Team 12: The Sultan of Swat


People Inspiration
Most of our ideas came from observing robots from past contests.
Strategy
We wanted our strategy to be simple yet effective. We chose to build a stationary non-orienting robot that spins a long arm. The arm would hit all the balls off the pedestal and send them to our side. We also wanted the arm to be strong enough to prevent other robots from gaining a foothold on the pedestal.

Implementation
It took us about 4 days to build a working robot. On the first day we built a turret module capable of supporting the arm as it spun. The following day we put together the actual arm, being careful to make it not only long enough and strong enough to hit the balls, but also as light as possible. On the third day we assembled a gearbox around the turret; we tried a 105:1 gear ratio and found it ideal. Finally, we built a base that held the handyboard and batteries, and provided traction with the ground. Our exceedingly simple strategy provided for very, very easy coding. Check it! Note that the scrolling marquee that we implemented for our LCD(which sadly remained unused) takes up more lines than the code that actually makes our robot go.

Bird's eye biew Turret and gearbox
Bird's eye view of the robot
The turret and gearbox of the robot

We were careful to pay attention to the mechanical aspects of our robot. In particular, the turret and gearbox were carefully put together to minimize wear and failure. Though some trouble spots were left in the design due to limited room (specifically, the 24-tooth gears that mesh with the turret are mounted very badly), we at least knew where to look for signs of trouble.

The information our robot needed from the environment is whether the arm is spinning or stopped against an obstacle. We first used a breakbeam sensor to detect rotation of a pulley in the gearbox, but due to different lighting in 26-100 this sensor failed to detect rotation so the arm thought it was perpetually lodged against an opponent. The following day the breakbeam sensor was replaced by two mechanical switches that are pressed twice each time the arm completes one rotation. To hell with fancy optical sensors, we said.

If the robot did decide it was stuck against an obstacle, we wrote code that would stop the arm from going forward, rotate it backwards for a half turn, and then try to go forwards again. We hoped this would create a nice battering effect on opponents.

Sultan of Swat
The final product

Results
Despite the relatively high ability of our robot to score an additional three points consistently, its lack of any defensive strategy was its eventual demise.

Mock Contest: We scored a double win in the first round of the mock contest (3-3, they stole one ball and knocked the other into the neutral zone while we collected all three balls off the pedestal), but we were eliminated in the second round when we faced a robot with a car that drove onto the pedestal.

Round 1: We faced off against team 10. Unfortunately, due to the different lighting conditions in 26-100 our optical breakbeam sensor failed to record the rotation of the pulley, so our robot went forwards for 3/4 seconds and then back for 1/2 second, for 60 seconds. It nearly tore itself apart. Team 10 collected the first ball off the pedestal and stole one of our balls, while we collected the remaining two balls in the middle for a final score of 4-3. The following day, The Tech described the match:

"Team 12's masochistic robot used its arm to hit the balls onto it's own side. However, the uncontrollable flailing motion was so powerful that the robot began to lose pieces of its arm with each whipping motion."

Round 2: A stroke of good luck pitted us against a team we could handle, team 7. They failed to orient correctly so we won easily with a score of 5-2.

Round 3: More bad luck for us as we are matched up against team 42, D.P. Team 42 mopped the floor with our arm robot. It was very very mean, and it stole all our balls. Bastard. Final score, 7-0. Note that team 42 was probably one of the best 5 robots and, according to the TAs, had the best code of any of the robots int he competition. D.P. was later taken out by team 10.

The 6.270 Page