3.253 seconds: Robot solves Rubik's Cube in record time at Birmingham fair (w/ video)

Mar 17, 2014 by Nancy Owano weblog

(Phys.org) —A robot that works with Lego Mindstorms modules and a Samsung smartphone smashed the world speed record in solving a Rubik's Cube, it was announced Saturday.

Cubestormer 3, performing at The Big Bang Fair in Birmingham, UK, unscrambled the cube in 3.253 seconds. The Cubestormer creators are David Gilday, principal engineer at Cambridge-based ARM, and Mike Dobson, an engineer at Securi-Plex, which specializes in electronic security systems. Dobson worked on the design. Gilday worked on the algorithm and robot software. The two engineers took 18 months, in their spare time, to design and build the robot. (Their efforts were rewarded on Saturday with a show of speed that smashed the previous record of 5.27 seconds set by the same design team in 2011 with their Lego robot, Cubestormer 2.)

Guinness World Records editor-in-chief, Craig Glenday, was invited to adjudicate at the event. As Engadget's Saturday story title marveled, "Lego Cubestormer robot solves Rubik's Cube in less time than it takes to read this headline."

A Samsung Galaxy S4 smartphone and ARM processors are key elements. The S4 device is the "intelligence" component, which serves to analyze the cube. The phone is powered by an Exynos 5 Octa application processor with an eight-core ARM big.LITTLE implementation, featuring four Cortex-A15 and four Cortex-A7 processors (Exynos 5 Octa is an ARM based Octa-core mobile CPU).

This video is not supported by your browser at this time.

After analyzing the cube the phone tells four robotic hands to do the manipulations. ARM processors also power the Mindstorms EV3 bricks.

Gizmodo noted how the robot hands must be "amazingly precise" to move so smoothly and quickly. This is but one of the challenges that confronted the engineers in crafting their high-speed entry. Commented Gilday in a published statement: "As well as working out the solution, the ARM-powered Exynos processor has to instruct the robot to carry out the moves. This is more complex than it seems because Cubsestormer 3 uses a speed cube which allows twists before the sides are fully-aligned. It means the robot is effectively mirroring the same kind of judgment and dexterity that a human speed cube has to apply. "

In the end, he said, the efforts the two made to ensure motor and intelligence functions were properly synchronized paid off. "Our big challenge now is working out if it's possible to make it go even faster."

The speed-record event took place at The Big Bang, which is a technology and engineering fair held at the National Exhibition Center in Birmingham, to inspire young people as the future generation of scientists and engineers.

Explore further: Rubik's Cube solving robot at Scienceworks

More information: www.businesswire.com/news/home… Solving#.UyXSRvldUu7

add to favorites email to friend print save as pdf

Related Stories

Rubik's Cube solving robot at Scienceworks

Dec 04, 2013

The world's fastest Rubik's Cube-solving robot, developed by students at Swinburne University of Technology, is now permanently on display at Scienceworks in Melbourne.

Exynos 5 Dual chip is unrobed by Samsung

Aug 11, 2012

(Phys.org) -- Samsung has released details about its next-generation Exynos 5 Dual. This is a dual-core mobile CPU based on ARM Cortex-A15 architecture. That ARM Cortex A15 word string is no small differentiator ...

A robot to beat humans at foosball

Aug 26, 2013

At first glance, the foosball table located in the middle of the Automatic Control Laboratory looks perfectly normal. Looks can be deceiving. In defense, one of the levers has a mechanical arm capable of ...

Recommended for you

Will tomorrow's robots move like snakes?

15 hours ago

Over the last few years, researchers at MIT's Computer Science and Artificial Intelligence Lab (CSAIL) have developed biologically inspired robots designed to fly like falcons, perch like pigeons, and swim ...

Robot Boris learning to load a dishwasher (w/ Video)

Sep 12, 2014

Researchers at the University of Birmingham in the U.K. have set themselves an ambitious goal: programming a robot in such a way as to allow it to collect dishes, cutlery, etc. from a dinner table, and put ...

Deep-sea diver hand offers freedom and feedback

Sep 12, 2014

Bodyskins and goggles are hardly the solution for divers who need to reach extreme depths. The Atmospheric Dive Suit (ADS) gives them the protection they need. Recently, The Economist detailed a technology ...

User comments : 4

Adjust slider to filter visible comments by rank

Display comments: newest first

Noumenon
3.4 / 5 (5) Mar 17, 2014
I used to do this in about a minute and a half by hand using a memorized 'algorithm', and noticed that the time completed varies depending on the algorithm used and 'luck of the draw' in the scrabbled arrangement of the cube,... i.e. sometimes a 'level' would be nearly already complete. (and sometimes the cube disintegrated in my hand)

I think that is what happened here because I didn't see alot of rotations. What the contest should do is require ten trials and determine the official time based on an average.
DistortedSignature
4.3 / 5 (3) Mar 17, 2014
@Noumenon I understand what you're saying but at the same time isn't that a slippery slope? When world records come into play, whether it be something like this or an olympic event, there are always different variables that come into play. I mention olympic events because there are several factors that can affect events like the 100m sprint that'll affect the speed of the runner to give them a faster time. For example elevation and wind direction, to types of shoes, what they ate, and physique of the runner.

As long as the cube is (pseudo) randomized by the same standard every single time, then I don't see a problem with this. Speed solvers usually get it into several different finishing positions and not a level/algorithm approach, e.g. corners first, top-middle-bottom, top-bottom-middle, etc. With a fast enough computer it would only take 20 moves to solve every combination. Should the record then be to see how fast a robot can do 20 random movements?
TheGhostofOtto1923
1 / 5 (1) Mar 17, 2014
I used to do this in about a minute and a half by hand using a memorized 'algorithm', and noticed that the time completed varies depending on the algorithm used and 'luck of the draw' in the scrabbled arrangement of the cube,... i.e. sometimes a 'level' would be nearly already complete. (and sometimes the cube disintegrated in my hand)

I think that is what happened here because I didn't see alot of rotations. What the contest should do is require ten trials and determine the official time based on an average.
Sounds like there's a maximal scrambled state. Just have the robot set the cube in this state and then have it solve it. Don't worry - computers don't cheat.
Noumenon
3 / 5 (2) Mar 17, 2014
With a fast enough computer it would only take 20 moves to solve every combination.


I did not know this problem was solved.

Should the record then be to see how fast a robot can do 20 random movements?


Is that not what they are doing then given the above solution?