I have been trying to get some local avoidance working. It’s such a crucial part of AI movement and very close to pathfinding.
So after reading quite a lot of papers about it I started coding my own attempt at it.
Here’s the result so far.
It uses a variant of Velocity Obstacles (VO) and I currently sample a number of point around the desired velocity to try to find a good velocity. It has got a number of problems still, but it’s pretty good for a first attempt i.m.o.

I plan to integrate this into the A* Pathfinding Project later on.