42 Technology and Partners Develop Automated System for Pharma Manufacturing

42 Technology (42T) has partnered with Balluff, Synaptics, and Arcturus Networks to develop and demonstrate a new approach for AI-powered automated line clearance in pharmaceutical manufacturing.

With input from pharmaceutical manufacturers, the partners have developed a lower cost, flexible, and scalable alternative to existing AI-powered automated technologies. Most dedicated systems designed to automate line clearance processes are too expensive for widespread use because of their high hardware costs, complex integration, and the need for extensive training on production lines.

42 Technology has built a technology demonstrator (shown) to showcase this new AI-driven approach for automated line clearance. The unit will be launched at this year’s Embedded World event. Courtesy of 42 Technology.

“With some pharmaceutical manufacturers operating up to 100 production lines with 20 or more cameras on each, cost and complexity have been major barriers to automating line clearance,” said David Sandner, product marketing manager for machine vision at Balluff.

The collaborators' approach addresses these challenges by using a fully autonomous, deep-learning AI platform. Unlike conventional systems, the inspection system is designed to enable in-process learning without the need for lengthy setup and training routines on live production lines.

42T has designed and built a technology demonstrator that showcased the AI-driven system at Embedded World earlier this month. The AI-powered approach combines Balluff’s high-performance image acquisition system with Arcturus’s deep-learning models and the Synaptics AstraTM SL1680 edge AI processor.

The platform eliminates extensive on-site training requirements by using deep learning as a stand-alone method for continuous model improvement. This creates a fully automatic process where datasets are continuously developed in real time.

Further, the system’s embedded intelligence compensates for variables such as shadows and changing light conditions, while its unique self-learning capability continuously improves accuracy and reliability during real-time operation to further boost production efficiencies for pharmaceutical manufacturers.

The technology could also be used in other regulated industries such as medical devices, and food and beverage; the system uses fixed position cameras at key locations for monitoring, and is flexible enough to accommodate ‘pan and tilt’ cameras to scan different production areas depending on the application.