One of the biggest problems with the development and use of the current era’s robot is finding the best methods to teach them. Basically, a robot is programmed on a certain set of codes which instructs him on how to manipulate with an object but as you give them a completely strange robot, they will go dumb. This issue occurs because most of the modern day robots aren’t very good at learning and using skills for understanding completely strange objects.
But some of the researchers at the MIT have found a way out of this issue. They have developed a one- armed robot which can study a handful of different shoes by moving back and forth to visually study the object. So, in case you give this robot a completely different shoe and command it to pick up then the machine will identify the tongue of the shoe and then pick it up on its own, without the need of any human interference.
This new innovation by the researchers of the MIT is big news as most of the other robots are still struggling to get familiar with the human world.
The training process of a robot requires a lot of hand training which basically involves you to joystick around. This process of learning is known as the imitating learning in the tech world. Another way is to just leave the robots on its own so that he can learn by trying again and again. The random movements of the robot are noticed in this case and each time the robot reaches near to the goal, it is rewarded with points.
But this new technology involves no hands as the researchers have to just leave the shoe and robot alone. The roboticist at the MIT computer science, Pete Florence said that “It can build up—entirely by itself, with no human help—a very detailed visual model of these objects,
Working with objects for which a robot hasn’t been programmed is one of the greatest achievements in the world of robotics as it will give a new way of making these robots learn and interact with the complicated and wide human world.
When you place a shoe in front of the robot, then it will take a look at the shoe from different angles through its in-built camera. During this process, the robot collects the data which it needs in order to build rich internal description of the meaning of particular pixels.
In order to use a robot more conveniently, you don’t only want it to understand different objects but you also want it to understand what these different objects are made of. For example, if you want your robot to pick up a chair with a loose handle, then you would have to instruct it on picking up it up from its legs. But in this case, you will have to instruct him on what is a leg of table and each time, you will have to instruct.