# Simplest way to generate a random path

What's the simplest way to generate a random "path" for a level in a game? I don't know what search terms I should use to read more. I've looked into maze generation but thats not quite right.

The end result should look like this:

• For a more "controlled" random, you might want to use something akin to Perlin Noise. – notlesh Nov 26 '11 at 23:35

As long as you don't want to go "back", meaning that your path goes only into a single direction (let's assume it is going downwards on your image), you can use the following C# code to easily generate those paths:

[Flags]
public enum Direction
{
Down=1,
Left=2,
Right=4
}

public class Path : List<Direction> {}

public Direction GetNewDirection(Direction allowed, Random rnd)
{
Direction newd;
int maxd = Enum.GetValues(typeof(Direction)).Length;
int[] vals = (int[])Enum.GetValues(typeof(Direction));
do
{
var t = rnd.Next(0,maxd);
newd = (Direction) vals[t];
}
while ((newd & allowed) == 0);
return newd;
}

public Path GenerateRandomPath(int startx, int starty, int endx, int endy, double prob)
{
Path newpath = new Path();
Random rnd = new Random();

int curx = startx; int cury = starty; Direction curd = Direction.Right;
Direction newd = curd;

while (!(curx == endx && cury == endy))
{
if (rnd.NextDouble() <= prob) // let's generate a turn
{

do
{
if (curx == endx) newd = GetNewDirection(Direction.Left | Direction.Down, rnd);
else if (cury == endy) newd = Direction.Right;
else if (curx <= 0) newd =  GetNewDirection(Direction.Right | Direction.Down, rnd);
else newd =  GetNewDirection(Direction.Right | Direction.Down | Direction.Left, rnd);

}
while ( (newd | curd) == (Direction.Left | Direction.Right)); // excluding going back

curd = newd;
switch(newd)
{
case Direction.Left:
curx--;
break;
case Direction.Right:
curx++;
break;
case Direction.Down:
cury++;
break;
}
}

}

return newpath;
}


You can generate the path by calling it:

var mynewpath = GenerateRandomPath(0,0,10,10,0.3);


Here the first four parameters are start and end coordinates of your field (determining the dimensions) and the last parameter is the curving probability. Hope it helps...

• +1 For having a variable called newd. Also, for a fairly straightforward and simple solution. – Richard Marskell - Drackir Nov 27 '11 at 6:57