13,354,605 members (62,308 online)
alternative version

#### Stats

151.4K views
48 bookmarked
Posted 17 Mar 2005

# Very simple A* algorithm implementation

, 17 Mar 2005
 Rate this:
Simple A* pathfinding algorithm implementation for beginners

## Introduction

This is about A* algorithm implementation which is about the way how we can find a best path between two positions. I already know that there are other A* implementations in this codeproject site. They are good, but I bet this is more simple and an easy implementation for beginners to understand.

There is no unnecessary code in this implementation, I just implement the A* algorithm pseudocode step by step in very intuitive ways. As you can see in the picture above, '#' means wall, each dot means available road, and 'o' means path that AI finds.

Shorty, A* is the most popular pathfinding algorithm to go from start to goal, based on efficiency of movement cost.

You can visit A* Pathfinding for Beginners(http://www.policyalmanac.org/games/aStarTutorial.htm) to learn how this algorithm works.

You can see pseudocode for A* that I used in this implementation at Justin Heyes-Jones A* tutorial. Visit here (http://www.geocities.com/jheyesjones/pseudocode.html) to see orginal pseudocode. The print-out of this pseudocode will help you a lot to understand this implementation.

## Class Design

There are three primary classes in this implementation.

• Map - This class represents a map.
• Node - This class represents each tile on the map. It has two primary methods.
• `CompareTo` - This method allows us to decide which node is better. We can say current node is better if this method returns negative number, which means current node has lower cost than the other node being compared.
• `isMatch` - This allows us to decide whether two node's geomatical positions are same or not.
• SortedCostNodeList - This is a list that stores `Node `object list. We need to get off the lowest cost node from the list, so this list is implemented as sorted list order by cost value of the node, and we don't need to examine the costs of all elements in the list every time to pop the node which has the lowest cost because they are already sorted. Just pop one.This list has two primary methods.
• `push` - add node elements to the list at proper position order by node cost.
• Node's `CompareTo` method is used internally to sort order by cost.
`pop` - just returns the lowest cost node from the list and remove it from the list.

## Implementation

This is the core loop of the algorithm.

```ArrayList SolutionPathList = new ArrayList();

//Create a node containing the goal state node_goal
Node node_goal = new Node(null,null,1,15,15);

//Create a node containing the start state node_start
Node node_start = new Node(null,node_goal,1,0,0);

//Create OPEN and CLOSED list
SortedCostNodeList OPEN = new SortedCostNodeList ();
SortedCostNodeList CLOSED = new SortedCostNodeList ();

//Put node_start on the OPEN list
OPEN.push (node_start);

//while the OPEN list is not empty
while (OPEN.Count>0)
{
//Get the node off the open list
//with the lowest f and call it node_current
Node node_current = OPEN.pop ();

//if node_current is the same state as node_goal we
//have found the solution;
//break from the while loop;
if (node_current.isMatch (node_goal))
{
node_goal.parentNode = node_current.parentNode ;
break;
}

//Generate each state node_successor that can come after node_current
ArrayList successors = node_current.GetSuccessors ();

//for each node_successor or node_current
foreach (Node node_successor in successors)
{
//Set the cost of node_successor to be the cost of node_current plus
//the cost to get to node_successor from node_current
//--> already set while we were getting successors

//find node_successor on the OPEN list
int oFound = OPEN.IndexOf (node_successor);

//if node_successor is on the OPEN list but the existing one is as good
//or better then discard this successor and continue

if (oFound>0)
{
Node existing_node = OPEN.NodeAt (oFound);
if (existing_node.CompareTo (node_current) <= 0)
continue;
}

//find node_successor on the CLOSED list
int cFound = CLOSED.IndexOf (node_successor);

//if node_successor is on the CLOSED list
//but the existing one is as good
//or better then discard this successor and continue;
if (cFound>0)
{
Node existing_node = CLOSED.NodeAt (cFound);
if (existing_node.CompareTo (node_current) <= 0 )
continue;
}

//Remove occurences of node_successor from OPEN and CLOSED
if (oFound!=-1)
OPEN.RemoveAt (oFound);
if (cFound!=-1)
CLOSED.RemoveAt (cFound);

//Set the parent of node_successor to node_current;
//--> already set while we were getting successors

//Set h to be the estimated distance to node_goal
//(Using heuristic function)
//--> already set while we were getting successors

//Add node_successor to the OPEN list
OPEN.push (node_successor);

}
//Add node_current to the CLOSED list
CLOSED.push (node_current);
}
```

Once we get to the goal, follow parent nodes to find the solution path.

```//follow the parentNode from goal to start node
//to find solution
Node p = node_goal;
while(p != null)
{
SolutionPathList.Insert(0,p);
p = p.parentNode ;
}
```

This A* implementation is very simple and good for beginners who want to know how A* algorithm works. Change map data in variety of ways, and check out how AI is smart to find the good path. Enjoy your programming.

A list of licenses authors might use can be found here

## Share

 Korea (Republic of)
Working in SK Communications,South Korea
as a web site developer.

## You may also be interested in...

 First Prev Next
 Diagonals Member 1170670330-Sep-15 0:53 Member 11706703 30-Sep-15 0:53
 Freeze Member 1066570423-Feb-15 8:32 Member 10665704 23-Feb-15 8:32
 Re: Freeze Member 1066570423-Feb-15 8:35 Member 10665704 23-Feb-15 8:35
 Thank you! Member 879312930-Jul-14 19:07 Member 8793129 30-Jul-14 19:07
 Good job Member 1061706722-Feb-14 6:27 Member 10617067 22-Feb-14 6:27
 [Message Deleted] Danny Rodriguez27-Jan-08 10:19 Danny Rodriguez 27-Jan-08 10:19
 I realy needed this.Thanks. toslavdsf329-Oct-07 15:22 toslavdsf3 29-Oct-07 15:22
 What about isometric maps? worldspawn6913-Jul-07 18:18 worldspawn69 13-Jul-07 18:18
 Just what I was looking for. Henize18-Sep-06 6:27 Henize 18-Sep-06 6:27
 MapLittle Mistake Djooneagain15-Sep-05 23:39 Djooneagain 15-Sep-05 23:39
 An issue in the algorithm LynxRaven30-Mar-05 19:37 LynxRaven 30-Mar-05 19:37
 I have now seen this issue in two algorithms I have looked at... If someone is walking through a maze, why would they make an unnecessary diagonal step... I believe that a horizontal or vertical step should be worth 1, and a diagonal step should be worth 1.5 (as diagonal would we worth 1.414 more than a horizontal or vertical step). If you were to give a cost to a diagonal movement of 1.5, there would be no unnecessary diagonal movements unless they were shorter than a vertical or horizontal movement...
 Re: An issue in the algorithm Hyrum2-Dec-05 8:25 Hyrum 2-Dec-05 8:25