QuickTime Movie
of MazeBot
1.10 MB

Question 1: What was the general strategy or plan behind your bot design?

My bot followed the right hand wall of the maze.

Question 2: What technologies/mechanisms did you use?

I used an arm with a rotation sensor and clutch (slip) gear to feel if there was a wall on the right. If there was no wall it would turn right and go forward. If there was a wall on the right, it would go forward. If it ran into a wall in the forward direction (detected by a touch sensor bumper), it would back up slightly and turn left.

I also used a rotation sensor on one of the wheels to measure turning and forward displacement.

Question 3: What was the most challenging aspect of construction?

Nothing really tough. I did rebuild once to make it smaller. I was afraid it would not fit down the hallways, but it turned out to have plenty of room.

Question 4: Are there any special features you care to talk about?

The program (in NQC) kept track of the coordinates of the squares that were visited (up to 10) and could be downloaded to a computer. Due to memory limitations I didn't get much farther than that, but I would have liked to have it "forget" unproductive sidepaths so that it could run the maze a second time much more directly. I still think it can be done for the size of maze we were using. I would also love to be able to store that "maze map" in the RCX for use on another run after the RCX had been powered down.

Question 5: What surprised you about how your bot behaved/performed during the event?

The wall-feeling arm did not perform as well as planned. The downward angle of the arm depended on how close the bot was to the wall, and the way I had the rotation sensor geared to the arm (directly) did not allow enough resolution to account for the slop. Also, sometimes it would get stuck in a corner, unable to turn properly. I have solutions in mind for both of those problems if we ever run the Maze again.

It was fun to watch though, which was maybe why it got "Crowd Favorite". Yay!