In pharmaceutical automation and manufacturing, maintaining the integrity of parenteral vials is critical to patient safety and regulatory compliance. Standard inspection methods were manual and required technicians to physically inspect each vial seal individually. This process was extremely strenuous and posed significant limitations in scalability, accuracy, and consistency. A pharmaceutical manufacturing systems integrator sought to alleviate this major bottleneck with their automated visual inspection (AVI) system. In this case study, we’ll explore how Premio’s edge computing hardware solution is enabling their AVI machinery with AI-driven defect detection for powder-filled parenteral vials.
Challenge:
The systems integrator recognized that implementing edge AI for defect detection in pharmaceutical automation was a complex challenge. They recognized that an industrial computer was needed and outlined a set of critical deployment prerequisites that needed to be met.
- Enable vision AI and unsupervised machine learning for anomaly detection
- Connect multiple high-definition vision cameras
- Aggregate store high-fidelity defect data for deeper analysis
- Durability and reliability for prolonged uptimes in factory environments
Solution:
Premio recommended the RCO-6000-CML-4NS, a high-performance, industrial-grade AI Edge Inference Computer tailored to meet the system integrator’s requirements.
- Real-time performance with a 10th Gen Intel Core TE Processor
- Modular EDGEBoost I/O with four RJ45 LAN ports to consolidate IoT cameras
- EDGEBoost Node support 4x NVMe U.2 SSDs for rapid data aggregation
- Super-rugged durability in harsh factory deployment settings
Benefits:
- 150,000 sq ft manufacturing scalability
- Proactive engineering and sales support
The Challenge
Traditional vial sealing inspection methods relied on technicians to visually examine each vial for defects. This approach was not only time-consuming but also introduced variability due to human fatigue and inconsistencies. The manufacturer needed an industrial computer to power their automated visual inspection (AVI) machine with AI and machine learning.
Complexities of Deploying Vision AI At The Edge
The system integrator's parenteral vial AVI platform required significant computing power to process intensive unsupervised machine learning and multiple high-resolution image streams simultaneously. Real-time performance was a necessity as defects needed to be ejected from the line within seconds of being inspected. Cloud computing was not an option as it introduced latency and consistent wireless connectivity to operate. This throughput requirement meant the edge computing solution needed to process multiple camera feeds while executing complex defect detection algorithms with sub-second latency.
Requiring IIoT Connectivity for Vision Cameras
The parenteral vial AVI system utilized multiple vision cameras to capture different angles of each vial seal for comprehensive inspection coverage. Each camera requires dedicated connectivity for high-bandwidth data transmission. This meant that the edge computing solution needed to offer compatible and the necessary amount of connectivity to support vision cameras and other IoT devices.
Necessitating High-Volume Data Storage
A critical requirement of the vial defect detection system was its ability to store large volumes of defect vial images for quality assurance, regulatory compliance, and AI model training purposes. While there are many server-rack configurations that can meet this demand, the manufacturing environment demanded a spatially efficient form factor that could retrofit into the respective space in the AVI. Additionally, the storage solution needed to have high-speed capabilities to aggregate and store data in real-time; preventing potential data bandwidth bottlenecks. With the sensitivity of the defect data generated, a form of redundancy is needed to safeguard mission-critical data and ensure the integrity of the overall system.
Requiring Industrial-Grade Reliability
Operating in a pharmaceutical manufacturing environment presents unique challenges for computing equipment. The solution is expected to have 24/7 operation capabilities and endure rigorous industrial settings while maintaining consistent performance. Key environmental considerations included resistance to dust, vibration, temperature fluctuations, and power inconsistencies.
The Solution
After understanding the system integrator’s AVI solution requirements, Premio recommended the RCO-6000-CML-4NS as its innovative features and industrial-grade design make it an ideal platform for AI-powered defect detection applications.
Enabling Real-Time Edge AI Performance
The RCO-6000-CML-4NS leverages a 10th Generation Intel Core TE processor paired with 64GB of DDR4 memory to deliver exceptional computing performance for AI-driven inspection workloads. The powerful processing architecture enables real-time AI inferencing for defect detection at rapid rates of up to 50 vial inspections per minute. The system's advanced processing capabilities ensure ultra-low latency for image analysis while efficiently managing multiple concurrent tasks including image acquisition, processing, and storage operations.
Modular EDGEBoost I/O for Seamless Connectivity & Compatibility
A key feature of the RCO-6000 Series is modular EDGEBoost I/O (EBIO) technology. This platform provides seamless I/O flexibility and compatibility to meet unique demand requirements. For the AVI system, the RCO-6000-CML is configured with an EBIO featuring four additional RJ45 LAN ports for direct connection to the multiple vision cameras. This modularity simplifies optimization and ensures support with standardized interfaces for seamless integration with various IoT devices.
EDGEBoost Node for Scalable & Redundant Storage
An industry-leading design approach that allowed the pharmaceutical integrator to avoid a costly OEM design, is Premio’s EDGEBoost Node technology. Without it, the system integrator would need to utilize server-rack solution that was not compatible with the available space or opt into an OEM design.
By following the same modular design principle as the EBIO, this technology enables configurable performance acceleration with GPU, NVMe storage, and/or PCIe expandability. The RCO-6000-CML-4NS integrates an EDGEBoost Node to support four 15mm hot-swappable NVMe U.2 SSDs, enabling the storage of terabytes of vial seal defect imagery. The hot-swappable feature allows operators to offload data with minimal downtime.
The RCO-6000-CML-4NS achieves high-speed data transfer rates with NVMe technology, eliminating storage bottlenecks even under intensive imaging workloads. To ensure data integrity, software RAID provides redundancy, protecting critical inspection data from drive failures while facilitating rapid data offloading.
Industrial Fanless & Cableless Design
The RCO-6000-CML-4NS features a rugged design centered around reliability and longevity. Its fanless cooling design eliminates major points of failure such as moving parts and prevents ingress of dust and debris. Pharmaceutical manufacturing operations are mission-critical and abrupt downtimes are costly and disruptive to production. Designed to withstand harsh deployment environments, the RCO-6000-CML-4NS ensures continuous, dependable performance with wide operating temperature ranges, MIL-STD-810G shock and vibration resistance, and wide power input ranges with built-in power protection. Additionally, UL certification confirms compliance with stringent safety standards, offering confidence for deployment in regulated manufacturing environments.
The Benefits
Premio not only delivered a fully optimized edge AI computing solution but also provided comprehensive support as a trusted business partner. With dedicated engineering assistance, scalable U.S.-based manufacturing, and ongoing technical support, Premio ensures a long-term partnership and with utmost reliability and quality assurance.
Comprehensive End-to-End Support
Premio's dedicated sales and engineering teams provided end-to-end support throughout the project lifecycle. This included initial consultation to understand specific requirements, optimization of firmware settings for the vision AI platform, and ongoing technical support to ensure optimal system performance. The proactive support approach minimized integration challenges and accelerated time-to-deployment for the AVI system.
Scalable Manufacturing Capabilities
Premio's 150,000 square foot manufacturing facility in Los Angeles, California provides the systems integrator with a reliable, scalable production partner capable of meeting growing demand for AVI systems. This domestic manufacturing capability ensures consistent quality control, rapid fulfillment, and the flexibility to scale production as market demand increases.