A*算法Java

smu*_*shi 1 java algorithm a-star

我整个周末都玩这个.我试图将节点存储在PriorityQueue数据结构中.我的astar函数似乎没有做它应该做的事情.有人介意看看吗?

public void aStar(Node from, Node to) {

    PriorityQueue<Node> exploreList = new PriorityQueue<Node>();
    ArrayList<Node> visited = new ArrayList<Node>();
    ArrayList<Node> successors = new ArrayList<Node>();

    Node current = from;
    System.out.println(current.getName());
    while (current != to) {
            successors = current.getConnected();
            Collections.sort(successors);
            for (Node n : successors) {
                    if (!visited.contains(n)) {
                            exploreList.add(n);
                    }
                    for (Node n1 : successors) {
                            if (n.fSum() > n1.fSum()) {
                            exploreList.remove(n);
                            exploreList.add(n1);
                            }      
                    }
            }
            visited.add(current);
            current = exploreList.remove();
            System.out.println(current.getName());
    }
Run Code Online (Sandbox Code Playgroud)

节点类在这里

   public class Node implements Comparable {
   private String name;
   private int travDist;
   private int straightDist;
   private ArrayList<Arc> arcs;

/**
 * Constructor for a new node
 * 
 * @param n
 */
   public Node(String n, int aTravDist, int aStraightDist) {
       name = n;
       travDist = aTravDist;
       straightDist = aStraightDist;

       arcs = new ArrayList<Arc>();
  }

/**
 * Adds a new arc
 * 
 * @param to
 * @param c
 */
public void addArc(Node to, int c) {
    arcs.add(new Arc(to, c));
}

/**
 * Gets the list of connected nodes to this node
 * 
 * @return
 */
public ArrayList<Node> getConnected() {
    ArrayList<Node> returnData = new ArrayList<Node>();
    for (Arc a : arcs) {
        returnData.add(a.getNode());
    }
    return returnData;
}

@Override
public int compareTo(Object o) {
    //return name.compareTo(((Node) o).getName());
    Integer sum = ((Node)o).fSum();
    return sum.compareTo(fSum());
}

public int fSum () {

    return travDist + straightDist;
}

/**
 * Gets the name of the Node
 * 
 * @return
 */
public String getName() {
    return name;
}


}
Run Code Online (Sandbox Code Playgroud)

Bor*_*jev 6

你在做什么不是一个合适的A星算法.

Collections.sort(successors);
Run Code Online (Sandbox Code Playgroud)

你不应该这样做.在A star中,你总是考虑所有的继任者.您不必担心订单 - 优先级队列将负责处理.但是,添加此行会增加算法的复杂性.

for (Node n1 : successors) {
  if (n.fSum() > n1.fSum()) {
    exploreList.remove(n);
    exploreList.add(n1);
  }      
}
Run Code Online (Sandbox Code Playgroud)

这是完全错误的.你在这里做的是:你只添加最接近的所有后继者.这将是一个光束搜索,光束大小为1,而不是一个星星 - 只需将它们全部保留.