It's a puzzle

A Chicago man spends months building a robot of Legos and programing it to solve the Rubik's Cube . . . but why?

By MIKE WILSON

© St. Petersburg Times,
published October 29, 2001

[Photo courtesy of Jonathan P. Brown] Using a digital camera, an electric motor and lots of Legos, consultant Jonathan P. Brown has mastered the nemesis of many, the Rubik’s Cube.

The Kind of Genius the World Needs Right Now answers the phone with the brevity and precision you'd expect of a visionary engineer: "This is J.P."

Not "Hello." Not "Jonathan P. Brown speaking." Just the initials -- perfectly functional, all you need to know.

It's a Tuesday morning, and Brown is in his office in Chicago. We're calling to talk to him about creativity, about curiosity, about initiative, about dreams. We're calling to tell him we believe he is the Kind of Genius the World Needs Right Now, and see what he says.

But first we need to ask him how -- and why -- he built a Lego robot that can solve Rubik's Cube.

"I was downstairs in my neighbor's apartment. They had a Rubik's Cube kicking about," he says. "And I was just casting around for something interesting to do with the Lego system at the time."

For kicks, and for months, Brown, 37, tinkered with his Lego MindStorms set (which includes a programmable microprocessor), his Rubik's Cube and his home computer. He built a robot out of Legos, put the cube in its "hands" and set up a digital camera that looked at the cube and told his computer which colors showed on its face.

The computer, loaded with Rubik's Cube-solving software, told the robot which way to twist the cube. It took about 20 twists for the robot to solve it.

It took about five minutes for Brown to become a celebrity among Lego hobbyists. We heard about him when the New York Times ran a story along with a picture of the robot and Brown. "It's a "geek in the headlights' picture -- please don't use it," he says. So we're not.

On the phone, Brown explains that he is not an architect or engineer, though he knows a thing or two about those subjects. He works in one of the most specialized fields we've ever heard of: He's a consultant on issues of temperature and relative humidity in historic buildings.

When Independence Hall in Philadelphia put in a new air conditioning system, Brown conducted studies to see whether the cool air was damaging any of the historic artifacts. He has done similar work at the Mount Vernon George Washington Mansion and Abraham Lincoln's summer cottage in Washington.

"All quite ironic because I'm actually British," he says. He is married to Rachel Fulton, a professor of medieval history at the University of Chicago. Their 5-year-old son, Rush -- named after Benjamin Rush, a signatory to the Declaration of Independence -- has his own Legos.

"As you can imagine, they are kept separate. Underlined. We do borrow bits from each other, under very controlled, mutual-hostages situations," Brown says.

We asked Brown how long he has been playing with Legos.

"I prefer the term "building,' really. Since '99," he says. "I was at a party in Research Triangle Park in North Carolina, and the host's son had some really fabulous Lego Technic. I just thought it was really nifty."

So nifty that he went out and got a MindStorms set, with its programmable component.

"What was really disappointing about it from a grown-up point of view was that the Lego programming environment was for people who had never done any programming at all. It was very limiting. But later I discovered that there's this whole subculture of adult Lego fans -- AFOLs, or adult fans of Lego, which I find a bit tacky. (Their Web site is www.lugnet.com.) One of these people had written a more grown-up programming language, essentially an implementation of the C programming language, that would work on the bricks.

"So then I really started getting interested in it."

Then came the visit to the neighbor's apartment, where he saw the Rubik's Cube, one of those cubes you have to twist until each side is a single color. Brown wondered how hard it would be to get a computer to tell a robot how to solve the cube. (Why anyone would wonder something like this is a question we'll get to by and by.)

"I remembered seeing an article on a BASIC program that could generate a solution to the Rubik's Cube. I thought, if I just could find that article again I'd have a solution. And since one of the things Lego is quite good at is making grippers and grabbers, I thought I could come up with a way to (implement) the solution."\

The grippers and grabbers weren't as strong as he had hoped.

