Djikstra's and Best First run as you would expect, but I modified the A* algorithm so that the priority was: backDistanceToStart + distanceToGoal * 3. This gives the A* algorithm a much faster search time, but at a slight penalty of "stupid AI" when they curve into certain areas instead of making straight paths like what you get with Djikstra's.
This is the A* search. You can see the results of what I described above in the very first "room" that the agent starts in. Instead of pathing directly downward, the agent goes a bit into the room. It isn't quite as bad as best first, but it is a price you pay for performance. I eventually want to add a slider to A* so that you can individually weight the two different costs for the priority, as it is a fairly important piece of making a good A* algorithm.
A* is the best of both words in my eyes. It is more customize-able to the situation at hand and is both efficient and produces a "semi-intelligent" path.
A* is the best of both words in my eyes. It is more customize-able to the situation at hand and is both efficient and produces a "semi-intelligent" path.
Djikstra's search, as you would expect it. This search should always return the fastest path to run through, but it definitely isn't the fastest to calculate.
Best First. Not the most elegant of solutions, but definitely a fast one given the right circumstances. This is the kind of search you would use for a map like a paintball course, where there are many small obstacles throughout the map, but not very many walls.