/* ---- Google Analytics Code Below */

Sunday, March 24, 2019

Motor Babbling Generates Data for Walking

Seeing some of these kinds of examples always makes me think of early AI work.  If we could design very good learners (and re-learners) we could start them like a human baby,  without knowledge,  put them in a context and just let them learn.    But there is still today much work to do to arrange the data, check it for correctness,  figure out what features are important, test it for bias, ethics  and consistency.    And much more.

A Robotic Leg, Born Without Prior Knowledge, Learns to Walk 
USC Viterbi School of Engineering
By Greta Harrison; Amy Blumenthal

Researchers at the University of Southern California (USC) have developed a bio-inspired algorithm that enabled a robotic limb to learn a new walking task by itself after only five minutes of unstructured activity, and then adapt to other tasks without additional programming. This breakthrough is similar to the natural learning that happens in babies, as the robotic limb was first allowed to understand its environment in a process of free play, known as "motor babbling." The random movements of the leg that take place during motor babbling allow the robot to build an internal map of its limb and its interactions with the environment. The robots use their unique experience to develop the gait pattern that works well enough for them, producing robots with personalized movements. Said USC researcher Francisco J. Valero-Cuevas, "Because our robots can learn habits, they can learn your habits, and mimic your movement style for the tasks you need in everyday life—even as you learn a new task, or grow stronger or weaker.". ... '

No comments: