Catch up on stories from the past week (and beyond) at the Slashdot story archive

 



Forgot your password?
typodupeerror
×
Robotics Science Technology

Evolutionary Robotics 30

schwit1 sends news that researchers from the University of Cambridge have developed robots that emulate natural selection. The "mother" robot designs and builds its "child" bots, and then tests them against particular criteria. The child bots that are the most successful have their traits carried on to the next generation. "For each robot child, there is a unique ‘genome’ made up of a combination of between one and five different genes, which contains all of the information about the child’s shape, construction and motor commands. As in nature, evolution in robots takes place through ‘mutation’, where components of one gene are modified or single genes are added or deleted, and ‘crossover’, where a new genome is formed by merging genes from two individuals." By the final generation, the fastest robots were able to perform their task twice as fast as the average robot in the first generation.
This discussion has been archived. No new comments can be posted.

Evolutionary Robotics

Comments Filter:
  • by Anonymous Coward

    Give it a few million years and maybe we will see something!

  • by Anonymous Coward

    Programmed for a purpose, the mother robot...

  • I for one welcome our new terminator overlords.
    • by PPH ( 736903 )

      Skynet becomes self aware 2:14 a.m. Eastern time, August 29th.

      Mark your calendars.

  • by jeffb (2.718) ( 1189693 ) on Saturday August 15, 2015 @11:06AM (#50322161)

    This doesn't seem especially interesting if the "child" robots can't in turn produce and select their own children.

    At that point, I guess it gets even less interesting, because the most "successful" mutation will be a simple one that causes the mother to select all her offspring.

    • the most "successful" mutation will be a simple one that causes the mother to select all her offspring.

      No, the most successful mutation will be the one that causes a robot to kill its competitors.

  • by Anonymous Coward

    Except for the coolness factor how is this different from a computer simulation? The robots can't replicate without humans involved. Why would you waste your time and build robots when more in depth studies can be done in software ( you don't have to mess around with hardware, if you need a new parameter, you don't have to build pieces for your robot)

  • by Jim Sadler ( 3430529 ) on Saturday August 15, 2015 @02:33PM (#50323027)
    Ok so we have had computer programs designed to evolve against a standard for well over a decade. Snippets of code were simply transferred between mating pairs and the best were allowed to survive while the worst were taken out of play. Obviously the same thing could be done with robots and in fact one could use programs to evolve each and every part of a robot and then create a robot with the newly designed superior parts and testing it out. So I don't really see anything new in this at all. Matter of fact we have entertainment with "battle" bots in which survival in the arena gives feedback to builders who then make changes to make a better warrior robot. So far these devices are like drones with human operators but we should see battle bots who have only automated control built in with no humans steering the devices. Perhaps automated battle bots would be a quite measurable method of advancing robotic abilities.
  • This is not natural selection at all.

  • Sims presented similar work at SigGraph in 1994 [karlsims.com]. He did physical simulations using a Connection Machine CM-5.

    This video shows results from a research project involving simulated Darwinian evolutions of virtual block creatures. A population of several hundred creatures is created within a supercomputer, and each creature is tested for their ability to perform a given task, such the ability to swim in a simulated water environment. Those that are most successful survive, and their virtual genes containing co

  • is reproduce itself. That is the one and ONLY reason to do this sort of research physically, and not in simulation.
  • I didn't think Framsticks was all that new

Work is the crab grass in the lawn of life. -- Schulz

Working...