It's not an RRT[1] if it doesn't employ a Voronoi bias to sample within the largest Voronoi cell. Since the first paragraph describing the "RRT" implementation

It's not an RRT[1] if it doesn't employ a Voronoi bias to sample within the largest Voronoi cell. Since the first paragraph describing the "RRT" implementation admits it didn't explore properly, it really wasn't an RRT. This is a pity, because a proper RRT (or better still, a PRM[2] with the monster walking every edge) would really have been a great way to handle this! [1] Rapidly-Exploring Random Tree. Without the Voronoi bias it's not rapidly-exploring, so it's just a random tree (RT) instead. [2] Probabilistic RoadMap. Similar to an RRT except that samples are connected to all close samples within some neighbourhood function.