In the quiet before the whir of motion, a cube sits scrambled. Then, in the time it takes to blink—actually, even less—it’s done. Solved. Six uniform faces, each a solid hue. It took just 0.103 seconds. A mechanical blur and a new world record.
That is the story of Purdubik’s Cube, a robot built not in a corporate lab but in a university workspace, by four students from Purdue University’s Elmore Family School of Electrical and Computer Engineering. Their achievement has earned a Guinness World Record for the fastest robot to solve a Rubik’s Cube—nearly three times faster than the previous title-holder from Mitsubishi Electric in Japan.
“We solve in 103 milliseconds,” said team member Matthew Patrohay. “A human blink takes about 200 to 300 milliseconds. So, before you even realize it’s moving, we’ve solved it.”
Shaving Off Milliseconds
Purdubik’s Cube didn’t win by simply being faster. It won by being smarter.
Where past record-holders focused on hardware—faster motors, stronger grips—Purdue’s team tackled an overlooked bottleneck: vision. A robot must see the cube first before turning it. Unlike human speedcubers who can study the puzzle before starting the clock, robots begin timing the moment the puzzle comes into view.
Instead of using high-resolution images, the students turned to low-resolution, high-speed sensors—Flir machine vision cameras with exposures as short as ten microseconds. Each camera captured just enough of the cube’s surface to be effective while keeping processing demands tiny. The image size? A cropped 128 by 124 pixels. No image enhancement, no full-frame scans.
They even skipped image processing altogether. Instead, they read raw RGB values straight from the sensors, feeding the data into a custom-built color detection system. Even artificial intelligence wasn’t fast enough.
“It’s sometimes slightly less reliable,” Patrohay admitted. “But even if it’s 90 percent consistent, that’s good enough as long as it’s fast. We really want that speed.”
Solving the Unsolvable, Faster
With colors identified, the robot had to know what to do. Here, the team turned to an existing cube-solving algorithm built for machines, Rob-Twophase, created by Elias Frantar. Unlike human solvers, robots can do things like spin two perpendicular sides at once—an edge humans can’t match.
But speed comes at a cost.
<!– Tag ID: zmescience_300x250_InContent_3
–>
One risky technique the team employed is called corner cutting—beginning the rotation of one side of the cube before another has completely stopped. Done well, it saves precious milliseconds. Done poorly, it shatters cubes.
So the students had to build a cube tough enough to survive their machine.
They customized the internal structure using SLS-printed nylon, a plastic strong enough to withstand brutal forces. The cube was lubricated to counteract the extreme tension. In fact, it was so tight, it couldn’t be turned by hand.
“The cube we use for the record is tensioned incredibly tight, like almost hilariously tight,” Patrohay told The Verge. “You have to really get your wrist into it.”
Each face was controlled by a servo motor connected to a metal shaft mounted in the cube’s center. The robot moved precisely. It used a trapezoidal motion profile, meaning it could ramp up to high speed quickly, then decelerate slowly for exact positioning.
The acceleration is a staggering 12 million degrees per second squared. That’s fast enough to twist plastic apart if you’re not careful.
From Student Project to World Record
The team—Patrohay, Junpei Ota, Aden Hurd, and Alex Berta—first unveiled Purdubik’s Cube at Purdue’s SPARK design competition, where it won first place. But they didn’t stop there. After the event, they continued refining the system, drawing on experience from Purdue’s Cooperative Education Program and support from Purdue’s Institute for Control, Optimization and Networks.
They used their own funds and co-op earnings to build the robot. They sought corporate sponsors and worked late.
“Back in high school, I saw a video of MIT students solving the cube in 380 milliseconds,” said Patrohay. “I thought, ‘That’s a really cool project. I’d love to try and beat it someday.’ Now here I am at Purdue—proving we can go even faster.”
Their creation is an interactive experience. Using a Bluetooth-enabled “Smart Cube”, users can scramble the puzzle in real time. The robot mirrors every move. When the user stops, the cube snaps back to perfection.
“This achievement isn’t just about breaking a record,” Nak-seung Patrick Hyun, assistant professor of electrical and computer engineering, said. “It pushes the boundaries of what synthetic systems can do. It brings us closer to understanding ultra-fast coordinated control systems like those found in nature.”
What’s Next?
Could the team go faster? Yes. But not with plastic.
“If you were to make a completely application-specific Rubik’s Cube out of some sort of carbon fiber composite,” Patrohay said, “then I could imagine you being able to survive at higher speeds… and just being able to survive at higher speeds would then allow you to bring the time down.”
For now, the team is proud of what they’ve done. As Milind Kulkarni, head of Purdue’s electrical and computer engineering school, put it:
“Take brilliant students, give them the tools and opportunities, and they’ll blow your mind. Four undergraduate ECE students, in less than a year, crushed a record set by a world-class team at Mitsubishi. I always say we have the best ECE students in the country — and this proves it.”
