According to data from the Internet of Things Observatory of the Politecnico di Milano, the IoT market in Italy (in 2017) reached 3.7 billion euros, with a 32% growth from the previous year. With the increasing spread of the number of smart objects companies, not just in Italy, are starting to glimpse the potential of data made available by smart devices. Proof is in the market launch of a rising number of new IoT solutions that integrate advanced data analysis platforms with Artificial Intelligence algorithms (AI) capable of simplifying the management of connected devices, a traditionally complex issue particularly with the increase of smart objects.
Imaginable applications capable of integrating IoT and AI are innumerable and will have a radical impact on companies, public authorities and consumers. An example is fitting rooms with transparent and touch displays (Smart Retail) that can, not only provide any information required by the user in real time, but also understand their preferences and show products of greater interest over time through new Artificial Intelligence algorithms. Another example is the possibility of implementing wearable devices in industry sectors, such as within a factory (Smart Factory), to collect information on the work environment and autonomously understand if workers are exposing themselves to a dangerous situation and promptly warn them.
Over recent years we have witnessed the inception of startups focused on developing platforms / hubs enabling multi-protocol integration of heterogeneous devices to manage complex scenarios. These solutions – and in general leveraging the “cloud” for interoperability – are still valid but they will be boosted even more by new Artificial Intelligence algorithms. Something like what already happens today between different PC applications: based on email contents and user interests an AI agent can automatically set an appointment in the calendar. This approach could be replicated in the Internet of Things world. To manage smart devices within the household, for example: a hub that decides when it is best to run the pre-loaded washing machine or turn on the heat, based on users’ habitual behavior. Or in a factory: a platform that, depending on input data detected from the number of units going through the production lines and based on the number of incoming orders, decides when is the best time to change the manufacturing schedule, thus streamlining costs and service levels.
Therefore, there is some initial evidence of Artificial Intelligence (AI) applications within IoT solutions, but there is still a long way to go: more often than not, today there is a tendency to limit the range of action of IoT systems to automating simple existing functionality or remotely managing connected devices without exploring more advanced scenarios. The use of machine learning and other similar techniques on which Artificial Intelligence is based will in fact acquire an increasingly important role to combine more traditional services with new logic capable of meeting – and in many cases anticipating – the needs of users and consumers. Tangible examples are the development efforts in recent months of both OTT (Over The Top) and startups, that tend to progressively leverage information generated by components and sensors connected to the web to build advanced and predictive analyses. This is all extremely fascinating but obviously gives way to important cyber security and privacy issues.