I.T. Vibe
Latest Business Communications Gaming General Security Technology Virus  
   Member Services
Login
Register
   General Services
Contact Us
Merchandise
Toolbar
RSS Feeds
Other Formats
   Site Search
 
Advanced Search
   News Alerts
Enter your email address to receive news alerts
 
View Privacy Policy
Unsubscribe
   Information
Latest Virus Alerts
Internet Threat Level
Internet Traffic Report
   Opinion Poll
Macs - Love Them or Hate Them? Place your votes now.
Love 'em
Hate 'em
Indifferent
Reader Comments: 0
View All Polls
Meet The Robots That Learn From Their Own Mistakes!
Thursday, August 07, 2008 at 10:46 by Peter Smith
Researchers from Leipzig have come forward with a breathtaking software package which can be used to teach robots how to move. The program mimics the human brain in what is been hailed as a ‘neural network' allowing computer simulated creatures to learn how to use their limbs, much like a new born baby or animal would. So what does this mean in real life?

The uses for this revolutionary software could be far reaching if the first tests are anything to go by. The researchers issued a video which shows a variety of simulated animals learning how to walk and jump, with one even showing a simulation dog which learns to jump over a wall. The video is amazing to watch with the dog slowly learning to lift its paws higher and jump further up the wall until finally it is able to clear the obstacle.

There is also an interesting video which shows a simulation human, with the same number of joints and limbs, trying to walk. Much like a baby it learns how to steady itself by moving its feet and body weight until it is able to stand up straight and walk. The potential for this software is frightening as it could be the first step towards robots which do not need human input and can ‘think' for themselves – something which could be very dangerous in the wrong hands!
 
No reader comments posted Reader Comments: 0 Contact Peter Smith, the author of this article View a printer friendly version of this article Email this article to a friend RSS Feeds

Your Verification Number:


Please enter your Verification Number: