Focus on First Open Call winners: Evolution experiment to migrate and adopt an Edge-Computing architecture into our AI-powered automated digital microscope for the agrifood processing industry.

WHAT ABOUT THE EXPERIMENT? 
 

There are many different analyses in the agrifood processing chain which are done with traditional microscopy with the aim to detect, count, or even classify microscopic particles/microorganisms in the samples. Microscopy is key in the food manufacturing industries. Typical examples are those with active fermentation processes within the manufacturing process, for instance. But there are many other examples like detection of contaminants, quality assessment, or microorganisms counting, in which traditionally there is an expert technician, observing a sample in detail through a standard microscope, during long and exhausting periods (sometimes even an hour per test!).

These traditional operations in the food manufacturing industry are labour intensive, tough, repetitive, must be carried out by hand by widely trained experts, and are expensive if externalized to specialized labs. In addition, when these kinds of analysis are externalized to specialized labs, the manufacturing companies receive the results deferred (days, or more than a week), so they lose real-time control of the product, and the process.

All these microscope analyses within the food manufacturing industry are perfect candidates to integrate autonomous AI-powered devices (coupled with image processing algorithms and trained neural networks) to allow faster and cheaper operations, while increasing control of their products. Unfortunately, the commercially available solutions of autonomous scanning microscopes are too expensive for the food sector (typically ranging from 25k€ to 120k€), since they are generally conceived and designed by large firms specifically for the MedTech and BioTech industries, which are more complex/demanding, and typically count with larger purchasing budgets for this type of laboratory equipment.

In view of this clear business need for affordable, easy-to-use and autonomous scanning microscopy device, at Microfy Systems our team started by 2020 to design and manufacture specific solutions for the automated quality control specifically designed for the agrifood industry, by means of robotizing a basic digital microscope (to act as a self-driven autonomous device), coupled with an AI-based image processing pipeline hosted in the cloud, which acts as the “intelligence” of the system. Our solutions are specifically conceived for non-lab end-users.

Considering our technology platform (robotized hardware + AI software) we aim to assist humans within the control checks usually performed in the food manufacturing industry, with different branded solutions for each different application/market.

In AI REDGIO 5.0 experiment we aim to evolve some of our devices to an edge-hybrid architecture.

​

WHAT IS THE EXPECTED IMPACT?  
 

With this project, it is envisaged 3 different types of impacts:

COMMERCIAL IMPACT: With a stand-alone device we would be able to reach new applications and collaborations faster, since we are not dependent on cloud architecture, nor variable costs of on-demand GPU. There are specific applications for laboratories in the public sector that demand for standalone alternatives, as well as partnerships with companies that have already developed the AI pipeline for other applications but are very interested in collaborating with us in regard to the automated affordable HW. Thanks to this experiment we could definitely migrate to a full edge-computing architecture.

ECONOMIC IMPACT: The impact on the economic side would be related with the reduction of variable costs per analysis on a AWS hosted full AI-pipeline. By reducing the requirements of GPU on AWS our company would we able to increase profit margin per analysis and offer the service even cheaper to clients, thus increasing the number of customer portfolio.

 Impact on our clients, food manufacturing SME

DIGITALIZATION IMPACT: As previously summarized, our technical solution addresses focus area of Human-machine co-working, in which there is a smart use of an automated robot to avoid tough, repetitive labours in the companies, with low added-value. Our solution assists food processing companies to improve their workplaces, increase control of their production processes with almost real-time data, and facilitates human work to implement real added-value tasks.

The food sector is very traditional and has strong resistance to adopt new technologies and approaches to their daily operations, mainly due to the low commercial margins considered in this industry. The reluctance to invest in emerging technologies such robotics, AI, edge-computing, etc. is very high in this sector, but the potential advantages for them are very high, such as productivity increase, product’s control increase, and also economic savings.