Hi! Any updates regarding this? Would really appreciate any help. Thank you!
Ooh, you have set the heuristic scale to 5. While not incorrect per se, it makes the search very greedy and if it needs to search a large region it may need to evaluate a given node many times. This is probably why you get the error.
I’d recommend that you change that back to 1.
Alright I’ll try this. Thank you!
Seems that the heuristic scale was the culprit. I haven’t got any errors when I changed it back to 1. Thank you!
1 Like
Hello. I thought I would give an update regarding this issue. Right now, we encountered the Binary Heap Size problem again (please see the very first image) ever since we switched to bigger maps. We now have approximately around 720k nodes in a map. If an agent calculates a path from edge to edge, the Binary Heap Size error pops up. That being said, I found the issue to be the Manhattan heuristic, I don’t know why, but after switching to Euclidean, the problem was gone. The new problem now is the agents are making to many zigzag paths which is very annoying. So I tried to solve this. I switched back to Manhattan heuristic, but changed the heuristic scale to 0.5 instead of 1. The error does not show anymore but it takes around 2 secs to calculate pathfinding from edge to edge (which is really not a problem for me, I just thought to mention it anyway). So with that, maybe this info can help you find the problem with Manhattan heuristic? Or is there any explanation to why this happens? It appears I am also the only one having this problem. Now, since I changed the heuristic scale to 0.5, will new problems arise? Because the last time, when I changed the heuristic scale to 5, the “1,000,000 nodes searched” error happens. Thank you very much!
1 Like
BTW, the Diagonal Manhattan heuristic also has the Binary Heap Size issue. I also tried changing the heuristic to None. It works great only if the destination node is not surrounded by unwalkable nodes. If it is, it will not try to even go to the nearest walkable node from the destination, instead, it will just stay where it is and the path fails. What we want with our project is to have the agent still move to the destination as close as he can. If he cannot reach the destination, he will stop nearest walkable node. This is achievable with all heuristic. If you can tell me how to achieve the same result without using any heuristic, please please, do so as I am out of ideas. It will be a great help to my project. Thank you again!
Hi
One thing you could try is to download the beta version (https://www.arongranberg.com/astar/download). It’s possible that there have been some fixes related to this.