Giorgio Metta he is 51 years old, he has been scientific director of the Italian Institute of Technology since 2019, but he began to be interested in robotics long before. Much earlier, but not early enough to see it born in Italy, as he told us with a smile: “In our country, what was then called cybernetics has been talked about for a long time, long before I was a researcher”. Since when? “The American mathematician Norbert Wiener (one of the fathers of this discipline, ed) was invited here at the end of the sixties, therefore more or less since then ”.
One of the most important moments in the history of Italian cybernetics was the Cnr Robotics Finalized Project, started in 1989 and over time involved over 600 researchers a year, engaged in the study and construction of prototypes that would allow to deepen the structure and management of robots and their control, also through sensors and actuators.
Clearly, the birth of the IIT (which turns 16 in 2021) accelerated all this: “We have a real robot family, which can be divided into 4 macro-areas – Metta explained to us – There is wearable robotics, that is, machines dedicated to rehabilitation, such as prostheses; there are robots capable of intervening on the scenes of a disaster or an accident, which can be quadrupeds or even centaurs with wheels, arms and hands, and those for the so-called precision agriculture ”. Again: “There are robots used for surgery and there is soft robotics, machines that simulate animals or plants, which have their own zero impact life cycle, which are capable of self-sustaining and do not pollute and can be used for environmental monitoring or space exploration “.
From iCub to R1, from baby robot to walking robot
The most famous of all is of course iCub: unveiled to the world in 2009, is a humanoid robot that looks like a 5-year-old child and in the latest version is just over a meter tall, weighs 33 kg and is equipped with 53 movement joints, most concentrated in arms and hands to allow him to take and manipulate objects; he can see and hear, he is the only robot in the world covered with artificial leather, which allows him to understand if and how he is touched and react accordingly, and he is able to walk alone. Which for a car is not exactly an easy task. iCub is a robot, but it is also one research platform on humanoid robotics, which is open source and whose updates are shared with 40 laboratories in Europe, the United States, Japan and South Korea, who use it for artificial intelligence studies.
iCub is the best known of the IIT robots, but it is not the only one: it exists HyQReal, an aluminum and Kevlar quadruped capable of walking, running, jumping, climbing / descending stairs, carrying weights, designed to intervene in emergency situations; there is Centaur, which as the name implies has 4 legs and a humanoid torso and can also perform tasks where the use of force is required, such as breaking a wooden plank or transporting material, and he too will serve as a support to the rescue teams in case situations of danger to humans or in hostile environments. And then there is R1, a prototype of a humanoid robot designed to work in hospitals, shops and even in our homes and therefore characterized by an appearance that is as similar to ours: head, torso, two arms and wheels instead of legs to allow it to stand up .
Italy after Morandi
Inspection and Wash, how the robots that monitor the safety of Ponte San Giorgio work
by Emanuele Capone
The problem of standing
Which, as we said, for robots is not at all easy: “It is rare that a biped robot does not fall and cause disasters,” admitted Metta. Explaining that it happens mainly for two reasons: “For a matter of hardware, because the human body is very well made, it is complicated to simulate and it absorbs the environment well through the muscles and the skeleton ”. Above all, as a matter of software: “We don’t fall because somehow we know what happens by putting one foot in front of the other, what awaits us. A machine cannot do this, because there are not yet such advanced AIs that allow it ”. Not yet, at least.
However, there is no joke about this, you have to understand them: “We also fall, if we can’t anticipate what will happen, if we suddenly meet a step or if we run a ladder”. So, is the idea of the robot announced by Musk unworkable? “I honestly think it’s hard to really see it in 2022: to make their bipedal robots, those of Boston Dynamics took 20 years of trying and I’m still behind. And they are the most advanced of all ”. However, for Metta the arrival date is not relevant: “It is not important when it will be done, it is important that the journey has begun and that someone like Musk has made a commitment like that. When he said he wanted to go to space, nobody believed him and even if he didn’t make it in the times announced at the beginning, in the end he did. And this is how we progress and evolve ”.
Our advice
Ten artificial intelligence movies (and a series) to watch in streaming
by Emanuele Capone
The future and positive robotics (also in movies)
Speaking of this, how will machines evolve, especially from the point of view of the intellect? “Already today, the AI do amazing things and they have reached levels of understanding and problem solving that in some cases surprise us researchers as well ”. And who knows that one day they will not come to understand our emotions, as seen in the cinema and as Alessandra Sciutti anticipated us a couple of years ago, which within Iit deals precisely with human-robot interaction, according to which “in the future, by including contextual aspects in the analysis, we can hope to have machines that are able to understand us more deeply: to have a robot capable of to understand if something he has done has bothered us, guarantees a more fruitful and effective interaction ”. According to Metta, the question is not so much whether the machines will be able to do this, but for what purpose: “We have to understand how all this can help us humans. In case of intervention on the scene of an accident, for example, it would be useful to have a robot capable of understanding whether an accident has occurred or that there is a person who is ill, even if only from their facial expressions “.
The problem, however, is not in the computing power nor in the processor that allows the machine to act and the AI to think: “It is only in the training phase of the algorithms that are the basis of artificial intelligences, which need super computers, very powerful, very large, very expensive – explained Metta – To use those same algorithms inside a robot, a normal computer, even a small and portable one, is enough ”. The secret is in the cloud, because “the robot collects data on the situation it has to face, which is processed in the cloud and then returned to it with the answer that allows it to act correctly”.
Hoping that everything goes well, a bit like it happens in Bicentennial Man, which is one of the films on the subject preferred by the scientific director of Iit: “Partly because it’s taken from a short story by Asimov, and you can’t do this job if you don’t love Asimov; partly because we talk about robotics in a positive, good way, without dystopias and dramas ”. This is unlikely to happen in another highly anticipated work both by Metta and by more or less all science fiction fans: “I’m curious to see how it will be Foundation, the series based on the novels of Foundations cycle (also by Asimov, ed). So curious that I’ll have to subscribe to Apple TV Plus just for this “. And who knows what iCub would say about this human weakness …
.