Is this tool a good fit for our project?

Hi everyone,

We are currently working on a multiplayer fantasy shooter, and I’m just about to start work on implementing bots for people to practice against.
To handle the pathfinding part of the equation, I have been looking at this tool, and from what I have read so far it seems like a pretty good match for what we need it to do, but I have a few questions before jumping into the deep end.

  1. How well does the pathfinding perform at avoiding dynamic obstacles? There will be players running around all over the place that the agent needs to avoid bumping into.

  2. Is it relatively simple for an agent to get back onto the recast graph once it gets knocked off? There are several skills in our game that knocks characters around the map, allowing an agent to be knocked on top of trees, buildings and rocks. It should ideally be possible for that agent to get back onto the graph without it looking completely out of place.

  3. How much control do you have over recast graphs when generating them at runtime? We have a map editor planned, which will allow both us and our players to make maps for the game in a tool that is separate from the unity editor, which means we have to rely on generating recast graphs only at runtime.

Thanks in advance!

  1. There is local avoidance included in the project, but unfortunately it is more suited towards crowd simulation and higher density scenarios. It does not work that well with low density and high velocity scenarios (which your use case seems to be).

  2. That depends on the movement script, you can detect if it is not on the navmesh anymore and get the closest point to the navmesh, but how you play animations to make them move back to the navmesh is very game specific so this system doesn’t handle that.

  3. Everything you can do in the editor can be done during runtime. Including scanning graphs and changing settings.