The One Robot's (TOR's) design was guided by the team's strategy to avoid any climbing obstacles by driving under the collapsing portion of the bridge. This approach also reduced travel time by taking a more direct and reliable route (could ride the north wall) to the target area with fewer turns. The same line follow program could be used to get the 4 corner loops, the high loops, and set up the run under the bridge, simplifying programming and consuming less on-board memory. TOR was designed with simplicity in mind, trying to pick up all the loops with the same basic programming and attachment (lift arm).
It was obviously a risky strategy, because TOR would have to be no more than 3.4" high to fit under the bridge with a very tight plan volume to maneuver in front of the loops on high poles. This meant designing more of a swiss watch than an FLL robot. There would be very little free space inside for passengers, design changes, and working room.
The microprocessor (brick) didn't fit in a conventional orientation, even with small tires. The brick depth directly on top of any drive motors was taller than the height criterion. Finally it was decided that the brick would be located sideways on the back. This orientation did cause a bit of occasional inconvenience when trying to push buttons with the back of the robot flat against a wall and when exhanging batteries, but on the positive side the brick provided significant lateral stability against sideways racking.
The yellow, field kit bridge roadway pieces were harvested to make very strong, but light plates for the roof and front touch plate. As a result, our field always looked like it had suffered significant collateral damage. The plates could be ordered in red or yellow. Emma felt a red robot would draw more attention, and possibly inspire fear, hopefully prompting the competition to forfeit at the starting line. That never happened, but TOR IS very recognizable.
Our Tasmanian friends sent us the article attached below that outlined the principles of creating a PID (Proportional, Integral, and Differential) line following program. Kjersti had created our line following program last year, without any conceptual help, other than to focus on the light sensors' inherent area averaging, and to target the grey instead of bouncing off black and white. Based on that experience, and the principles of PID control systems, she was able to create a fairly smooth and quick line following program.
During the first tournament it was recognized that the arm would slip teeth occasionally, and a more secure attachment between the lift arm main gear and drive gear would be required. Violet created this in just a few minutes one day, and gear teeth slippage was eliminated.
During the second tournament, TOR inexplicably lost the line 3 out of its 5 runs, the first time we'd seen that behavior. Antipodes could not replicate line loss on the practice tables, so it was difficult to troubleshoot during the tournament. Before the Norcal Championships Kjersti tried many different recipes for P, I and D factors, but realized the values she had been using all along were were optimal, and just reduced the line follow speed slightly to hopefully ensure reliability. Concurrently, Violet added a light shield around the light sensors, resulting in light readings with 99% accuracy. TOR has not lost the line since.
Emma built a CAD model of TOR, as a means of providing a collaboration tool with our Australian counterparts.