24/7 Customer Service (800) 927-7671

Autonomous learning puts human-like dexterity within robotic reach

by • May 9, 2016 • No Comments

On the other hand humans perform intricate hand movements like rolling, pivoting, bending and grabbing various shaped objects without a 2nd idea, such dexterity is yet beyond the grasp of many robots. But a team of desktop scientists at the University of Washington has upped the dexterity stat of a five-fingered robotic hand that can ape human movements and learn to improve on its own.

Built at a cost of around US$300,000, the University of Washington’s create is based on a Shadow Hand skeleton, that is actuated with a custom pneumatic process that gives it the competence to move faster than a human hand. On the other hand the hand is already too expensive for commercial or industrial use, it does allow researchers to test control strategies that mayn’t otherwise be possible.

“Usually folks appear at a motion and try to determine what precisely needs to take place — the pinky needs to move that way, so we’ll put a few rules in and try it and if a fewthing does not work, oh the middle finger moved too much and the pen tilted, so we’ll try another rule,” says senior author and lab director Emo Todorov. “It’s almany like manufacturing an animated movie — it appears real but there was an army of animators tweaking it. What we are via is a universal approach that empowers the robot to learn of its own movements and requires no tweaking of us.”

Having utilized an algorithm that was able-bodied to version complicated five-fingered behaviors, like typing on a keyboard or catching a falling object, and simulate the movements needed to complete a desired result in real time.

This desktop version has now been applied to the robotic hand hardware to see how it performs in the real world. Additionally, the robot hand has been given the competence to learn and improve its performance without human input thanks to a range of sensors and motion capture cameras that feed data to machine learning algorithms. These assist version the actions required to complete a goal and the basic physics involved. This is in contrast to other approaches that involve guide programming of every individual movement of a robotic hand.

Today, the autonomous learning process is able-bodied to improve at a specific task that involves manipulating one set object in a much like way to what it has done previously, but the team is aiming to demonstrate global learning, that may allow the hand to work out how to effectively manipulate unfamiliar objects via its 40 tendons and 21 joints.

The research was funded by the National Science Foundation and the National Institutes of Health.

A video of the hand in action is at a lower place. To see the robot learn how to twist an object, skip to 1:47.

Source: University of Washington

Latest posts

by admin • March 5, 2017