The Development Process
At the beginning, we brainstormed many different ideas for bots, some of which were seen at the contest itself. We considered leaving the center ball and going for the opponent's side; we considered the quick-whack-and-grab method that the top two contestants used. However, we finally decided that with an arm to get the three center balls, we would have a simple objective--no complex line-following, bump-sensing, or worrying about if we're really moving when the motors are turning. Ideally, the three center balls should give us a score of 5-2, so we would still win even if the opponent managed to steal one of our balls.

Once we had the basic idea of an arm, we began to modify it somewhat. We wanted a wedge mounted on top of the arm in order to deflect the balls as they fall rather than trying to sweep them off after they fell. We decided on an immobile base--we didn't need to move for our strategy, so we planned to weight our base heavily to make sure that we would not, even if another bot rammed us. We toyed with the idea of a secondary, weaker arm that would block the entire length of the arena, but we put that off until the basic robot was constructed.

We decided that since our arm must sweep across--and we wanted strength to not only push balls and overpower other arms, but also to withstand the impact of a falling ball--we would have the arm be powered, with its own wheel and gear train. After a look at the size restrictions, we knew our arm had to initially be bent, but we realized that if it were bent at an obtuse angle, if the first part deployed to a horizontal position, the second part (containing the wedge) would drop down by gravity. We were not completely sure what gear ratio to use for our motors, but not until we had the individual parts mostly assembled did we appreciate the sheer weight of our bot. We knew we had to produce some serious torque, so we constructed a few serious gear trains.

The initial design had all three of the gear trains at 1080 to 1, hence the name of our bot. We did a few trials to determine how many motors we needed in each location; we quickly found that one motor would suffice to turn the wheel on the arm, and we would probably need three to extend the arm and two to turn the base. We then added our secondary arm, which could extend to the opposite side of the table by means of a servo. We put a wheel on the end of this arm to press against the opposite wall and create some serious friction to oppose...whatever might want to knock our arm out of the way. We also mounted the HandyBoard and battery pack (partially as a counterweight since our bot tended to tip over while deploying due to the massive arm).

This caused a bit of a crisis, as the entire arm was now significantly heavier. Two motors could not effectively turn the base in a reasonable amount of time, and three motors--even three motors geared down 1080:1, with a servo helping out--could not deploy our arm! This posed a major problem, as we were lacking in power. We had already determined to use gravity to help us deploy our arm. This time, the power needed was eventually found by looking at the rules, which allow the use of rubber bands to store energy.

Not one, not two, but TEN high-tension rubber bands were mounted to the arm trying to pull it down. This is not enough to pull the arm down from its initial position, but two motors could give the arm an initial "kick" and the rubber bands would end up doing almost all of the work. The rubber bands would not pull the arm past horizontal since the arm would land on the raised platform. This idea worked beautifully; we ended up using two motors to deploy the arm and moving the third to help move the base. Also, at this point, we noticed serious wear on the worm gear in the deployment gear train, so we restructured the gear train to put the torque into two separate worm gears, halving the strain on each.

After this, we made minor adjustments--we found that we still needed a bit more torque to turn our bot, so we geared the base down to 1800:1. Conversely, we found that we could reduce the arm gear ratio down to 360:1 with no ill effects, making our arm swing across a bit faster. Though we were loath to abandon the 1080:1 gear ratio (1080 to 1 being our robot's unofficial nickname at this point--later, for reasons of laziness, adopted as its final name), a friend pointed out that the average of 360, 1080, and 1800 is still 1080. Problem solved.

Main Page