How can obstacles be made effective in findPath calculations?

[liushzhql]

Hello,

I am currently developing a feature for ground navigation lines, where players control movement using a joystick controller. This feature involves recalculating and rendering the path from the character's current position to the target point every second. As a result, I utilize the findPath function exclusively, without the implementation of an Agent.

I have encountered an issue where obstacles added after the mesh build do not participate in the findPath calculation. Could you advise on how to resolve this problem?

[isaac-mason]

Hi @liushzhql!

Paths should respect tilecache obstacles - I've created a new example to confirm this is working ok: https://recast-navigation-js.isaacmason.com/?path=/story/tilecache-obstacles--path-example

One thing to check is whether you've called tileCache.update() after adding/removing obstacles to update the navmesh: https://docs.recast-navigation-js.isaacmason.com/classes/index.TileCache.html#update

If you can share a reproduction of your issue I'm happy to take a look 🙂

[rob-myers]

Hey @liushzhql , @isaac-mason I think I just ran into the same problem.

When using a smaller cs of 0.05 I had to update the tileCache many times after adding an obstacle.

const obstacle = api.nav.tileCache.addBoxObstacle(position, extent, angle);
for (let i = 0; i < 5; i++) {
  console.log('addBoxObstacle', id, `update ${i}`);
  if (api.nav.tileCache.update(api.nav.navMesh).upToDate) break;
}

These updates could be spread out, to avoid blocking the main thread.

This should probably be mentioned in the README:

https://github.com/isaac-mason/recast-navigation-js?tab=readme-ov-file#temporary-obstacles

[GGAlanSmithee]

I might also be hitting this. After adding obstacles and calling tileCache.update, nothing happens. I will try your method @rob-myers to see if I can brute force it to work.

What's the logic behind it working after updating the tilecache many times btw? Is it a race condition? If so, maybe the update could be done after a timeout instead?

[liushzhql]

@rob-myers I apologize for the delay in seeing this, but it seems the issue has already been resolved :)

[liushzhql]

@isaac-mason I'm sorry for that the performance testing hasn't been conducted yet. As I mentioned earlier, it’s just a fallback compatibility solution.
But I will share the findings immediately once available. Thanks for your understanding!

[isaac-mason]

No worries @liushzhql! I was just curious 🙂

[rob-myers]

I believe this is expected behaviour from recast/detour. Maybe the re computation of the tile should be spread over several frames, to avoid jank.

…

On Fri, 5 Apr 2024, 15:23 Alan Smithee, ***@***.***> wrote: I might also be hitting this. After adding obstacles and calling tileCache.update, nothing happens. I will try your method @rob-myers <https://github.com/rob-myers> to see if I can brute force it to work. What's the logic behind it working after updating the tilecache many times btw? Is it a race condition? If so, maybe the update could be done after a timeout instead? — Reply to this email directly, view it on GitHub <#272 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAMU5KLXFGSHU5DAIHGYFODY32XXRAVCNFSM6AAAAABD3RC5CWVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TAMRSGM2TO> . You are receiving this because you were mentioned.Message ID: <isaac-mason/recast-navigation-js/repo-discussions/272/comments/9022357@ github.com>

[GGAlanSmithee]

thanks for the info @rob-myers, this is working for me now! I used a setInterval without an interval to split it over several frames:

let times = 0;
const interval = setInterval(() => {
  tileCache.update(navMesh);
  if (times++ >= 5) clearInterval(interval);
});

[GGAlanSmithee]

or, more cerrectly

const interval = setInterval(() => {
  const done = tileCache.update(navMesh).upToDate;

  if (done) clearInterval(interval);
});

[isaac-mason]

Here's some related detour docs: https://recastnav.com/classdtTileCache.html#a93f931ad563e5cadcc389acdffd4892e

I think we can improve the documentation in our readme

Updates the tile cache by rebuilding tiles touched by unfinished obstacle requests.
[out] - upToDate - Whether the tile cache is fully up to date with obstacle requests and tile rebuilds. If the tile cache is up to date another (immediate) call to update will have no effect; otherwise another call will continue processing obstacle requests and tile rebuilds.

[isaac-mason]

I've made some improvements to the TileCache docs in the readme, let me know if we can make it clearer: 6b8eae7

[liushzhql]

Thanks for the improvements. The documentation is much clearer now and here is a clear description of how tilecache.update works.