Team 25: Mr. Glitch

If you are wrong, he will eat you

Rupert

Our robot had two incarnations. The first was Rupert. Rupert was designed to pick up blocks and sort them into bins, later to deposit them in preferable places. Rupert died a horrible and gruesome death at the hands of his creators when they decided he was too complicated, and he didn't work in the middle of the third week. In fact, he showed no signs that he was ever going to work. RIP Rupert.

Mr. Glitch

Our second attempt was much simpler. Two closable bins on a bulldozer-type robot. The complicated part was the programming - when we finished, we could tell Mr. Glitch where he was, where he was facing, and where he should go, and he would figure out the simplest path there and get there (reliably). Unfortunately, this relied on Mr. Glitch being able to see. In 26-100, Mr. Glitch began to have hallucinations in two of his nine eyes because of poor shielding. He got lost.

Both Rupert and Mr. Glitch shared the same drive train - an adder/subtracter made with two differential gears. Driving one differential would cause both wheels to turn at the same speed in the same direction. Driving the other would cause both wheels to turn at the same speed in opposite directions. Driving a combination of both added the speeds of driving them separately on each wheel - so one motor would drive straight, and the other would adjust turning. We found numerous advantages to this drive system:

• Single motor drove both wheels - the robot could drive almost perfectly straight without any feedback (no turn corrections)
• Both turning in place and turning while driving possible - line following and shaft encoding were both simple.
• Back driving almost impossible - no need to worry about being pushed The only problem we found was that the number of gears in the gear train allowed for a lot of slack - the wheels could turn 10 degrees or so before the gears tightened up when the motors weren't running.

  Our sole method of navigation was line following. Shaft encoding would have been easier, but boring. A shaft encoder robot can only follow very fixed paths, and our entire strategy was based on a dynamic path. We used an array of 9 light sensors on the bottom of the robot to follow the lines, although usually only six were watched at a time. In lab, the sensors picked up the blue jail as black, and we coded for that. However, during the contest (under much different lighting) they picked it up as white. We didn't figure this out until after impounding, and thus couldn't fix it. It was frustrating to watch Mr. Glitch banging against a wall because he couldn't figure out he was in the jail.