Mistake in pacman implementation

I found a mistake today in the way I was counting the collisions when pacman is stationary and planning in the metaplanner code. I set the pacman location to the head of the arc being computed rather than the origin position.

The result of this could be why the pacman data does not indicate that it is being eaten as much during planning when compared to A* and Random. I will generate new results with the correction.

I have also implemented anytime repairing A* (ARA*) for a map navigation problem and am in the process of integrating it with the pacman problem as the anytime algorithm to compare against.