At the DARPA Robotics Challenge (DRC) Trials in Miami in December, I talked with Marc Raibert, head of Boston Dyanamics about what it takes to pull off extreme innovation.

Raibert co-founded Boston Dynamics in 1992 and his mission ever since has been to create animal-inspired robots, including humanoid robots.

Eight of the teams at the DRC Trials used the Boston Dynamics Atlas robot in the competition. The company’s LS3 (for Legged Squad Support System) has finally realized DARPA’s decades-long dream of a robotic pack animal that is capable of going where wheeled vehicles cannot. Just before the trials, Google announced that it had purchased Boston Dynamics.

I asked Raibert to shed some light on what has made his company such a success over the years. He sounded theme that I have heard many times among successful innovators: build a little, test a little. Raibert explains:

“We have a strategy that we frequently call ‘Build It, Break It, Fix it.’ We make a robot. We expect it to have problems. We take it out in the field, and it breaks. And then we figure out what went wrong. We make a lot of measurements and then we fix it, and then we build it again, break it, fix it. And we go around that loop hundreds of times, maybe break it five times a day.

“That’s why you can see, if you look around at our staff, it’s not only paper and pencil, blackboard guys. There’s also lots of field testing guys, repair guys, analysis people.

“I think that’s the way to make progress; not to try and have the perfect the right out of the gate, but you make a stab and then you can see what you can learn by getting some data back from operating it, and make up for the things that are wrong.”

Watch the full 5-and-a-half minute video of the interview above or at