Anytime Dynamic A* Algorithm

This Program implements the Anytime Dynamic A* algorithm based on the Psuedo code presented in the following path planning paper [1]

"Anytime Dynamic A*: An Anytime, Replanning Algorithm"

How to use the program

Example graph for this demo is as shown below

Step 1: Initalize graph in the Graph.py as shown below

    #define and start and goal node names
        #the same needs to be used in the graph creation
        self.start_node_name = "xs"
        self.goal_node_name = "xg"

        #Create the graph using the node objects
        #Ex node creation: 
        #   node(node_name, {parent:action_cost, ...}, G, rhs, heuristc_cost_to_come)

        #Add created nodes into the node dictionary
        self.nodes["xs"] = node("xs", [], math.inf, math.inf, 0)
        self.nodes["xg"] = node("xg", {"x4":1, "x7": 1, "x8": 1}, math.inf, 0, 5)

        self.nodes["x8"] = node("x8", {"xs":7}, math.inf, math.inf, 1)
        self.nodes["x7"] = node("x7", {"x6":6}, math.inf, math.inf, 3)
        self.nodes["x6"] = node("x6", {"x5":1}, math.inf, math.inf, 2)
        self.nodes["x5"] = node("x5", {"x1":1, "xs":13}, math.inf, math.inf, 1)
        self.nodes["x4"] = node("x4", {"x3":1}, math.inf, math.inf, 8)
        self.nodes["x3"] = node("x3", {"x2":1}, math.inf, math.inf, 6)
        self.nodes["x2"] = node("x2", {"x1":1}, math.inf, math.inf, 5)
        self.nodes["x1"] = node("x1", {"xs":1}, math.inf, math.inf, 1)

Step 2: Run AD A* algorithm and press enter

 python3 AnytimeDynamicAStar.py

The algorithm shall find a fesible path for the given value of epsion (default 4) and same has been illustrated for the example graph as below

Program output:

Illustration:

Step 3: The user will be asked to press enter to reduce the epsilon value and perform iteration again. The results after 5 iterations ie. for epsilon equals 1 is as shown below

Step 4: The user can modify the graph with updated cost for each link seperated by semi-colon as shown below

ex: parent_node,cost,child_node;parent_node,cost,child_node

 Please enter edge updates (ex-format : x1,100,x2;x2,100,x3 ): x1,100,x2

For the above cost a new path will be discovered as shown below.

References

[1] Maxim Likhachev and David Ferguson and Geoffrey Gordon and Anthony (Tony) Stentz and Sebastian Thrun. (2005).

The D* Lite Algorithm (basic version).

Anytime Dynamic A*: An Anytime, Replanning Algorithm, Figure 4.