Introduction: How to Win MIT Robot Competitions (How to Do MASLab)
Documenting Team 22's entry in MIT's Mobile Autonomous Systems Lab, 2008.
Step 1: Rhett Creighton's MIT Robot Competition-Winning Tips:
From Rhett, winner of 6.270
Here is some advice if you are deadset on winning:
1. First, the best robot often does not win in single or double
elimination tournaments. If there are 5 rounds of single elimination,
that means that you need to have a better than 95% chance of winning
each round to have an 80% chance of winning the whole thing.
2. Most people will build non-functional / barely functional machines.
The same people could have built much better machines in a single day
if they had went with a simpler design.
3. Start planning before IAP. Look at old contest
rules/pictures/tapes. Make sure you have a schedule set up.
4. Make something that works fast. Halfway though IAP, you need to
have something that can score points, and probably would make it to
top 10 in the contest.
5. Spend more time on it than anyone else, from day one.
6. The people running the class want to help you. Use their help
early, as they will have nothing to do early on, and then later the
whole class will want their help to get something working.
7. Get rule clarifications often. In 6.270, we got a clarification
that we could run our machine off of the board power, which made it
much better. The ruling was announced to the all the teams, but no
one else took advantage of it, so it really only benefited us. If you
ask for clarifications, it's likely to be about points that could
8. Day of the contest, use a checklist. Create a setup procedure.
Every year, several teams in 6.270 lose a round because they forget to
turn the power switch on, on their machine.
I asked the staff a bunch of things -
"If my robot were small enough to drive into a goal, I would get double points for both having balls with the chassis, and for having balls in the goal?" - no.
"Can I use biology on my robot?" - sure, bacteria.. whatever.
"What about higher-level animals like mammals: dogs and rodents?" - um.. sure.
So I immediately out to find a dog to train to chase after balls and stash them in goals.
Unfortunately, they modified their decision to:
"Biology that does not smell, goop up your computer, mess up the field, make
sounds or are considered harmful to the organism is considered okay (ie, a
plant) but any use of biology should be cleared with Ben first."
and since I'm unwilling to train a dog to not bark, I listened to Nellie McKay's "The Dog Song" while putting the finishing touches on my mechatronic fetch-bot, instead.
Step 2: Make the Pegbot
Make the pegbot quickly out of the stuff that's given in your kit. This is just to play around and see how things can be put together. Use velcro and cable ties, because it's quick and you will take it apart.
At this point, or even before it, learn about what teams have done successfully in past years.
Step 3: Make a Prototype
Day 3! Assignments 1-3 finished, we can move, sense, and stop when we find a red ball.
Our prototype is finished, and one team member is still controlling the robot from all the way in Mexico.
Step 4: Code Your Robot
Step 5: Design the Robot
You'll want to talk to a whole bunch of smart people. Talk to everyone you know who could have something to contribute.
Writing on a blacklight blackboard can help your design process!
Step 6: Build Your Actual Robot
Design, then build a prototype the next day.
Step 7: Make Your Own Life As Easy As Possible
The basic kit includes a big blue tub. I added backpack straps to my robot box, so I could carry it home, and so my unicycling teammate could bail if he needed to, without throwing the box down. Some people use hand trucks. Other people install X on their robot computers, so that they don't have to deal with a shell. Do what you need to do to build in comfort.
Somewhere around here, I started eating for $3 a day.
Step 8: Feed Your Ears.
Listen to some music, for your own good, and for the good of the robot you're trying to bring into the world.
Unrelated to this step, there were a number of times after at least 4 hours of building with no break that I couldn't go on without watching youtube parodies of Scientology (nsfw), or fake interviews between Weird Al Yankovic and Kevin Federline.
There is no explanation.
Step 9: Celebrate
Our robot, in 3 days, chases after red balls and knows when it's reached one. Hooray!
Step 10: Change Your Design
After drawing it out on paper, we decided to change some things - this is good, but in an intense competition setting, limit it to the first week. We were able to change because we quickly had a working prototype and saw what things could go wrong, and what parts we just didn't like and wanted changed.
Now I'm drawing CAD files of the final design. CAD is recommended, but I'm only using it so I can cut the parts with CAM.
I think it's best to build your prototype without CAD, because drawing takes a long time, and it's easy to have a computer make your stuff, but anyone can do that, and you're not an engineer if you don't know how to make things yourself.
Step 11: Talk to Smart People
I traveled up and down California for a few days. I wanted to travel, but that doesn't mean I took a break. I talked to mechanical engineers and physicists up and down the coast, from Los Angeles to Berkeley, from Idealabs to Squid Labs. I grilled them on my design, and their ideas. The best part about engineering is learning how it's done, and you don't get far unless you're willing to tell someone why you think they're wrong, or ask them why they think they're right, and then learn their thought process. For example - springs do the same thing in compression as under tension. I didn't understand why two mechanical engineers in a row told me "So, have the springs pushing.." on my roller assembly. I asked the second one, 'why pushing?' and he was startled. He hadn't wondered before, that was just how it is taught. So we figured it out right then - springs under compression won't fatigue and wear out like a spring under tension will. Same forces, different lifetime for the spring.
Step 12: Be Ingenius
A high quality speed controller for my team's robot would have cost us around $200, cheap.
So what has speed controllers in it already? Drills do! Good thing I'm stripping the motors out of this $15 battery drill, for my roller assembly. Now I have a use for the speed controller, too. I feel sort of like an Indian after using all parts of a buffalo. There is nothing left here but some grease and a plastic shell. I think I'm even going to use these bearings.
Plus, drill speed controllers are practically the juiciest electronics pre-assembled you can get for cheap. These things can handle 12V at around 7A!
Here's another example -
I needed a roller. I said, well, what has rollers in it? Printers do! So i grabbed a printer (they pretty much rain from the sky around MIT's labs) took out the roller, and stuck it onto my robot.
Then I lost the belt for it, and I figured out how to make and join a new one using heat shrink tubing. I thought, wow, this should be its own instructable!
Step 13: Feature Freeze
Do your teammates think your robot would be a whole lot better if it had an EMP and several rocket boosters on the back? They're probably right, unless you only have a week left to build and test!
STOP building. That's important.
In the end, since most of my teammates weren't around for most of IAP, we didn't stop building, and we didn't ever test, and we did about as expected given that preparation.
You can see my team's logs and final paper
We have a be nice policy.
Please be positive and constructive.