|Ideas and Preliminary Design|
My computer programmer friend Mark Phillips and I built a two motor walker
together in the early part of 1998. He was interested in learning about robotics and
electronics in general, and I was helping him with building a simple two motor walker. The
walker we built was fairly cool, but we were interested in trying something more
Since he programs for a living, he naturally wanted to make something that
he could program for. I'm strictly a mechanics and analog electronics guy, myself, though
I like using computers. We went to a Seattle Robotics Society (Henceforth referred to as
the SRS) meeting one Saturday, and we talked about some of the different robots being
built for the up and coming fire fighting robot competition. One robot was supposedly
going to use a color Quickcam to process information about the maze and find and put out
the candle. Mark, however, is sharp on computer systems and recognized that the guy didn't
know what he was talking about, as the Quickcam didn't use the NSTC standard for video
information as the man had assumed, rather it had already converte the image to a binary
stream. A lady from Newton labs also caught the misake, and pointed it out.
Newton Labs, a regular at the SRS meetings, specializes in designing and
building machine vision processing systems and it's potential applications of it. They
have built several robots that are amazing. One robot they built will chase an orange ball
around, collect it, and deposit it between a Newton Labs employee's arms. It uses the
Cognachrome vision system to recognize the shape and color of the ball (at 60Hz!), and is
able to follow the ball and grab it with good accuracy, as if it were a little robotic
dog. I downloaded a video once that showed a robotic arm that (I assume) used the same
system to catch a ball thrown through the air at it!
After the meeting, we became interested in the idea of using a Quickcam or
similar camera mounted on a robot to provide all kinds of spiffy functions for our robot.
He assured me that he would be able to design a simple vision system that would be useful
for possible collision detection, etc. Since mechanics are my strong point, I set out to
design us a walking robot. The came the design constraints.
Mark guessed that we would only get a 3FPS frame rate, even if our karma
is especially good. So, I needed to design as stable a walking platform as possible, which
ruled out 2, 3, and 5 motor standard BEAM walkers. They simply bounce around too much.
That left 8 and 12 motor walkers. We didn't really feel like building a four legged, two
motors per leg robot, so the 12 motor, six leg design eventually won out. After some
surfing around on the web, I found:
Which sells a walker called the HexaPod II. We had notions of buying the
kit, but the $375 price tag put us off, especially since we had decided to build two
robots so we could each have one. However, I liked the kit so much, I modeled our design
after it. The first version of the design (One leg only)looked like so: