> Important side note: A* or other "best path" algorithms are not always preferable - there are some situations where it makes sense to use even less ideal paths in exchange for greater speed and less memory usage - even ones that are faster than greedy-best-first search with A shown here.*
Another example is when you are dealing with large crowds. Vector fields tend to shine there. A simple example is using a vector field when dealing with a large swarm with one goal:
Something more slightly more advanced yet incredibly powerful would be "continuum crowds", where the agents in the swarm are also obstacles for each other, affecting the vector field:
The intriguing part here is that these potential field can have any source - so it's relatively easy to get complex behaviour, such as pedestrians avoiding cars, or prey/predator behaviour.
Again, these are more relavant with large crowds, so as stated before: the best approach all depends on the particular use-case.
A* is typically for single source, single destination. For crowds, if you have lots of agents that want to go to a single destination, you can often use Breadth First Search or Dijkstra's Algorithm. That tutsplus article uses Breadth First Search.
That is an interesting example with the fluid pathfinder - had not considered a fluid solver for that application but it seems to work really well - even perhaps mimicking nature. If you look at a large flock of Starlings their movement is almost fluid-like: http://www.youtube.com/watch?v=XH-groCeKbE
Another example is when you are dealing with large crowds. Vector fields tend to shine there. A simple example is using a vector field when dealing with a large swarm with one goal:
http://gamedevelopment.tutsplus.com/tutorials/understanding-...
Something more slightly more advanced yet incredibly powerful would be "continuum crowds", where the agents in the swarm are also obstacles for each other, affecting the vector field:
https://www.youtube.com/watch?v=lGOvYyJ6r1c
http://grail.cs.washington.edu/projects/crowd-flows/
The intriguing part here is that these potential field can have any source - so it's relatively easy to get complex behaviour, such as pedestrians avoiding cars, or prey/predator behaviour.
Again, these are more relavant with large crowds, so as stated before: the best approach all depends on the particular use-case.