Follow Slashdot blog updates by subscribing to our blog RSS feed

 


Forgot your password?
Close
typodupeerror
Robotics Math

Student's Robot Obliterates 4x4 Rubik's Cube World Record (bbc.com) 24

Posted by BeauHD from the you-get-an-A dept.
An anonymous reader quotes a report from the BBC: A student's robot has beaten the world record for solving a four-by-four Rubik's cube -- by 33 seconds. Matthew Pidden, a 22-year-old University of Bristol student, built and trained the "Revenger" over 15 weeks for his computer science bachelor's degree. The robot solved the cube in 45.305 seconds, obliterating the world record of 1 minute 18 seconds. However, the human record for solving the cube is 15.71 seconds.

Mr Pidden's robot uses dual webcams to scan the cube, a custom mechanism to manipulate the faces, and a fully self-built solving algorithm to generate efficient solutions. The student now plans to study for a master's degree in robotics at Imperial College London.

Student's Robot Obliterates 4x4 Rubik's Cube World Record More | Reply

Student's Robot Obliterates 4x4 Rubik's Cube World Record

Comments Filter:

  • 4x4? (Score:2)

    by ddtmm ( 549094 )
    Or 4x4x4
  • This was 16+... and is not the current record on video. Humans have beaten this number on video since 2019.

  • Why is the robot much slower than the fastest human? Is the problem that each robot turn is slower or that the algorithm for determining the sequence of turns comes up with longer sequences? I'm guessing that the problem is the relative slowness of each robot turn. If the robot turns could be done really fast, say at 1/10th the speed of the fastest human, then the turn sequence length becomes irrelevant.

    However, I imagine that if a high priority and unlimited funds were devoted to building a fast-turn ro

    • Re: (Score:2)

      by necro81 ( 917438 )
      Partly it is the speed of the robot. By comparing the BBC video against human-cubers on youtube, you can see this robot is relatively sedate. (Still fast in absolute terms, but slow compared to human moves.) Moreover, this robot is, shall we say, single-threaded. It can only execute a single move at a time. Human cubers will often use all their fingers to execute more than one move at once.

      If you built a robot that could do the (single-threaded) moves 10x as fast as a human, you'd actually have to s

    • Interaction with the cube is key here. The reason why a robot can obliterate a human in a standard 3x3 cube is you can fix the rotational axis on 6 fixed points on a 3x3 cube, then you're limited only by how quickly you can turn the 6 axis independently. Even ordered cubes like a 4x4 don't allow you to do this, there's no central stationary point meaning you don't just turn the cube face, you need to calculate how to hold the cube put the correct clamps in position on the cube, and remove the others in the

  • is to peel off the coloured squares and put them back - one colour per face. This, unfortunately, takes more than 15.71 seconds.

    • It isn't that hard to solve a 3x3x3 Rubik's cube. It just takes some practice and some algorithms.

      Technically you only need to learn a single algorithm [rider.biz]: R U R' U'.

      But it is best to learn at least T-Perm and Ja-Perm.

      Alternatively, there is also the Old Pochman Method [speedcubereview.com] for solving a cube blindfolded:

      For corners
      * Altered Y Permutation: (R U' R' U') R U R' F' (R U R' U') R' F R

      For Edges
      * J Permutation: (R U R' F') (R U R' U') R' F R2 (U' R' U')
      * T Permutation: (R U R' U') R' F R2 (U' R' U' R) U R' F'

    • Re: (Score:2)

      by Ogive17 ( 691899 )
      When I was a kid - I'd take the rubik's cube apart and reassemble in the correct order.

  • It should have done it with a humanoid hand, that would advance the field more. Right now robot hands are terrible and can't even do very basic dextrous tasks.

    • Re: (Score:2)

      by pz ( 113803 )

      It should have done it with a humanoid hand, that would advance the field more. Right now robot hands are terrible and can't even do very basic dextrous tasks.

      Did you watch the video? This is an application were a non-humanoid hand has clear superiority. His manipulator design is spot-on.

      The limiting factor was that his motors weren't fast enough: the linear positioning of the hands was much slower than I can imagine is possible, and the rotation speed, while better, was still slower than I'd expect. With bigger stepper motors, I would easily expect his design to go twice as fast and be in the realm of human speed. Faster than that might take substantial addit

      • Re: (Score:2)

        by pz ( 113803 )

        It also looks like he didn't shorten the wires to the stepper motors at all, meaning he may have found he could not reliably drive them faster than he did, a problem which could have been addressed by cutting them to proper length and ensuring the driver circuitry was up to the task of dumping massive currents to move the rotors quickly and absorbing massive flyback spikes when breaking the circuit. It also wasn't clear that he was using any advanced techniques for stepper motors, like ramping up and down

  • Takes me more than 45 seconds LOL
  • Is the start position always the same? Or some are easier to solve than others?

    • The starting setup should be randomized sufficiently. Otherwise, you could simply skip the analysis part in your algorithm - you could simply precalculate the optimum path solution to a known problem. Having to figure out the correct (or the best) solution is an important part of the challenge.

      Of course, it is important that the randomization is done by rotation. If you remove the pieces and reassemble the cube in a random configuration, the chances are you are out of luck. Only a very small subset of poss

Slashdot Top Deals

New systems generate new problems.

Close