Humans Help to Complete Robot Learning

Featured Life Style

In the Gates Computer Science Building at Stanford University, a screen attached to a red robotic arm which lights up and its eyes blink. Ajay Mandlekar, a Ph.D. student in electrical engineering introduced the robot ‘Bender.’

These researchers are using this robot arm for testing two frameworks together which can make the learning basic skills faster and easier. One of this framework, RoboTurk allows people to direct the robot arm in real time and control by their smartphones or browser by showing their way to complete their tasks. Another framework, SURREAL speed up the learning process by running multiple experiences at the same time which makes the robot to learn many experiences at the same time.

With both the framework, the researchers were able to push the boundary of the ability of robots by combining numerous data and collecting together from the human and large support of leaning, informed Ajay Mandlekar. The group of researchers will present this arm with this framework on Oct 29.

These two frameworks – RoboTurk and SURREAL – are the features to make easier to learn basic skills. However, the basic lessons are not that perfect; thus, humans are not the optimal mean to complete the learning process of the robot but this experience is an integral part for robots, said Mandlekar.

Additionally, the SURREAL is another way to make this process in a faster way.

The team expects this robot will be an integral part of life in future as assistance in household chores and performing assembly tasks in manufacturing or completing the dangerous tasks without risking the life of humans.

Leave a Reply