The Flying Spaghetti Monster

What is 6.270?

6.270 is an annual class and competition offered at MIT during the month of January. It is a Lego Autonomous Robot Design Competition, in that the robots are built from legos, can operate in the playing field without any input during gameplay, and were designed by us for a competition at the end of the month.

This year's board included six different goals in which to score; each was assigned to one of two robots on the playing field at any time, and each one had different point values, but every goal had the same markings, and a robot had to have enough sensing ability and intelligence to figure out where it was and where it wanted to be. The objective was to come as close to a value set at the beginning of the match as possible. Going over was as bad as being under. In addition, a "skunk ball" was placed on the far side of the board; if the skunk ball was in your starting corner by the end of the round, you would receive 4 additional penalty points to your total score.

Who are you guys?

"Us" above refers to MIT students enrolled in 6.270 over January 2008. The contest is open to all MIT students, although you must enter a lottery to get in, as the equipment is heavily subsidized and spots are limited. Our team, the ones responsible for making a robot in his Noodlyness' image, is:

Sid Creutz, '10

Kendra Pugh, '09

Noah Silverman, '10

Sid enjoys caving, Kendra enjoys EMS and Noah enjoys running. We all like robots.

The Robot

Top

To the right you'll see our submission for the contest, the Flying Spaghetti Monster.  He has a 2-wheel, 2-motor differential drive train, and a 125:1 gear ratio. 

The top-mounted hopper holds six 1.5" balls that are kept in place by a servo-controlled arm, which can direct all balls down either the left or right side of the robot.  Ramps extend from each side to above the board's wall, to shoot balls into baskets behind each goal.  Our robot is intentionally designed to not individually dispense balls, but instead to drop all of them in a single goal.

Bump sensors on either side of the robot are located directly below the ramps and at two different heights to aid in goal detection.  IR sensors are located on each corner of the robot pointing down so that the robot can orient itself at the beginning of a match. 

Bump sensors are located on the front and back of the robot, in case the robot bumps into a wall or competing robot during gameplay.  A servo controls an arm that reaches out from the base of the robot and locks the robot in front of a goal, effectively blocking the other team from scoring.

Many, many hours went into the making of our robot, which had many more sensors than most other robots.  Because of all our components, our robot ended up having a ton of wires grouped together.  We maintain that the robot's similarity to the Flying Spaghetti Monster was not coincidental and in fact an indication of its destiny for greatness.

Our Strategy

Our basic strategy was simple: have our robot pull out of its starting position and sit in front of the opposing team's 4-point goal for the majority of the match, then pull forward and put all of our balls into our own 2-point goal, resulting in a score of 12 points for us.  We figured that this strategy would counter most other robot's strategies, since the 4-point goal was essential to reaching any of the projected scores.  In addition, we figured that most robots would attempt to score their necessary points at the beginning of the match, and that worry about scoring ourselves later would pay off.

It turned out to be a very good idea to decide on this simple strategy early and stick with it, as we didn't waste time trying to implement things we weren't going to use in our final design.  We did have ways to build upon this strategy, such as the locking mechanism that we did develop, rushing our own 4-point goal at the end of the round in an attempt to score more points, etc.

In general, our strategy was very good. Implementation, however...

Our Struggle

Suffice to say we went through 4 bump sensors and 3 IR sensors in the 36 hours before impounding, and stayed up the 26 hours before impounding.  We fell behind, a lot, but were still able to achieve our basic goal and improve upon it a little before we had to turn our robot in.  We suggest to other teams to take their project seriously, early!  But don't forget to have fun.

The Results

We got off to a promising start, winning easily in the first round.  Our robot still had some weaknesses, however, and did not perform well twice in a row.  We were unlucky enough to be assigned to the same board, in the same corner, in the same configuration twice, and after watching our robot fail the first time, we were pretty sure we were done.  It was good to see our robot in action, and somewhat vindicating (if very frustrating) to know that the team that did win used a strategy very similar to ours.  We also beat the 2nd place team in an unofficial round the day before the competition, which, as hallmates and rivals, was fairly satisfying anyway.

In Retrospect

Our plans were good.  Our implementation needed some work, but was adequate, as the competition rounds proved very frustrating to many teams.  6.270 is a great course for those looking for intro to robotics and skills, or just something really fun to do over IAP.  A recommended class!