming cycles).
- The robot steams targeted zones according to zone-specific instructions.
- Afterwards, AI verifies the steaming results.
- Whenever necessary, the control parameters are automatically adjusted, creating a continuous optimization loop. (see also: AI-supported self-learning loop)
Consequently, production and quality managers benefit from a solution that not only automates seat steaming but also improves process stability, quality, and efficiency over time. This makes the system particularly suitable for high-volume production lines with demanding quality standards.
Implementation and Boundary Conditions
The production line processes numerous seat models featuring textile, vinyl, and leather covers in a wide variety of shapes, designs, and seam layouts. Although the overall cycle time is fixed at 53 seconds, only 48 seconds are effectively available for the steaming process.
Under these conditions, the key challenge is to detect wrinkles reliably, apply the optimum steaming parameters for each material and ironing zone, and continuously improve process performance through AI-based wrinkle detection and intelligent robot control.
Verification Test After Project Completion
The final validation included a one-hour production test with 63 seats:
- Approximately 99% of all wrinkles were detected correctly.
- The false-positive rate remained below 6%.
- Only 0.28–1.58% of the total seat surface was steamed unnecessarily
Furthermore, the solution significantly reduces manual intervention, lowers rework, and protects sensitive materials such as leather from unnecessary steaming.
Conclusion
SCIIL VISION reduces the workload for production and quality teams while delivering standardized, reproducible results.
At the same time, the self-learning control loop continuously optimizes the process, making the solution highly scalable for additional production lines.
Key benefits include:
- Lower Cost of Poor Quality (CoPQ)
- Reduced unnecessary steaming
- Complete process traceability
- Standardized and reproducible quality
- Easy rollout to further production lines
