LuAstar, a simple A* pathfinding class

[kikito]

This lib has those inconveniences (the flexibility part; I don't know about the speed) because it is coupled with a particular map implementation. As soon as you get out of those needs, it's just not flexible enough.

I'm not saying this to bash Roland; he's sharing a lib so others can use it, and that's great. I actually started doing my own A*, but doing it "the flexible way" takes some effort - I still haven't finished.

Maybe you guys want to give a look at TSerial. I just gave it a cursory look myself. At the very least it uses threads, which is something Jasoco is requesting.

[Taehl]

I think you mean TLpath?

I will remark that my implementation, while maybe not being perfectly easy to use (I tried, though! It's a very tough problem!), is quite optimal in the path-finding itself (or so I believe).

[Jasoco]

I was thinking of taking the LuaStar code and dissecting it and implementing path finders into my "person" objects so the part that does the finding can be run in stages from within their update function instead of in a closed loop that pauses the code. So basically it'll be the person will just stand still while it thinks, and then will move. At least it wouldn't freeze the rest of the game for every person.

This of course means figuring out exactly how it works at the code level.

[kikito]

I think you mean TLpath?

I will remark that my implementation, while maybe not being perfectly easy to use (I tried, though! It's a very tough problem!), is quite optimal in the path-finding itself (or so I believe).

I meant TLPath. I got distracted by the other link to TSerial, apologies.

[coffee]

I was thinking of taking the LuaStar code and dissecting it and implementing path finders into my "person" objects so the part that does the finding can be run in stages from within their update function instead of in a closed loop that pauses the code. So basically it'll be the person will just stand still while it thinks, and then will move. At least it wouldn't freeze the rest of the game for every person.

This of course means figuring out exactly how it works at the code level.

You probably already saw this around. For sure this can help you (a lot) if you want to do it from scratch or improve one.
http://www-cs-students.stanford.edu/~am ... html#paths

[Nixola]

I tried to use this library to generate a tile-based map with a walkable path (tile number 2 is walkable), but I get the following error:

Code: Select all

Error: ./lib/astar.lua:39: attempt to compare number with table
stack traceback:
	./lib/astar.lua:39: in function 'inBound'
	./lib/astar.lua:233: in function 'setInitialNode'
	./lib/functions.lua:16: in function 'create'

The code in function.lua (the file that generates the error (and the map) is this:

Code: Select all

function map.create()
	for y = 1, tilemap.h do
		map[y] = {}
		for x = 1, tilemap.w do
			map[y][x] = math.random(1,5)
			if map[y][x] ~= 5 then
				map[y][x] = 1
			else map[y][x] = 2
			end
		end
	end
	astar(map)
	map.start = {math.random(tilemap.w),math.random(tilemap.h)}
	map.goal  = {math.random(tilemap.w),math.random(tilemap.h)}
	-- If I uncomment the lines below, I have the error...
	astar:setInitialNode(map.start)
	astar:setFinalNode(map.goal)
	enemies.path = astar:getPath()
	if not enemies.path then
		love.graphics.print('Loading...',0,0)
		map.recreate()
	end
	-- If I uncomment the lines above, I have the error...
end

function map.recreate()
map.create()
end

Link to the .love file: http://dl.dropbox.com/u/15081595/Doesn%27t%20work.love

[Roland_Yonaba]

I've been working on this one. I've re-started it from zero.
Actually I am trying trying to have Jump Point Search working.
It works fine, but there are still some odds with diagonal moves.
I'll give heads up soon.

[Roland_Yonaba]

Done.