Researchers develop self-learning robotic hands using Deep Learning

June 18, 2017, 10:36 a.m. By: Pranjal Kumar

self-learning robotic

Bielefeld University researchers have developed a system with the robotic hands that autonomously familiarizes itself with novel objects. This system works without previously knowing the characteristic of the objects. This system is developed as the part of a large-scale project called Famula at Bielefeld University.

The scientists have been working on developing robots for a very long time. But the one problem that they were having is that robot has no idea about the things that he is touching or picking up. It has to be informed about the characteristic of the object earlier. But soon this problem can be solved using the latest developed technology. This can enable the scientist to develop the robot that can independently adapt to working in a new household.

According to the researchers,” The system is like a small baby. It learns by trying out and exploring on its own.” The researchers are Woking on the robot that has two hands which are quite similar to the human hands in terms of mobility and shape. The robots learn to distinguish between the various objects on the basis of their color, shape, size as well as what matters when attempting to grasp the object.

According to the researchers,” Humans identify the object on the basis of shape, size, and color rather than weight. This robot also does the same. These robotic hands can interpret both the spoken language and the gestures. This can help in giving the input to the robot about its task.

This can be a great development in the field of robotics as it can help the robots to understand their surrounding better. They can be trained to the reaction the basis of their surrounding environments.

Image Source: ScienceDaily