Sep 27, 2007

Guest Post - Steve Hassenplug

Steve Hassenplug (www.teamhassenplug.org) has an interesting little item to share. FLL competitors - take note:

-----

I know this blog usually addresses the LEGO MINDSTORMS NXT, so please bear with me, while I describe a recent event involving a LEGO RCX (predecessor to the NXT).

I just got back from Robothon in Seattle. It was a very fun event. It's always great to hang out with builders who really appreciate each other's robots.

One of the competitions was the Micromouse maze event. In this event, a robot has to navigate through a maze, as fast aspossible. I have a robot that will very reliably get through a maze,if it can simply follow a wall (see Right-Hand or Left-Hand rule), however, one major challenge in a Micromouse event is that the end of the maze can not be reached by simply following the wall. I won't go into all the details about how to solve a maze, but that is an excellent topic for readers to research.

[NOTE: Steve has shared a version of the robot - not the latest, but close here.]

I planned on taking my old reliable robot (called A-Mazing) even though I know there's no possible way it can make it through the maze. Then I had a really interesting idea. Instead of just following the wall all the time, every once in a while (I picked every 30 seconds) the robot would turn 180 degrees, and start following the wall on the opposite side. You may need to draw out some mazes to see exactly what the result of this would be, but it turns out the robot goes from a zero percent chance of getting through the maze, to something greater than zero. Yes, it's much like going through a maze blindfolded, with no real pattern for getting through. But, as it turns out, with a good idea, and more luck than you can calculate, anything could happen.

And, to my surprise, it actually worked. The robot wandered throughthe maze for 30 seconds, and at EXACTLY the right time, made a random turn, which led to the middle (finish) of the maze. The robot went from start to finish in about 45 seconds. That's a far cry from the 17 seconds the winning robot took to finish, but good enough for second place.

I'm not sure I can explain how lucky that timing was. There was a window of about 2 seconds, where the robot could make that turn. Sure, if it missed, it could wander through the maze for several minutes, and maybe get to the end, or maybe not. But, in this case, it worked.

Sometimes, luck plays an important role in robotics. Usually, it's bad luck. But not always!

3 comments:

CATpit contractor said...

Got it, how is that expressed in NQC?

On(OUT_LUCK);

Hery said...

I like your blog. Recently I started my own blog at http://www.legoindonesia.com

Would it be okay if I take some of your post and post it in my blog for my readers?

Jim Kelly said...

Sure, Hery, but please provide a link in each post to the original post, okay? Thanks.

Jim

Related Posts Plugin for WordPress, Blogger...