Machine dialogue

Machines can provide a lot more data than we have been used to if we listen to them carefully; thanks to the development of smart technologies, today we can navigate this sea of data in search of Unknown patterns and find the logics governing Big Data. Assembling and cross-checking this data to get even more information, in a process that is yielding important results in the fields of energy and mobility, is also possible. This is an important step towards achieving smart grids and smart cities.

USA, Massachusetts, Boston, detail view of Financial District office building, dusk

Machines with their own digital identity, networking and interacting with each other through a connection capacity and widespread processing: these are the new 2.0 machines, able to connect independently, which process data and enable new functions. At the end of the nineties, the first scenario where Machine to Machine communication allowed for capturing and processing an immense amount of data in real time was developed with the evolution of distributed sensor networks, like a sort of digital voice of the machines. Even today, listening to machines enables many functions and applications, and in the future a new potential, which still has to be discovered, will be activated, thereby transforming our homes, our cities, and our power plants into ‘sensitive’ environments. Among the many existing, there are two elective areas of these applications: the field of energy, where listening to machines already allows for the optimization of the management and maintenance of the generation and distribution systems, and the field of urban mobility, where new applications and uses will transform our cities and our lifestyle.

Energy signals

Thousands of new plants powered by sun, wind, water, and biomass have revolutionized the world of electricity generation: among these, wind energy is the source in Italy that has seen the largest growth in the last decade. Enel Green Power (EGP), one of the major global players, with over 750 sites of wind power production glob- ally and a total of 5,000 wind turbines installed to date, has equipped each of these with about 500 sensors. A network has been made up of approximately 2.5 million sensors that detect magnitudes of different nature and send millions and millions of bits of data without any temporal interruption. Just as what happens with the human cognitive system which detects, processes, and manages millions of signals and then stores them as experience, likewise millions of signals travel from the network of EGP sensors towards infrastructures dedicated to storing them, thus effectively creating a wealth of experience now known as Big Data. Processing the data acquired by machines, together with environmental data, allows building up models which today aim to understand the past but which will be able to provide us with useful elements to predict the future, projecting it together with the demand in the ecosystem of the maintenance which, in fact, has shifted from the models of forecasting to models of the previously inconceivable nowcasting. This paradigm shift is radical, especially in the approach to the analysis of the data, in that it passes from samples of aggregate data, theoretical models of analysis, and inference for monitoring and making decisions to that of the detailed analysis of real events within the entire universe of data. Big Data is nothing more than a large container of unstructured information coming from different sources. The concept of Big Data Analytics concretizes its added value, allowing for the real time analysis and interpretation of the entire universe of available data: from that which is already organized and structured – such as geo information and the assets of the fleet – to that which is unstructured, highly variable, and unpredictable – like the weather, machine data, and the price and demand of energy.

It is a radical change of paradigm: thanks to the development of smart technologies, today it is possible to find the logics governing Big Data

Thanks to the exponential increase in computational performance, followed by ever-increasing cost savings, the development of cloud computing and programs able to navigate this sea of data in search of unknown patterns, it is possible to think of finding the logic that governs Big Data. The information collected can be made available through Machine to Machine dialoguing, for example, by sharing the fault models, thus equipping the entire system of collective intelligence with advantages in terms of savings and optimization of resources. The signals processed in real time may show operating conditions that are potential causes of blocking production well in advance compared to the past; this technique of Predictive Maintenance is now a reality in Enel Green Power through its Predictive Analytics project. Therefore, the simple control of individual parameters such as operating temperatures, vibrations, or an alert etc., has been exceeded in order to find correlations be- tween phenomena that we were not aware of, by providing real-time data through a continuous dialogue with each machine.

Traffic signals

Over the past couple of decades, cars have evolved from the mechanical contraptions that Henry Ford imagined into veritable computers on wheels. Today’s automobile is bristling with hundreds of onboard sensors that monitor passengers and surroundings, and often interfaces with digital systems through 3G/4G, short-range wireless and GPS. In addition to moving the population from origin to destination, urban vehicles have become a networked fleet of mobile sensors that suffuses the city and generates a tremendous amount of data: Traffic Signals. Collecting, analyzing and interpreting these signals – what could be thought of as ‘listening’ to our vehicles – is leading to a better understanding of our cities, our streets and our populations… ushering in an era of Intelligent Transportation Systems.

Listening to machines allows optimizing the management of energy and urban mobility generation and distribution systems

Even a relatively simple and familiar technology like GPS provides expansive opportunities for research: urban traffic flows can be tracked, analyzed and modeled. For the first time, real-time maps of human mobility can be drawn, unveiling the living pulse of the city. Applications of data analytics could be as diverse as real-time travel time prediction, pothole and noise pollution detection, and more. One area of active research is vehicle sharing. ‘Listening’ to digital traces can show us how vehicles move, but those patterns themselves may be changing. Thanks to real-time fleet management systems, cars can be accessed by a community of users: an example of the ‘sharing economy’ that is emerging in many domains today. Rather than owning and leaving a car idle in a parking lot for 23 hours each day, vehicles can be shared and used on-demand. Simply ob- serving this trend – that people are readily willing to share through digital platforms – is opening fertile lines of research based on urban mobility data that will increasingly help the adoption of electric vehicles interacting with smart grid recharging stations. A recent project applies an unprecedented dataset – over 150 million digital traces from New York City’s taxi fleet – to the question of shareability. If passengers were willing to share rides and still arrive on time (within a window of 5 minutes), what is the minimum number of cars required? A network-science approach led us to propose the notion of a shareability network, a mathematical model of sharing opportunities representing rides as nodes and the possibility of sharing as a link connecting two nodes. The results are staggering. An empirical analysis of the data demonstrates the significant potential for ride sharing in New York City. Over 95% of taxi trips can be combined, with a minimal passenger inconvenience (a maximum of 5 minutes delay, as compared with a solo ride). More generally, the results suggest that urban mobility is intrinsically social: populations tend to move between similar origin and destination points at similar times.

For the first time, maps of human mobility can be drawn in real time revealing the vital throb of the city

Data reveals the communities and patterns that have long occupied sociologists and urban designers. More importantly, this knowledge will become crucial when vehicles are driven by digital systems: not only will vehicles create data, but they will run on data. Far beyond their familiar role in futurology and science fiction, self-driving cars are quickly rolling into the marketplace. The major OEMs plan to integrate varying degrees of autonomy into their fleets as early as 2016. As autonomous vehicles suffuse our streets, real-time information transfer between cars, streets  and pedestrians will allow for optimization in a variety of ways, from reducing accidents to managing intersections for smooth traffic flow. Digital systems will then form a mesh between our cities, our vehicles, and even ourselves – what many call ‘smart cities’. As research in this field develops, data can reveal how we move and how we interact, even paving the way for new mobility paradigms.

by Matthew Claudel
Senseable City Lab, Massachusetts Institute of Technology

by Carlo Ratti
Senseable City Lab, Massachusetts Institute of Technology

by Paolo Santi
Senseable City Lab, Massachusetts Institute of Technology

by Roberto Tundo
Enel Green Power