Mistake in pacman implemetation (follow up)

I corrected the mistake of counting the number of collisions from the wrong starting location during planning for the metaplanner. The results are slightly worse as expected.

I found another error in planning location for the corrected random planner (i.e. the random planner cannot execute until a feasible plan to the goal has been found). Rather than setting the start location to be the head of the arc being planned, the start location was always the pacman origin. This would cause pacman to have to plan much more than was necessary resulting in more planning collisions. I expect the results of the random algorithm to improve.


I found an even more egregious error: I never ran the corrected version of the random planner which was in a script file named testRand2. Instead the results generated was from the original testRand. I will be fixing this too.