Terry Dodson, the Director of Engineering at New Frontier Technologies, discusses the cutting edge of technology for industrial automation and controls with Lonnie Worthington, the Motion Control & Vision Group Manager. This project uses Motion Vision and AI to detect Foreign Objects on Conveyor Belt systems. New Frontier Technologies trained a learning algorithm, aka a neural network, to detect protein sources and foreign object particles like bag tails, metals and plastics.
Working with protein products can present unique quality, safety and sanitation challenges to food processors.
Client Challenges
For one of NFT’s clients, detection of physical contaminants was difficult when utilizing conventional human and metal detection systems because of the high variability in the appearance of the products. This can result in costly misidentification of foreign objects and unnecessary conveyor shutdowns. NFT’s client wanted to ensure that the most advanced technology was integrated in the production process in order to differentiate between “product” and “non-product” objects.
NFT Solution
Working closely in tandem with their client, NFT considered multiple “detection” and “inspection” technology solutions.
Detection Solutions:
- Standard Vision System Tools
- Hyperspectral Imaging
- Convolutional Neural Network (CNN)
Inspection Solutions:
- Overhead Imaging
- Waterfall Imaging “whole object”
After developing baseline specifications, detailed research and ”worst-case” performance testing was conducted on the detection and inspection options. NFT and their client determined that implementing a CNN-enhanced vision detection system with simultaneous Waterfall “whole object” inspection methodology would maximize the benefit of Machine Vision enhanced with AI/Deep Learning, while leveraging the experiential value of human inspection and decision making.
Client Outcomes
By implementing the CNN & Waterfall inspection technology, NFT’s client was able to significantly enhance the efficacy of their inspection methodology, while simultaneously reducing the rate and corresponding high costs associated with foreign object “false positives”. This allowed for critical “detection and inspection” to be positioned for future operational advancements and regulatory requirements.
Key Performance Data Conclusions
- Minimizes need and variability of supervisory oversight
- Reduces expense associated with product rejections and spoilage
- Allows for Preemptive Control, Command and Reduction of costly production line shutdowns
- Provides protection of company brand reputation and mitigation of legal and recall potential
New Frontier Technologies is an end-to-end industrial systems integrator, offering turnkey operational & technology solutions to the Food & Beverage industry. If you'd like more information about how NFT can develop solutions for your food or beverage plant that uncover and integrate real-time operational data with business information, helping you improve profits, please get in touch with us!