"The problem is that Rubik's Cubes are somewhat stiff. And it turns out they're sufficiently stiff that if you grab the cube with the Lego grabber and try to turn it, either the fingers just get prized apart, or no rotation happens because the motor locks up. It isn't strong enough to turn the face around. It became clear that something was going to have to give in this process," Brown says.

So he did what anyone would do who was having trouble getting a strong grip on a Rubik's Cube with a computer-controlled robot. He got on the Web.

"I found a very interesting Web site (www.lar5.com) by a guy named Lars Petrus, about speed cubing -- these guys who get together and try to solve the cube as fast as they can. They lubricate the cube with silicon lubricant to make the faces move more easily.

"What's tricky about that is that the solvent seems to actually dissolve the plastic of the Rubik's Cube, so while you're waiting for the solvent to dry you have to keep turning the cube to keep the plastic from deteriorating.

"What I ended up doing was wedging plastic knives between the faces of the cube to try to stretch the springs out a bit. I was desperate at the time. It worked."

The grippers grabbed and the cube twisted. Now it was time to program the computer.

Brown found software that could solve the cube. But to get the computer to send instructions to the robot, he first had to tell it which colors were showing on each of the six faces of the cube -- and each of the six faces has nine colored squares. This meant coding in a 54-digit number.

"It was very easy to get one of the orders for one of the faces wrong," he says.

That was when he got out his Lego digital video camera, which is actually made by Logitech. He thought he'd be able to find color-recognition software that would help the camera "see" the cube, but he couldn't. So he wrote his own. It took about three months.

"When it's properly calibrated I get about one mistake in every two cubes. It mistakes white for yellow or orange to red," he says.

We asked Brown to talk about the first time the robot solved the cube.

"I didn't really have a "Eureka!' moment. It was very incremental. It was very very slow and gradual until I got something that worked. And then I submitted it to the ultimate test. I asked my neighbors to come up and see it work, and it jammed. . . . I just kept changing small things, and eventually we got it to the point where it worked."

That was in July. Now he calls the robot the Cube Solver. But don't get the idea that it's a highly sophisticated piece of machinery.

"We're really sort of at the edge of what Lego will do here," he says. "We're not talking about the Boeing 777 of robots. We're talking about the Wright Brothers and Kitty Hawk. I've done about 40 or 50 cubes now. It turns out that it's worked every time. But it still has a paper-clips-and-chewing-gum feeling to the whole thing."

We asked Brown if he could solve a Rubik's Cube without a computer and a robot.

"Oh God, no," he says. "And I felt quite bad about that initially, because I was interacting with a lot of people who are real Rubik's Cube nuts." He caught himself. "Enthusiasts would be a much better word. I felt this terrible inadequacy. . . . But as time went on through 2001, I started to feel much more comfortable with the idea that I was going to make a machine to do it. I can't run 17 miles an hour, but if I could make a car to do it, I could be a help to people who can't run 17 miles an hour."

The cube wasn't going to help anybody do anything. So -- this is the big question -- why bother?

"It's something you do just to see if you can do it," he says. "I thought it was an amusing thing to do."

It occurred to us that this impulse -- the simple wish to know what you can accomplish -- is at the very root of creativity and innovation. Without that impulse in clever human beings, we wouldn't have computers or the Hoover Dam or the Sears Tower. And without it we'll never get the things we need to continue surviving on this torn planet. That impulse can save the world.

We're not saying J.P. Brown is the genius the world needs right now. Maybe he is, maybe he isn't. We're just saying he is the Kind of Genius the World Needs Right Now.

To him, that just sounds weird.

"It didn't feel very creative at the time," he says. "It felt like a very hard slog."

Well, all right. One last question: Does the Cube Solver have a future?

"If anyone wants to give it a future, that's fine with me. I do have an invitation to demonstrate it at the International Puzzle Conference in Chicago in 2002," he says. And Lego has asked him to make copies that it can use to promote the line at trade shows. But mostly, Brown says, "this Cube Solver thing is for fun.

"Though if someone takes the idea and makes a lot of money, I'd be happy to have some. I'm not disdaining money."

© St. Petersburg Times. All rights reserved.