Our robot had a simple design; it was 12" wide by 8" long, with back-wheel differential drive powered by 2 motors on a 75-to-1 gearbox per wheel. Three sides were solid lego walls, with the fourth side on the front using a one-way garage to trap every ball in its path. Some pictures can be found in the slideshows below.

Seeding. There were two seeding rounds. We won one round, only to have it cancelled by the opponent's failure to start, and we then did not score in the rematch. We also lost 3-0 to the Cube, since our robot's wheel alignment was off and it could only veer to the wrong scoring area.

Qualifying. We immediately went back from seeding to an empty lab; since all the other teams at seeding had been up all night preparing, they went home to sleep, while those of us who were well-rested worked. An abridged and modified version of our code enabled the robot to turn toward the nearest scoring area, drive over the four balls in its path, and shut itself down in the scoring area. We qualified!

The Tournament. Our morning round was marked by a solid 3-0 victory, which was again taken away due to the other team's failure to start. The second time, we were not so lucky, and even though they again failed to start (incurring a loss), we this time placed two red balls in our scoring area even though green was the winning color. The result was a -2 to 0 double loss. The evening proved no better; we again scored -2 points on ourselves, but the opposing robot gave us an additional -9. We lost, -2 to -11, ending The Dream ™ and our IAP.

Lessons learned. Each team member had valuable learning experiences gained through the hard work put in throughout January 2005. In particular:

▪ Jonathan Karr learned that joining a class late can be tough, especially when you join a week into a month-long course and discover that it's not as exciting as you had hoped.

▪ Gary Sivek learned that he should stay away from all hot objects, especially the powerful combination of solder and batteries, and stick to math. Wonderful, safe math.

▪ Igor Ginzburg didn't really learn anything.

January 25. Intentionally lulling the other teams into a false sense of security, we show up in lab just an hour and a half before impounding, make a few code changes, and test none of them.

January 24. We take the day off again. We need our beauty rest!

January 23. The seeding rounds! Gary and Jon brave a blizzard, while Igor shows up four hours late because he was watching The Godfather. Our robot wins a glorious 3-0 victory, only to have it thrown out because the other team failed to start. We lose the rematch, 0 to 0, and change our robot to only score the four balls behind it and stop. This strategy qualifies, and we then change it back, but this was the only time we ever saw the robot do exactly what it was supposed to do.

January 22. Facing reality, we agreed to focus on a more simplistic strategy given our relative lack of effort throughout the previous three weeks. The strategy is for the robot to:

1. Orient itself in the starting area, and
2. Back up and push four balls into the scoring area immediately behind the starting area.

January 21. Facing mechanical difficulties (specifically the resistance of a 10' length of wire), we abandoned our two ball voting strategy and returned to our one ball voting strategy of January 19.

January 20. Finding the strategy of only voting one ball limited, we decided to think of ways to vote two balls into either bin immediately prior to the end of the game, again using a detachable process. The strategy remanied the same, only now we would be able to vote two balls in to either voting bin.

January 19. Having finished a robot capable of the January 11 strategy, we amended our strategy to include a method for changing the vote count in our favor. Namely we decided to introduce a process to our robot which would:

1. Lower at the start of the game,
2. Slide into position between the two balls immediately in front of the starting area,
3. Detach from the rest of the robot (a wire a lego chain would still connect the two pieces), and
4. Depending on the vote count immediately prior to the end of the game, vote one of the two balls.

January 11. With two weeks left before impounding (336 X 3 = 1008 man hours) we presented an ambitious strategy to our organizer. Specifically we planned for the robot to:

1. Orient itself using four phototransitors positioned at four corners of the robot,
2. Drive forward to catch the two balls in front of the starting area,
3. Drive backward to push the four balls behind the starting area into the closest scoring area,
4. Decide which side of the board to drive to based on the position of the opposing robot,
5. Pick up balls on that side of the playing field which it decides to pursue,
6. Wait until three seconds prior to the end of the game to decide which color scoring area to place the stored balls based on the current vote count, and
7. Repeat steps 1-6 over several rounds and win the competition.

January - December 2004. Dominate the competition.

