The only logical thing I can see this being useful for is in the case of Roblox pathfinding failing to make an alternative route/path – then again using :MoveTo or attempting a different route position are alternatives already without a new system entirely. Without having the ability to query the NavMesh I do not think it will be superior then Roblox’s own systems.
I’d be unable to give comparative feedback compared to Roblox on the limited detail here as @tyridge77 points out I believe.
My game implements an in-house Dijkstra’s algo for mapping roads as it’s simply better for our use case and distance but that is not feasible universally hence why it’s in house. Therefore I do not think your approach is going to be useful universally as PathfindingService and I would be critical of how this can be used as a better approach compared to the already existent ones.
As I mentioned in my argument with tyrdige. I’m comparing it with pathfinding service based on accuracy and the ability to make a path with complex map geometry.
Therefore, it’s simply useless to repeat the previous argument once more.
If you have any valuable feedback by which I could potentially improve this, like pointing out actual flaws then please list them. If you have don’t, then please read the above argument and go on with your day.
As @tyridge77 said, navmesh would be 10x better and more performance optimal. I do think this is a really good demonstration of an A* pathfinding system, but this won’t help your NPC interact with the environment, it seems like mostly what your doing is point and click navigation which is a really good idea for certain games, navmesh would allow for it to find it’s way around basically every obstacle, now that’s not to say yours can’t but I would use navmesh with this.
You’re right, just like M_etrics and tyridge77 (not pinging to avoid disturbance). And I’m also right in my case, which is that this is simply a showcase, it’s not done, it’s meant to show off the fact of how accurate my implementation is and how it can solve obstacle courses. I won’t go for a navigation mesh system because it’s a bit trickier than a simple voxel-based system, and it’s not always accurate (Roblox navigation mesh and Unity navigation mesh as examples.)