Thought you guys may be interested to see my (non-navmesh) AI path-finding in action for my upcoming update:
Because bases in my game can be free-built by players and NPC’s, navmeshes didn’t seem like a viable option as I would have to continually update them during run-time. Or at least in my experiments I wasn’t finding much success.
I’ve instead built a somewhat elaborate system (using a lot of traces) which stitches paths together using detection of adjacent floor pieces, cancelling out paths on a floor piece’s 4 axis as it hits dead-ends. I added a bunch of fallbacks for blocking during the pathing. I’ve tested on some fairly complex base-designs and they are able to navigate effectively.
That looks great, Coldscooter. How random was the hunters’ base? It looks pretty well organized. Do the bots ever invade the player’s base? Your game is multiplayer, right? I assume all of the pathfinding must be done on the host computer. (You don’t have dedicated servers, do you?) Do all of the traces have any noticeable cost? How often do the bots update their path? Do you stagger it to avoid hiccups?
Edit: Also, you need to stop the constant tactical reloads! That was seriously hurting your DPS.
@Nathaniel3W I did a lot of bench-marking with traces and was surprised by how performant they are at scale. I can do many thousands per tick without any noticeable lag (not that i use that many).
The bots update their path for each movement. The hunter base is a fixed template. There are currently 4 different base templates they use. Planning on adding many more now i have the new base-interior nav system. The hunters don’t currently invade player bases (although they do attack them and try to destroy them). No dedicated servers yet, only listen hosting using Steam-sockets. That will come soon though.