January 25. Just before impounding, we hot glue the sensors to the robot, build a platform for the swatch, secure the RF receiver in place, and build a little flagpole with a tiny flag made of masking tape with the word "6jew70". Our robot is complete!

January 23. Gary wakes up just in time to write all of the code for the robot before the qualification and seeding rounds. Not really, but we have no idea what his timeline was for the past month.

January 22. Igor rebuilt the robot to increase the gear ratio to 75 from 45. The new robot is more powerful and less error-prone.

January 21. Igor and Jon abandoned the two ball voting detachable process due to unforeseen mechanical complications. They instead focused on the tether design for connecting the main part of the robot to the detachable process. Due to lack of progress on the software and a declining faith in our abilitiy to accurately position the robot, we abandoned the detachable process idea entirely later that day. All three go to IHOP at night and eat pancakes.

January 20. Igor and Jon worked on designs for a more robust detachable process capable of putting two balls into either voting area.

January 19. Igor and Jon worked on designs for the detachable process. Igor prefered a gravity-driven design capable of putting one ball into either voting area.

January 12. Gary leaves lab early to do math.

January 11. Frustrated by lack of success builing a sufficiently compact and reliable syncho-drive wheel, Igor and Jon began to pursue a differential drive design.

January 3. Gary defied all odds and followed published directions to assemble a front-end LEGO™ loader. The front-end loader was later destroyed when we left our work room unguared and unlocked one night. Possible suspects include:

• Matt Roitstein - Motive: Frustrated with his 6.270 team's disappointing 2004 performance and envious of our obvious upcoming 6.270 success, Matt unleashed his anger on our defenseless front-end loader.
• Jonathan Benezry - Motive: Unknown
• Emily Beman - Motive: Unknown, but we think it has something to do with Jonathan.
• David Held - Motive: This guy is generally destructive. MIT watch out.
• Steve Stern - Motive: Expand Google's database to include a winning 6.270 strategy as well as numerous pictures for its image database.
• Paul Peeling - Motive: Steal top secret United States 6.270 strategies for the British military.

The highly touted Team 4 was forged in January 2004 by Igor and Jon's envy of the apparent enjoyment their friends derived from 6.270 that year. Gary was discussed as a possible third team member. His participation was secured in early Fall of 2004. In November the team weathered the tough 6.270 lottery to make it into this year's competition.

Igor, Jon, and Gary met two and a half years ago thanks to the efforts of the 2002 undergraduate members of the Alpha Epsilon Pi Fraternity. Since then the three have become good friends, living as well as often working together, for the past year and a half. The three are looking forward to another exciting year and a half together before they each graduate.

Igor built the braun of the robot including its child process which never made it out of the design phase and wasn't used in competition. Busy working on the 6.270 robot, Igor never made it to Killington this IAP. He did however enjoy a fraternity trip to IHOP.

Igor is the only team member from the midwest. Since moving to the east, he has become a Computer Science and Physics major as well as an avid West Wing fan. He is a proud member of the Alpha Epsilon Pi Fraternity.

Jonathan's pre-IAP hopes of 6.270 glory were quickly dismissed upon returning to Boston a week after the competition began. Jonathan spent much of IAP working for his father's architectural firm, his fraternity, and for a class other than 6.270.

Jonathan hails from the suburbs of Washington, DC. He is a junior majoring in Physics and Brain & Cognitive Science. He is an overworked member of the Alpha Epsilon Pi Fraternity.

Gary gave us the multithreaded brain of our robot. With Igor he developed, debugged, and perfected our winning strategy. Gary also participated in a student-run, TJ-dominated number theory seminar during IAP. With so much free time during IAP Gary is getting bored and is eagerly awaiting his upcoming 84 unit Spring semester.

Gary, along with his evil twin brother, comes to MIT from the opposite side of the Potomac River. He is a Mathematics and Computer Science major. Gary is a classy member of the Alpha Epsilon Pi Fraternity, where he serves as Math Porn Chair.

• defs.c
• mylib.c
• color_sensors.c
• gyro.c
• move.c
• 6jew70.c

• 6.270
• Massachusetts Institute of Technology
• LEGO™