The Mahdad Miner

This is our robot, the Mahdad Miner. Raar!

The Goal:

Find a 10" by 16" 'lavatube' (drywall box) burried 2-4" beneath the surface of a 14' by 14' box. Excavate it. Drill a 4" hole in it. Plug the hole with an 'airlock'. Pressurize the box to around 3 psi. Disconnect from the airlock and drive away, leaving the box pressurized.
Official contest site:

The Sponsors:

Surveillance Solutions
RF Digital

The Bot:

This is the rack, which slides back and forth to allow drilling and airlock placement without having to drive the bot. As you can see, it is very modular. We used an advanced suspension substitute, originaly conceived of by Dan and prototyped in a 'dansaw special'. Luckily there are no pictures of the prototype, just the beautiful ones Tom made for us.
The bot is controled through an overly complicated serial connection. A laptop running Linux talks to a basic stamp, seen prototypes here. This base station stamp (the small horizontal board) cleans up the command and sends it out over the transceiver (the tall vertical board). On the bot are two more basic stamps, each listening on the same serial line. Each command is comprised of the stamp code, the subsystem it is controling, the command to the subsystem, and the argument for the command (usually duration in milliseconds). Thus a command to, say, drive forward for 3 second would look like: s1 0 1 3000. The client code on the laptop made entering these a little more intuitive: t drive forward 3000 The lava tube is pressurized by a cute little pneumatic system Ryland found, original intended to be used for model airplane landing gear.

The Construction:

The Mahdad Miner without the sliding rack. You can see the pretty but completely non-functional probing mechansim. With a multiple bot design, we could have had multiple non-functional probing mechanisms. The basic mechanical robot -- no electronics have been added at this point.
Sean looking like most of us felt. This was taken sometime early tuesday morning -- maybe 10 hours before the final round of competition. Notice the complete lack of control electronics. The hotel room, looking across the main construction area. Sean is explaining why the bot needs more exposed gears and Lincoln is working more metal shavings into Dan's bed.
More hotel room, detailing the grinding/drilling area. AIR! Yup, that room was pretty trashed. This is looking towards Dan's fortress of solitude. He sat back there for 3 days, soldering. Yay Dan!
Our robot was controlled by relays... and stuff. In the beginning it was fairly orderly... ...though it soon became clear they wouldn't all fit in the project box..
...and then we had to remove them rather forcibly from the box to fix some things... ...leading to the final package. At this point all of the drive electronics had been bypassed (but not removed!) to use a big chunk of relays hanging from the far side and the whole gob had been crammed into the remains of the project box and gaff-taped onto the bot. Well, at least it worked. Mostly. In this image you can see the mystery LED, which was always on for no apparant reason.

The Rivals:

The Eastern Kentucky University robot, first day of competition. We had crashed around 8 after deciding that we couldn't get the bot working in time. Morale was low. Dan and I were woken up around noon to go see EKU compete. We got there just in time to see their airlock fall off. Morale was suddenly much higher.
More of EKU
The EKU team at 'mission control'. They had matching team tshirts! I wish we had done that. The EKU airlock in place, thanks to an 'act of god'.

The Competition:

Somehow we got a very good score for the report/presentation. Don't ask me how! The judges must not care about flash or coherent language, as long as the information is in there somewhere.
The Mahdad Miner, ready for battle.
The FIRST run of the bot. We had literally NEVER tested it before this shot -- 5 minutes before the competition! I do love the tire tracks. Here we (Dan, myself and Lincoln) are in 'mission control'. I'm sorry, but typing on a laptop is orders of magnitude cooler than twiddling with an RC controler.
This was taken right after Jake cleared the sand away for us. Luckily I was talked out of driving straight onto the box -- the slope was much steeper than it looked in grainy b/w! Mahdad had two cameras on it: the obvious one on the boom on top, and this one, underneath. It points where all the drilling and plugging happens -- assuming the sliding rack works, anyway. Whoever thought of this deserves a medal. We never could have inserted the airlock without it. The next bot needs to be just covered in cameras. The boom worked well (we had the best vision systems by far) but was still limited. You just can't have enough POVs.
Success! Well, mostly. It handled like a dream. All of the control and electronics worked. We couldn't sense or excavate. The rack was permantently extended, so the drilling got complicated. But we got points for finding the box given XY coordinates, finding the center of the box, drilling, and starting to place our airlock. Time ran out, but we finished placing the plug to see if we could.
Beautiful, ne?
What it would have looked like, given another day of work and another 30 minutes of competition. Oh well.

Our team, with the spoils of almost-victory.
From left to right: Professor Dunston, Daniel Villa, Rylan Bryant,
Lincoln Ghioni, Jake Parks, Matthew Dockrey, and Sean Caughlan kneeling in front.