Machines’ Omniscience. What Robots Look Up on the Internet
Robots make life easier: they vacuum clean, work on factories and tell jokes. Scientists from different countries try to work with artificial intelligence so that the machines can perform more functions. One of these ways is to connect them to the Internet.
Top of the year, Las Vegas, about 10 degrees, it was nearly night. Not far from the empty parking lot by a building where consumer electronics exhibition took place, a robot was standing on the road. It was looking somewhere. On its interface, appeared something resembling a smile.
Together with other robots of the Promobot company, it had to get to the exhibition and take part in the stand design. But it changed its route and seemed to enjoy the air of freedom.
With injured head and hand mechanisms, it lay on the side of the road unresponsive. It was hit by an autopilot controlled Tesla.
Perhaps Promobot could be well if it were connected to RoboEarth, the ‘Robot Internet’, and analyzed the data on how safe it was to stand on the road when a car is racing towards you.
What ‘Robot Internet’ Is
In 1991, the first website was created. One can still access it.
28 years later, new Internet appears, and it’s no longer for people. It will be used by robots (however, it should be noted that about a hundred machines are already connected to the RoboEarth system).
RoboEarth has the same principle as the regular ‘human Internet’. The system allows robots learn from each other. Scientists from five European universities sponsored by the European Commission came up with the idea of creating the cloud space. They proceeded from the fact that most machines were very effective, but only under conditions that take into account all possible scenarios. If the task or conditions are changed, robots become useless.
If a company owns a certain number of robots, in conditions when their tasks change, it is necessary to reprogram all the machines. According to Heiko Sandy, who works at RoboEarth, the cost of such a procedure comes to 80 percent of the cost of new robots.
RoboEarth has to solve this problem, because it is enough to load new data into the system for robots connected to the system to use it.
The researchers working on RoboEarth have created a special language that stores all the information. These are not words familiar to us, but a set of commands, coordinates, parameters, and other calculations. For example, to navigate in a room, a robot loads the map of the room that was created by a robot who visited it earlier.
In addition, the robot will detect the door in the room. If so far it has not had information about such objects, it will start looking for it: what is this rectangular thing? How does it work? Why does one need it? How does this type of the door open? After that, the robot will look for possible options for action. Sandy describes this search as ‘a set of tasks that must be performed in a certain order, taking into account the evaluation of the parameters “if … then … ”’.
If the robot faces the task of opening the door and leaving the room, it will begin experimenting: what if I press the handle? There is no effect. What if I just pull the door? It opens!
This new experience will also be posted on RoboEarth, which means the next robot that will face such door will not have difficulty opening it.
The commercial potential of this technology is enormous: after all, according to scientists, if robots can learn on the go, then initially the manufacturer will not need to spend resources and time on complicated programming. In addition, the robots themselves will consume less resources required for complex calculations. It will be enough just to google, though it is the robots’ Google.
Sandy explained the principles of the RoboEarth technology: ‘We do not draw a map that can be used by a robot to navigate, instead, we teach the robot to create its own map’.
Aren’t Robots Fiction?
It depends on what one means when he or she mentions robots. If you talk about androids, as in the game Detroit: Become Human or the Blade Runner movie, the answer is no. Although there are already real technical organisms like robot Sofia, who gained citizenship of the United Arab Emirates. But there are already many machines that effectively solve tasks. And the further, the more.
Here are some examples of how robots have already simplified human life.
They have mechanical ‘hands’ with tools and sensors that make them ideal for working on a conveyor belt. Robots not only save manufacturer’s money, but also perform complex tasks at a pace that a human can’t have. They also make production safer, because they can perform difficult and dangerous work instead of people.
Military and police robots
This are the most high-tech robots that are used in our time. They perform extremely dangerous tasks without endangering the lives of the people who operate them. These are sappers, scouts, and drones. Police robots help fight crime without risking the lives of police officers. The police use different machines with remote control equipped with cameras, including infrared, as well as speakers and other technologies to search for criminals. They are able to neutralize an offender, as was the case with the Dallas sniper.
These robots can be unmanned spacecraft, like lunar rovers and Martian rovers, or they can be parts of a ship’s equipment — for example, mechanical ‘hands’ that help carry out complex and dangerous tasks.
Robotic surgery has expanded the capabilities of surgeons in ways inaccessible to people. The robots are controlled by surgeons who use a computer console to move the tools attached to the ‘hands’ of a robot. Medical technology is so developed that with the necessary equipment a surgeon can do everything remotely, that is, conduct a surgery from a different room and even a different country.
It’s still worth thinking about the future place of work, you don’t need to worry too much. Eriko Guizzo, senior editor at Spectrum, a media of the Institute of Electrical and Electronics Engineers, claimed that there is still a lot to do for robots to fully cope with most human tasks: ‘Robots are generally still too limited in what they can do. They may be great for carrying out repetitive tasks in clutter-free environments, but entering a rubble-strewn building, climbing ladders, using fire hoses—these operations are beyond today’s best robots.’
But What About Tomorrow?
Until recently, robots were not able to understand and cope with unstructured circumstances: their systems relied on knowledge of every possible situation that they might encounter. Each contingency response had to be programmed in advance, and the systems had to rebuild their model of the world based on sensor data each time they had to perform a new task.
This is one of the main reasons why robots are mainly shipped to strictly controlled and predictable environments, such as manufacturing facilities. The human world is too complex to be generalized within a limited set of specifications.
Soon the robots will be able to learn from their past experiences. They will be able to instantly share their knowledge with their own kind. RoboEarth can be used as a stepping stone for further research and development.
However, in trying to simplify your life, remember the words of James Barrat, author of ‘Our Final Invention: Artificial Intelligence and the End of the Human Era’. He wrote about the dangers that we may face because of robots: ‘RoboEarth increases safety, because we can essentially manage them from one point. But we need to ensure that the data nodes do not develop or otherwise improve their own software. It is necessary to study the consequences of robots sharing a common mind.’