I decided to build my bot on a synchronous drive platform. At the time this seemed like a good idea - I had always wanted to build a synchro-drive, and having one in the maze would eliminate having to worry about precise steering control on a conventional bot.

My basic plan was simple: on top of the synchro platform, place a touch sensor in each direction. I predicted that I might rub up against the wall when moving and thereby active a touch sensor, so I needed to be able to read each one individually. I was going to use a touch-sensor multiplexer from Hi-Technic, but Lego's decision to stop selling the wire assortment temporarily halted Hi-Technic's production. My solution was to wire the two opposing touch sensors to each input; this consumed two of my inputs, but would still allow me to not be confused if a side rubbed against the wall. Rotation sensors would also be used to provide exact turning and advancing movements.

Over the course of about four weeks, I built two prototype units. I hit a major snag when I discovered that I couldn't plug two rotation sensors into the same input on the RCX - this meant that I was one input short. So I ended up not using a rotation sensor on the drive motor, instead relying on dead reckoning.

As I built the final version, I was feeling pretty confident. The actual event disabused me of that notion, however. My bot had one major problem: the wrap-around bumper that I was so proud of tended to bind up on the walls. This, in turn, completely ruined the dead-reckoning logic. So I spent most of my time in the maze stuck against walls and corners.

So the bot never made it more than a few squares into the maze. None of the bots entered were resounding successes; most were better than mine, but I think the reality of the challenge was much different (and much harder) from what we had expected.

This image, and the next four, just show each side of the bot from the top. I wanted the entire bot to fit into a six-inch cube, but didn't quite make it - the three sensor inputs stacked up on input port 1 caused the bot to be slightly over six inches tall.
These next three pics show the bot from the top as I start to strip off the top pieces.
The geared motor in the corner drives the wheels; the 24-tooth crown gear in the center is mounted to the drive shaft, which runs down to the rather complicated distribution gearing in the bottom layer of the bot. You can also see the rotation sensor, which is driven by a shaft driven off of the turntables; this measures the current angle of the axles.
In this slightly blurry shot, another layer of the shell has been stripped off. You can see the top of one of the two micromotors used for steering peeking out. The gear visible in the front left are part of the power distribution gearing.
On the right side, you can see a little more of the distribution gearing.
A view from the bottom. The light sensor is used to stop the bot when it encounters a black mark on a square - the theory was that this would allow it to stop when it solved the maze. The problem, of course, was that it never actually solved anything.
One of the steering motors pulled out.
The next three shots show most of the distribution gearing. The multiple changes-of-direction were necessary to get all of the wheels to turn in the same direction. I could have simplified it by swapping some of the axles themselves, but I wanted a challenge.

Back to the Lego index.