Local Avoidance pt.2

I have been able to improve the local avoidance by some. I have changed how the limit (the part which is cut off from the VO cone) is calculated, and changed a bit in where the samples are put. I also tried the solution of using half-planes instead of cones as some papers suggest, but it didn’t work out that well, the sampling method with cones worked better.

Worth to note is that the agents don’t use any physics for collision, collision avoidance is purely done by the local avoidance.


  1. Anthony says:

    I rolled my own A* solution a couple of years ago but yesterday I purchased your implementation in anticipation of the local avoidance functionality; it’s important to support our fellow developers. Btw, good job on the A* pathfinding! 😀

Comments are closed.