If I were smarter, I'd have thought of this at the beginning of the season:
This might make for some worthwhile experiments for your teams to do. The challenge: prove (or disprove) my assertions below.
For example: start the robot in the NE corner of base (i.e. manually aimed) and have it run some simple program where the robot drives 5-6 feet in all.
Record the robot's ending position. Repeat several times, and record the results.
Next, start the robot in the SE corner of base (butted up against the W and S wall; use bumpers or a jig). Repeat the experiment. This will give you a good idea how much variability, if any, is added by aiming the robot.
You can similarly experiment with the light situation. Record the reflected light values when the color sensor is over various spots on the field. Open or close the shades as appropriate, then measure reflected light again. How do the values change? Then run the experiment again, this time calibrating the color sensor right before measuring. What changed?