An Exynos-powered robot breaks the Rubik’s Cube world record
Who remembers wishing they could break a world record when they were younger? If your inner ten-year-old is still fascinated by world record attempts, then we have some news that you’ll love – especially as it also involves Lego!
Did you know that the Rubik’s Cube was invented in 1974? Since then, it’s been a much-loved children’s toy and the subject of record attempts from around the world.
Last weekend, an Exynos-powered Lego robot beat the world record for solving a Rubik’s Cube, in a super-speedy 3.253 seconds at the Big Bang Fair 2014 in Birmingham, UK. The record-breakers David Gilday, a principal engineer at ARM, and Mike Dobson, a security systems engineer for Securi-Plex, built the CUBESTORMER III to solve the classic Rubik’s Cube puzzle, which was powered by a Samsung Galaxy S4 complete with Exynos 5 Octa processor.
The same team were responsible for the previous record of 5.27 seconds using their CUBESTORMER II robot the year before – which means they managed to shave a respectable 2 seconds off their previous record. The machine performed the movements needed to beat the record faster than the human eye can see, demonstrating the processing power of the Exynos 5 Octa.
It’s even more impressive when you consider that the machine needed to analyse the cube for the solution, calculate the correct sequence of moves and move the robotic arms to complete the sequence. The fastest human to be able to do this is Mats Valk, who managed to complete a Rubik’s Cube in 5.55 seconds in 2013.
We spoke to David Gilday, one half of the team that beat the world record, to ask him how he and his engineering partner went about it.
The overall process of solving the cube consists of three stages. “First the scrambled cube is scanned to determine the colors of the pieces”, says David. “Then, an efficient solution is calculated which describes how the faces of the cube should be turned to solve the puzzle. Finally this solution is transformed into a series of rotations that are applied via the motors to cause the mechanism to physically manipulate the cube.”
David explains that the Exynos 5 Octa-powered Samsung Galaxy S4 smartphone is central to the overall process. “Its camera is used to scan the cube, and the software running on the Samsung Exynos 5 Octa application processor with an eight-core ARM big.LITTLE™ implementation (featuring four ARM Cortex®-A15 and four Cortex-A7 processors) analyzes the cube. It then calculates the solution, and instructs the eight ARM-powered LEGO® MINDSTORMS® EV3 bricks to sequence the motors during the scanning and solve processes.”
According to David, the overall speed of the design was the main consideration when constructing the robot. “In addition to raw speed, the mechanism is designed to allow many physical operations to be overlapped in time. The camera is carefully synchronized with motor movements to minimize the time that the cube has to be held stationary for each captured image. The software algorithm is optimized specifically to take advantage of the multi-processing capability and large memory of the Samsung Galaxy S4. Also, the software optimizes the solutions specifically for the mechanical capabilities of the four independent grips.”
David explains however that they also spent a large amount of time making the robot as reliable as possible, spending many hours testing and adjusting the timings to make sure they were as prepared as possible. “It was important to run CUBESTORMER 3 on location before the attempt, to ensure that differences in the environment such as lighting conditions would not reduce its reliability. Mike and I were fairly confident that CUBESTORMER 3 would beat the previous record by quite a margin – we were hoping for a time well under 4 seconds, and were thrilled when it achieved an official time of just 3.253 seconds. The high performance of the Samsung Exynos 5 Octa application processor in the Galaxy S4 allows the software to calculate a number of efficient solutions to the puzzle in only around 50 milliseconds. Then using a software model of the timing characteristics of the mechanism, the fastest of these solutions is chosen.”
Owing to the complexity of the Rubik’s Cube puzzle, efficient solutions typically require an algorithm that searches for a solution rather than being able to directly calculate one single, short solution.
“The algorithm I developed is based on a two-phase approach using pre-calculated lookup tables,” says David. “The efficiency of the solution is dependent on the amount of processing time allowed for the search and on the contents of the lookup tables. Simply, the more processing power is available, the faster an efficient solution can be found. The Samsung GS4 was chosen because of the high performance of the Samsung Exynos 5 Octa application processor. The large RAM size also enabled the use of larger lookup tables with more entries optimized for CUBESTORMER 3’s physical capabilities to help find better solutions.”
“Mike and I were both initially inspired a few years ago by videos of other robots solving the Rubik’s Cube puzzle to create our own individual designs out of LEGO bricks. Our first joint project, CubeStormer II was significantly faster than the previous robot world record so it seemed natural to attempt the official record for that design. It was only during the preparation for this event that we decided to attempt a Guinness World Record, hoping that this might raise awareness of the Big Bang Fair* and maybe inspire even more youngsters to be interested.”
The Big Bang Fair exhibition is held every year at the NEC centre in Birmingham UK, focusing on the latest innovations in science.
As well as the amazing achievements of the CUBESTORMER III, another Exynos-powered Lego robot was a world record breaker last weekend, as the ARM team’s Multicuber 999 powered through the record for solving a largest 9 x 9 x 9 Rubik’s Cube. The feat had not been previously attempted by a robot. This mammoth task, involving a highly complex set of movements was completed in a record-breaking 34 minutes and 25.89 seconds, and was powered by an Exynos 4 Quad housed in a Samsung Galaxy S3.
So what does this mean for the future of the Rubik’s Cube record? We’d like to think that David and his team will come back next year and improve on this amazing innovation by completing it in even quicker time.
Want more news about events featuring Exynos products? Follow us on Facebook and Twitter for all the latest Exynos news.
*This article was originally published on Samsung Exynos Blog.
Products > Semiconductors
For any issues related to customer service, please go to
Customer Support page for assistance.
For media inquiries, please click Media Contact to move to the form.
