US researchers have developed a new way of learning robots
Education or training robots - no easy task. As a rule, scientists and engineers have to resort to complex algorithms to their mechanical movement wards meet certain requirements. A team of researchers from the American Rice University has developed a completely new method for teaching robot motion. In order that the robot was able to learn how to move, a person only needs to push his or her hands forward in the right direction.
Yes, the new method is also based on machine learning algorithms, but in this case comes in not only mathematics, but also the operator, which is like a parent who teaches his child's first steps, gently guides the robot and tells him how perform a task properly and without errors. Depending on how one directs the robot changes its trajectory. Real-time person brings new knowledge into machine code, and the robot is learning something new. Generally, if the robot is programmed to interact with a person, he will notice physical contact, but then try to return to the originally selected their path of movement. In this case, the robot is able to use this contact for their own learning. He will take into account a change in slope, path, direction, and the next time will act in accordance with them new knowledge obtained. Nowhere is this shown in the video released by researchers, where they demonstrate their system operation and explain how it works.