From manufacturing to Artificial Intelligence to make the difference

I changed my domain, but my position at Oròbix is very connected to manufacturing. I work as a kind of translator, a sort of facilitator between the world of manufacturing and the world of artificial intelligence. I believe this is a crucial position for the real spreading of AI in our companies. I use my knowledge in manufacturing together with the passion for data science resulting from my engineering background. In my previous experience, my passion led me to deal with innovation and digital transformation, but above all to experience it by building sensors and writing code. Back to the reason for this change, I think it’s because I want to have an impact, to be an active player in this revolution happening across all sectors. I am taking Artificial Intelligence where it is not yet present.

Because I tested you guys and, like a good engineer, without data I couldn’t make such a big decision! I experienced Oròbix’s consistency by working on a computer vision project (editor’s note: we talked about it here: https://orobix.com/en/ai-observatory-2020-2021-business-case/) and I realised that there was a similar approach, an attitude to problem solving and a great desire to get things done. Everything I needed to make this leap.

For me the AI is an “enabler of continuous improvement” (editor’s note: 😮). In a company people often waste weeks chasing after specific solutions, applying patches, without taking time to look outside the box and think about solutions that can really add value. We don’t have all the time in the world for one single problem! Artificial Intelligence is speeding up this process and allowing people to use data to objectify their expertise, two different kinds of intelligence coming together.

This is a crucial moment for Italian manufacturing, where competition and future opportunities are under discussion. Artificial Intelligence plays a fundamental part in this. Like any big innovation, it takes time before people understand the real benefits for companies. It takes patience to help customers discover Artificial Intelligence and, above all, it is necessary to let them experience its potential in person, and to adapt use cases to the companies’ real problems. You can’t remain generic and you have to go deep. This is exactly what Oròbix AI Solutions are for: to experience AI, to see it first-hand on specific business problems. A very important change that overcomes the traditional approach: “feasibility / PoC / model development / production deployment”. We start directly from the end, we go to production with a model to be specialised according to the customer’s data directly collected from the production line. We take risks, but we make the solution more reliable and, above all, we test its limits and potential together with the client. Excluding the customer, hiding behind the idea that we are specialists, doesn’t help the process of comprehension instead it creates diffidence. The current economic scenario requires us to be agile, very effective and to share strategies, risks and results with the client.

I understand the trouble on both sides. I’ve been there. Let me try to give you a practical example. Detectiv.ai, our solution for anomaly detection on data from IoT sensors, can be proposed in many ways, but not all of them are effective and comprehensible to the manufacturing world. If you remain generic, the client probably will not understand you. But if you try to get down to specifics, for example by showing a video of a milling operation with chips- you have to know that chips are  very important for those who work in the mechanical processing – and the related data collected with a simple accelerometer and showing the correlation between machining and data, everything becomes simpler. Basically, data analysis, anomaly detection and prediction tools are nothing more than the scientific objectification of input/output correlations: an incorrectly tuned tool will cause non-conforming machining. I like to see the physics principles in the data. These kinds of notions, acquired through experience, are the day-to-day life of skilled workers in our factories, but it’s challenging to teach them to an Artificial Intelligence system without being familiar with the domain. This is a journey of discovery, it takes patience on both sides, but it is possible and above all necessary. We are lucky to have someone on the other side who understands the importance of teamwork between humans and new technologies. People with the help of AI become capable of doing things better and sooner, but we have to experience it to generate trust!

First of all, I would suggest getting his/her hands dirty. Go to the shop floor, talk to the workers, drill down to the heart of the problems, intercept bad feelings at all levels. When I see a machine suffering I have to act! Comprehending a mathematical theory is not so different from  understanding how a mechanical machine works. We are always talking about human genius! Don’t remain superficial and strive for a comparison. Otherwise, you risk that models remain just cold mathematical functions disconnected from  reality, stylistic exercises that do not really help anyone. Then I suggest they find their own voice, a way to make their work comprehensible at all levels, from production to management. Communicating results and methods is the only way to ensure that projects do not remain in a box but become operative. The only way to have a real impact and generate value for all.

Keep informed as much as possible, today it’s straightforward to access technical information. Without awareness, you can’t make these decisions and you risk following a trend without understanding its value. And then make experiences, go out in the world. Ask for transparency, ask for monitoring systems. Involve the appropriate staff, including front-line workers, but also those in charge of innovation, because sometimes we just have to talk together to design solutions that can quickly scale up from simpler, short-term problems to more complex situations.