One of the most frustrating things this season has been entering the year with false expectations about robot performance and reliability. When our kids first reported problems with our NXT I told them 99% of the time they run into erratic behavior it’ll probably be user error. This is only a slight exaggeration based upon my experience as an engineer and technical manager in Silicon Valley.
However, early in the season when I tried to replicate their issues I was utterly frustrated by the results. Still believing user error was the most probable explanation, we burned much valuable time trying to extract a precision and reliability that is just not inherit in the NXT. By the time we upgraded to the eV3, we began to realize that LEGO Mindstorms Kits are not mil spec and a good part of the FFL challenge was selecting a mission solution/attachment that minimized reliance on precision movements and sensors.
The FLL Boards have a good thread on this and a senior member on that board, Dean Hystad, has repeatedly informed rookie coaches of this bug/feature of FLL. Here is one of his many great explanations:
Recipe for reliable missions
I noticed that some of the most frequently viewed posts have titles mentioning problems wit robot turns. Looking back in the archive I see the same was true last year and the year before. Lots of problems with robot turns.
My girls didn’t have many problems with robot turns. Not because they came up with a magical solution that makes the robot drive laser straight and do perfect turns. They didn’t have many problems with turns because they didn’t assume the turns are ever going to be perfect and designed solutions that don’t depend on perfect turns.
The girls had a rule of thumb of no more than two turns without some sort of navigation fix. A looser rule of thumb was no turn required to be more accurate than about 5 degrees and no move required to be more accurate than 5 cm. Put those together and you get missions that will work with a broken Tribot. Their equation for success:
Great Strategy + Good Programming + Crummy Robot = High Score
What I see many teams attempt is this:
Great Robot + OK Programming + Crummy Strategy = ???
Design and programming cannot overcome a bad strategy. Great robots are REALLY hard to build. Programming can’t contribute much when it operates in a vacuum (no information).
The cool thing about game strategy is all you need is logic and creativity. No programming or mechanical design knowledge required. Make up silly challenges and have the the team brainstorm solutions. Can you come up with a reliable way to win “Pin the Tail on the Donkey”? If the first round of suggestions doesn’t include “Remove the Blindfold” you need to break up that box they are thinking inside.
The first year my girls competed there was a mission to put a marble in a slot in a bone (deliver the cancer medicine). The bone was about 18″ from base. Try as hard as they might the mission hardly ever worked. At the qualifier they came in 3rd for robot design and 2nd for programming. The cancer mission failed every time and they finished in the lower half of the robot game.
Good Programming + Good Robot + Crummy Strategy = Poor Success.
At the tournament they saw a team that used a cool attachment that lined the robot up with one of the field models. Neat idea. Wonder if something like that would work with the bone? At the next meeting they tried a few variations on this theme and every one delivered a near 100% success rate. The same robot, the same program, a better strategy and a completely different result.
Good Strategy + Good Programming + Good Robot = Great Success.
Later, to save time, the girls removed some of their sensor use on close missions and added some to combine longer missions. The Cancer mission became a dumb odometry program that used the cool attachment to compensate for error. Even with a weak program the mission was still very successful.
Good Strategy + Good Robot + Poor Programming = Great Success.
I didn’t test the solution with a crummy robot, but I think it would have been successful on any platform that didn’t fall apart in route.
I only wish we had run into Dean’s numerous comments about this aspect of the robot game earlier in the season. Yet, we would’ve missed out on one of the many invaluable lessons Rookie teams only learn through hard knocks.