Fiery Dung-Beetle of the Apocalypse
This is a dung beetle. It is very good at what it does. Sadly, our robot was not.
I won't give a rewrite the long 6.270 speil, as you should have read it already. So. More to the point. Our team consisted of Tim Abrahamsen, James Wnorowski and Ross Glashan.
Our robot had a reasonably normal design - 2 differentially powered 1:125 wheels, a servoised castor and an overly complex claw/grabber mechanism. The ball collector was split into 2 parts - the large claws that were supposed to pull the ball in, and the forklift 'mouth' that held the ball while we moved. The claws started out closed then opened as we drove up the hill, and then switched between the open and closed position to try pull balls in. When a switch inside the forklift was triggered the claws would lock into the closed position and the forklift would lift the ball up. The forklift was the most mechanically complex component of the robot, but in the end, probably not why we lost. It's main feature was it's 'tri-state' design. Instead of just having 'up' and 'down' positions, it had 'down', 'up' and 'uplock' positions. The uplock position was used when the ball we had just collected was ours, and we wanted to try put it into the lavapit. The forklift would lift up and lock into position, so then when we lowered the forklift down it would just tip forwards instead of lowering, allowing the ball to roll into the cup. While the forklift design worked rather well, we had problems with the claws as the often did not open fully, pusing the balls away instead of pulling them in. This was the reason for our loss in the main competition. We also decided not to go for the lavapit but instead try for only 1 point because we had great difficulties consistantly turning at the bottom corner.
Sensor-wise our robot had 2 back bump sensors, 2 side bump sensors, ball colour / detection sensor, and a start light sensor. The back bump sensors were used to align the robot after reversing into a wall, while the side bump sensors were used to follow the wall up and down the hill.
Our robot's strategy changed during the competition. At first, we planned to place the ball inside the lava island, but later we decided to settle for just a single point. (looking back at the competition, we should have tried for the single point right from the start). After starting, our robot would back up until it hit the wall, and then turn towards the top of the slope. It would then wall follow until it got near the balls, at which point it would deploy its lift and begin using its claws to try and grab a ball. Usually, this is where our robot failed to work properly. At this point, a claw may have hit a ball away, or our robot may have run into the wall and become confused. If our robot did happen to work properly, and find a ball of the correct color, then it backtracked down the hill untill its shaft encoders told it that it was in the scoring area. We hoped to avoid interaction with the other robot, and we did for the most part.
While we did end up winning the mock competition (purely by luck), we did not fare so well in the real thing. The first round our robot competed, its one claw failed to open fully pushing balls away. With one loss already behind us, we were eliminated from the competition. Knowing what we knew after the competition, we would have done many things differently. Firstly we wouldn't have spent so much time trying for the lavapit. In the competition only one robot actually got a ball into the cup. We also would have gone for a simpler mechanical design, as, despite its coolness, the claw was extremely failure-prone. All in all, the competition was great fun for all involved, and despite our robot failing dismally soon into the competition, watching the other robots was highly entertaining.