An other A* library. You can use the node or map mode you like(gird, point, mesh or infinite map).
The library don't need to init a map and don't care the size of your map. We just tell the algo how to get node, neighbors and estimate cost.
Base algorithm logic referenced a-star-lua
A example for grid map
local AStar = require 'astar'
local map = {}
local cached_nodes = {}
-- Node must be able to check if they are the same
-- Cannot directly return different tables for the same coord
-- The library doesn't change nodes, so you able to reuse your node or create a C struct for faster
local function get_node(x, y)
local row = cached_nodes[y]
if not row then row = {}; cached_nodes[y] = row end
local node = row[x]
if not node then node = { x = x, y = y }; row[x] = node end
return node
end
local neighbors_offset = {
{ -1, -1 }, { 0, -1 }, { 1, -1 },
{ -1, 0 }, { 1, 0 },
{ -1, 1 }, { 0, 1 }, { 1, 1 },
}
-- Return all neighbor nodes. Means a target that can be moved from the current node
-- add_neighbor_cb: function(new_node, cost)
-- cost is optional, call map:get_cost to get cost if no cost value
function map:get_neighbors(node, from_node, add_neighbor_cb, userdata)
for i, offset in ipairs(neighbors_offset) do
add_neighbor_cb(get_node(node.x + offset[1], node.y + offset[2]))
end
end
-- Cost of two adjacent nodes.
-- Distance, distance + cost or other comparison value you want
function map:get_cost(from_node, to_node, userdata)
return math.sqrt(math.pow(from_node.x - to_node.x, 2) + math.pow(from_node.y - to_node.y, 2))
end
-- For heuristic. Estimate cost of current node to goal node
-- As close to the real cost as possible
function map:estimate_cost(node, goal_node, userdata)
return self:get_cost(node, goal_node)
end
local finder = AStar.new(map)
local start, goal = get_node(1, 1), get_node(3, 4)
local userdata = { 'mydata' }
local path = finder:find(start, goal, userdata)
if path then
for _, node in ipairs(path) do
print(node.x, node.y)
end
else
print("Not found path")
end
And you can try to run the main.lua
by Love2d
This article tells us how to generate a smooth path with a turning radius.