NPR logo

Competing in the Government's Robot Race

  • Download
  • <iframe src="" width="100%" height="290" frameborder="0" scrolling="no" title="NPR embedded audio player">
  • Transcript
Competing in the Government's Robot Race


Competing in the Government's Robot Race

Competing in the Government's Robot Race

  • Download
  • <iframe src="" width="100%" height="290" frameborder="0" scrolling="no" title="NPR embedded audio player">
  • Transcript

Renee Montagne talks with Michael Griffis, owner of Eigenpoint Company. Griffis helped design a robot that is competing in this year's DARPA (Defense Advanced Research Projects Agency) Grand Challenge, a government-sponsored race of autonomous robotic vehicles in the California desert.


Dozens of robots are competing for a $2 million prize in a race sponsored by the Pentagon to encourage development of unmanned vehicles. The robots are in qualifying rounds in the Southern California town of Fontana. They have to navigate without a driver or remote control more than two miles of narrow roads, tunnels and mountain switchback. Michael Griffis has a robot in the race, a rather sizeable one.

Mr. MICHAEL GRIFFIS (Owner, Eigenpoint Company): It's pretty big. It's pretty bit. We're--What?--six, seven foot wide and about six, seven foot high. It's an off-road vehicle, four-wheel drive, it's high off the ground and we modified the design of it to hold about 20 computers. So we needed to provide power and cooling and all the engineering stuff that you need to run the computers.

MONTAGNE: How exactly does it make its way through the course on its own?

Mr. GRIFFIS: We use GPS. We use inertial navigation. What we do is we use a lot of sensors to figure out where it is. Right before the qualifier, they'll give us a route, and in the route there'll be stuff like a tunnel. They'll find some abandoned cars and put them out there as obstacles that we'll have to navigate around.

MONTAGNE: How much are you linked to the robot, this unmanned vehicle?

Mr. GRIFFIS: During the event or during a run, we're spectators.

MONTAGNE: Huh. It's out there making its own decisions.

Mr. GRIFFIS: That's right. For example, we came upon a hill, and if you think about it, if you have a sensor looking out and you're seeing a hill, well, that's an obstacle. So your system has to be smart enough to see that you're actually changing elevation and then that's not really an obstacle because you'll be driving over it. Does that make sense?

MONTAGNE: Oh, yeah.

Mr. GRIFFIS: Yeah. So anyway, we came upon this one hill and the robot stopped and it took it about 20 or 30 seconds to realize, `Hey, this is just a hill.'

MONTAGNE: It's sort of endearing, almost, to think it was, you know, scratching its head and you really couldn't help it.

Mr. GRIFFIS: That's what it looked like. That's exactly what it looked like. Then it realizes, `OK, I'm just looking at a hill.' And then you know everything's good after that.

MONTAGNE: And how fast can it go?

Mr. GRIFFIS: The course that we have now is about two and a half miles, and it took our robot 13 minutes to complete it. But as a point of reference, 2004, it took about three days for anyone to complete the course, a much easier course last year. So we've really come a long way in a year--`we' being the competitors, all of the teams.

MONTAGNE: Michael Griffis, good luck to you.

Mr. GRIFFIS: OK. Thank you very much.

MONTAGNE: Michael Griffis owns the Eigenpoint Company, a robotics engineering firm in High Springs, Florida.

This is MORNING EDITION from NPR News. With Steve Inskeep in New Orleans,I'm Renee Montagne.

Copyright © 2005 NPR. All rights reserved. Visit our website terms of use and permissions pages at for further information.

NPR transcripts are created on a rush deadline by Verb8tm, Inc., an NPR contractor, and produced using a proprietary transcription process developed with NPR. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.