Using BFS to find number of possible paths for an object on a grid

MajesticButterfly :

I have a matrix that represents a grid and would like to find out all possible places an object can move to.

An object can only move horizontally or vertically.

Let's assume that the example below is the grid I'm looking at, which is represented as a 2d matrix. The object is the *, the 0s are empty spaces that an object can move to, and the 1s are walls which the object cannot jump over or go on to.

What is the best way to find all possible movements of this object provided that it can only move horizontally or vertically?

I'd like to print a message saying: "There are 9 places the object can go to." The 9 is for the example below, but I would like it to work for any configuration of the below grid. So all I have to do is give the current coordinates of the * and it will give me the number of possible positions it can move to.

A thing to note is that the *'s original position is not considered in the calculations, which is why for the example below the message would print 9 and not 10.

I have a isaWall method that tells me if the cell is a wall or not. The isaWall method is in a Cell class. Each cell is represented by its coordinates. I looked into using Algorithms like BFS or DFS, but I didn't quite understand how to implement them in this case, as I am not too familiar with the algorithms. I thought of using the Cells as nodes of the graph, but wasn't too sure how to traverse the graph because from the examples I saw online of BFS and DFS, you would usually have a destination node and source node (the source being the position of the *), but I don't really have a destination node in this case. I would really appreciate some help.

EDIT: I checked the website that was recommend in the comments and tried to implement my own version. It unfortunately didn't work. I understand that I have to expand the "frontier" and I basically just translated the expansion code to Java, but it still doesn't work. The website continues explaining the process, but in my case, there is no destination cell to go to. I'd really appreciate an example or a clearer explanation pertaining to my case.

EDIT2: I'm still quite confused by it, can someone please help?

dominicm00 :

While BFS/DFS are commonly used to find connections between a start and end point, that isn't really what they are. BFS/DFS are "graph traversal algorithms," which is a fancy way of saying that they find every point reachable from a start point. DFS (Depth First Search) is easier to implement, so we'll use that for your needs (note: BFS is used when you need to find how far away any point is from the start point, and DFS is used when you only need to go to every point).

I don't know exactly how your data is structured, but I'll assume your map is an array of integers and define some basic functionality (for simplicity's sake I made the start cell 2):

Map.java

import java.awt.*;

public class Map {
    public final int width;
    public final int height;

    private final Cell[][] cells;
    private final Move[] moves;
    private Point startPoint;

    public Map(int[][] mapData) {
        this.width = mapData[0].length;
        this.height = mapData.length;

        cells = new Cell[height][width];
        // define valid movements
        moves = new Move[]{
            new Move(1, 0),
            new Move(-1, 0),
            new Move(0, 1),
            new Move(0, -1)
        };

        generateCells(mapData);
    }

    public Point getStartPoint() {
        return startPoint;
    }

    public void setStartPoint(Point p) {
        if (!isValidLocation(p)) throw new IllegalArgumentException("Invalid point");

        startPoint.setLocation(p);
    }

    public Cell getStartCell() {
        return getCellAtPoint(getStartPoint());
    }

    public Cell getCellAtPoint(Point p) {
        if (!isValidLocation(p)) throw new IllegalArgumentException("Invalid point");

        return cells[p.y][p.x];
    }

    private void generateCells(int[][] mapData) {
        boolean foundStart = false;
        for (int i = 0; i < mapData.length; i++) {
            for (int j = 0; j < mapData[i].length; j++) {
                /*
                0 = empty space
                1 = wall
                2 = starting point
                 */
                if (mapData[i][j] == 2) {
                    if (foundStart) throw new IllegalArgumentException("Cannot have more than one start position");

                    foundStart = true;
                    startPoint = new Point(j, i);
                } else if (mapData[i][j] != 0 && mapData[i][j] != 1) {
                    throw new IllegalArgumentException("Map input data must contain only 0, 1, 2");
                }
                cells[i][j] = new Cell(j, i, mapData[i][j] == 1);
            }
        }

        if (!foundStart) throw new IllegalArgumentException("No start point in map data");
        // Add all cells adjacencies based on up, down, left, right movement
        generateAdj();
    }

    private void generateAdj() {
        for (int i = 0; i < cells.length; i++) {
            for (int j = 0; j < cells[i].length; j++) {
                for (Move move : moves) {
                    Point p2 = new Point(j + move.getX(), i + move.getY());
                    if (isValidLocation(p2)) {
                        cells[i][j].addAdjCell(cells[p2.y][p2.x]);
                    }
                }
            }
        }
    }

    private boolean isValidLocation(Point p) {
        if (p == null) throw new IllegalArgumentException("Point cannot be null");

        return (p.x >= 0 && p.y >= 0) && (p.y < cells.length && p.x < cells[p.y].length);
    }

    private class Move {
        private int x;
        private int y;

        public Move(int x, int y) {
            this.x = x;
            this.y = y;
        }

        public int getX() {
            return x;
        }

        public int getY() {
            return y;
        }
    }
}

Cell.java

import java.util.LinkedList;

public class Cell {
    public final int x;
    public final int y;
    public final boolean isWall;
    private final LinkedList<Cell> adjCells;

    public Cell(int x, int y, boolean isWall) {
        if (x < 0 || y < 0) throw new IllegalArgumentException("x, y must be greater than 0");

        this.x = x;
        this.y = y;
        this.isWall = isWall;

        adjCells = new LinkedList<>();
    }

    public void addAdjCell(Cell c) {
        if (c == null) throw new IllegalArgumentException("Cell cannot be null");

        adjCells.add(c);
    }

    public LinkedList<Cell> getAdjCells() {
        return adjCells;
    }
}

Now to write our DFS function. A DFS recursively touches every reachable cell once with the following steps:

Mark current cell as visited
Loop through each adjacent cell
If the cell has not already been visited, DFS that cell, and add the number of cells adjacent to that cell to the current tally
Return the number of cells adjacent to the current cell + 1

You can see a visualization of this here. With all the helper functionality we wrote already, this is pretty simple:

MapHelper.java

class MapHelper {
    public static int countReachableCells(Map map) {
        if (map == null) throw new IllegalArgumentException("Arguments cannot be null");
        boolean[][] visited = new boolean[map.height][map.width];

        // subtract one to exclude starting point
        return dfs(map.getStartCell(), visited) - 1;
    }

    private static int dfs(Cell currentCell, boolean[][] visited) {
        visited[currentCell.y][currentCell.x] = true;
        int touchedCells = 0;

        for (Cell adjCell : currentCell.getAdjCells()) {
            if (!adjCell.isWall && !visited[adjCell.y][adjCell.x]) {
                touchedCells += dfs(adjCell, visited);
            }
        }

        return ++touchedCells;
    }
}

And that's it! Let me know if you need any explanations about the code.

Using BFS to find number of possible paths for an object on a grid

Guess you like