Maybe a bug about astar 4.2.8

as shown in the gif and the test project
link: https://pan.baidu.com/s/1gQASdOqNFm4Nbi0aDWMjyA
code: a9r7

Hi

Hmm… That edge really should be connected. Can you visualize it using the ‘Show Connections’ mode and see if there is a connection between those two nodes.

More generally paths on navmeshes can be longer than the optimal route due to how pathfinding works on navmeshes. The RichAI movement script has some internal code (called funnel simplification) which can often make it find more optimal paths than the AIPath script.

See also https://arongranberg.com/astar/docs/getstarted2.html#navmeshnotes

Thanks, here is the connections picture, and it shows the edge not be connected.

Hi, I find an interesting thing about the bug, maybe it can help you to solve the issue, here is the gif.

Interesting. Would it be possible for you to share this test scene with me?

I can’t upload the project here, and I have shared it here, you can download it with the code below.
link: https://pan.baidu.com/s/1gQASdOqNFm4Nbi0aDWMjyA
code: a9r7

Hi, when walkableClimb is zero, MaxTileConnectionEdgeDistance * MaxTileConnectionEdgeDistance will be zero, and it will cause the bug.
bug0

Hi

Ah, I see. When you modified the character height it clamped the walkable height to 0 (since it has to be smaller than the character height). That’s what caused it. I have changed that setting to be a delayed float field now. So the clamping will not happen until you deselect the control.

This fix will be included in the next update.