Recastnavigation uses path cost and heuristic based on distance to find
shortest path by A* algorithm. Using raw speed values makes cost much lower
value than heuristic (1000 times less at maximum). Heuristic is defined as
distance from node to target * 0.999 that means area cost should never be less
than 0.999 or 1 for simplicity.
Use max speed to make lowest possible cost factor equal to 1.
Even when target is not reachable actor will try to run there either because
target navmesh polygon is selected within extended area or because partial path
is built to the closest possible polygon.
Actor may reach destination before collision. Get next path point from active
packet pathfinder and use point tolerance to reduce distance.
Use maximum of last destination tolerance and DEFAULT_TOLERANCE because
there are 2 checks made for path completion. First checks whether actor is
close enough to the destination and second drop last path point when actor is
closer than DEFAULT_TOLERANCE.
Division by zero causes float value to be infinite. When infinite cost is
multiplied by zero distance the result is NaN. After this pathfinding algorithm
state is broken.
When distance between start and end point is greater than max radius of area
possibly covered by navmesh there is no way to find path via navmesh. Also if
distance is greater than cell size navmesh might not exists withing mentioned
area because cell is not loaded therefore navmesh is not generated. So minumum
of these values is used to limit max path distance. Assuming that path
actually exists it's possible to build path to the edge of a circle. When
actor reaches initial edge path is built further. However it will not be
optimal.
This allows to distribute AI reaction calls over time.
Before this change actors appearing at the same frame will react in the same
frame over and over because AI reaction period is constant. It creates a
non-uniform CPU usage over frames. If a single frame has too many AI reactions
it may cause stuttering when there are too many actors on a scene for current
system.
Another concern is a synchronization of actions between creatures and NPC.
They start to go or hit at the same frame that is unnatural.
Refactoring related to "smooth movement"
See merge request OpenMW/openmw!285
(cherry picked from commit 6eaf0a389d5aed3b74ab1a7cf89574612f964bdf)
e847b4c8 Split getSpeed() to getMaxSpeed() and getCurrentSpeed()
a96c46bc Refactor calculation of movement.mSpeedFactor
03ee9090 Use getMaxSpeed instead of getCurrentSpeed where it makes sense.
a178af5c Create helper functions `normalizeAngle` and `rotateVec2f`
Assume actor is stuck when it's not able to move in the destination
direction with maximum speed. Approach to check moved distance from the
previous point doesn't work good for slow and big actors. When they face
obstacle they're trying to move along with oscillation so check is
passing but they don't get any closer to the destination.
Ignore z coordinate for not swimming nor flying actors to calculate
distance from actor destination to last path point. If walking actor
destination point is floating above the ground then a point on navmesh
may be too far away when z coordinate is included. In this case path
will be rebuild on each AI_REACTION_TIME.
