Saturday, 15 December 2018
 
  Home arrow Artificial Intelligence arrow Human-Like Robot Runs  
Main Menu
Home
About us
Email
Blog
Forum
Search
How To Order
Contact Us
Private Policy
Terms of use
Articles/Tutorials
Artificial Intelligence
Robotics
Products Categories
Robotics Solar Robotics

List All Products


Advanced Search
Product Scroller
JRT RP1062
JRT RP1062


JRT RP1064
JRT RP1064


Sumovore
Sumovore


JRT RP1082
JRT RP1082


Hyper Peppy
Hyper Peppy


JRT Shop
  New 
New
Top 
Top Ten
Special 
Special
Random 
Random
 
Robotic Arm
Robotic Arm

Avoider III
Avoider III


Robotic Arm
Robotic Arm

The Sumovore Baseplate Upgrade Kit
The Sumovore Baseplate Upgrade Kit


JRT RP1064
JRT RP1064


Download J Robotics's listed product's catalog. To download please click here. | For free E-Mail id like myname@jrt.in, register with us.  
Human-Like Robot Runs Print E-mail

A new two-legged robot that can walk with a human gait recently demonstrated it has the ability to run.

While other robots, such as Honda's Asimo, can also run, the French-based Rabbit relies on a completely different control system that could save researchers months of experimental time and ultimately produce more agile robots.

Most human-like robots, such as Asimo, are programmed to learn how to walk using trial-and-error methods.

Over months and even years, researchers painstakingly program a robot to move inch by inch and monitor what happens during every little action, such as lifting a leg, swinging it forward, or shifting the torso weight.

When a movement succeeds, it's saved in the programming. But more often than not, it fails — the leg steps too far forward, for example, causing the robot to stumble — and the scientists must go back to the drawing board to refine the robot's gait.

"They end up walking like an old man for want of a better word," said University of Illinois at Urbana-Champaign engineer Mark Spong, whose research focuses on robotics.

It can take months of lab time to make the robot walk well. Researchers at the Technical University of Munich spent approximately 18 months taking their robot, Johnnie, from its first steps to walking well.

But the 4 1/2-foot tall Rabbit, designed and developed within France's national Robotics and Artificial Entities project (ROBEA), does not rely on the popular trial-and-error methods.

Instead, it uses a control system based in math developed at the University of Michigan by Jessy Grizzle and Eric Westervelt, now at Ohio State University.

The team figured out a way to describe and predict the robot's motion using mathematical equations that on some level are similar to those that describe the motion of a pendulum on an old-fashioned clock.

To keep time, a pendulum must balance the energy it gains while swinging to the energy it loses at the end of each stroke. That is similar to what happens when each of Rabbit's legs swing forward and then lose energy as the stilt-like foot impacts the ground.

And if someone pushes on a pendulum, it does not speed up, but readjusts its momentum to absorb the energy, very soon re-establishing its regular motion. Rabbit does the same thing when pushed from behind.

Because the mathematical equations accurately described and predicted Rabbit's motion, the robot walked the very first time Grizzy set it in motion.

"It worked right out of the box," said Grizzle.

And, he said, "Within two weeks, it was completely routine to achieve a stable, robust, visually pleasing gait."

Eighteen months later, Rabbit ran six steps. The results of that work will soon appear in the International Journal of Robotics Research.

Rabbit still has a way to go. Currently it can only move in two dimensions: back and forth and up and down.

A stability bar holds it from falling sideways, and it cannot run more than six steps.

But Christine Chevallereau, Carlos Canudas de Wit, and Gabriel Buche of ROBEA are hard at work on unravelling the last puzzles with Rabbit.

Grizzle is in the process of building a robot similar to Rabbit for his own lab to give him more hands-on time to work out such kinks.

"I am confident that other researchers will soon adopt some of our results, modify others, and come up with even more exciting breakthroughs," says Grizzle.

 
< Prev   Next >
JRT Videos

You need to a flashplayer enabled browser to view this YouTube video  

more videos....

Newsletter/Register

Advertisement

© 2018 J Robotics Technologies, India

Get The Best Free Joomla Templates at www.joomla-templates.com