Search code examples
javacollision-detectionmaze

How to do collision detection with many walls (maze)?

In my game, the player navigates a maze. I can't figure out how to do proper collision detection with the walls. It is easy to do collision detection for staying in a certain area:

if (x > rightWallX - playerWidth) x = rightWallX - playerWidth;
if (x < leftWallX) x = leftWallX;
//...

But how would I do collision detection for many walls?

I can do plain collision detection without correction (like if (intersecting) return true;), but I can't correct this correctly. If I just store the old x and y and reset them, then

  1. The object never actually touches the wall
  2. If the object can go up but is blocked to the right, it won't go up, it will just not move.

How is collision detection in a maze done?

Solution

  • The easiest way, once you have solved collision detection, to fix the collision is to move the actor to the closest valid position to where the actor would be were it not for the object it collides with. This assumes no inertia, but it is sufficient for maze-like games or top-down map-crawling games.

    If you want to simplify your calculations further, you can limit yourself to detecting if changing the actor's x or y coordinate would be better. If your actor has an axis-aligned rectangular hit-box and all obstacles are axis-aligned rectangular as well (the simplest case), this assumption is indeed correct. However, the results might not be satisfactory in some other cases (potential artifact: speed boost from gliding diagonal walls - not the case in most maze games).

    Keep in mind multiple collisions could happen concurrently (pushing against two walls). If there are no sharp angles between two walls that an actor could both intersect (say, if all your obstacles are axis aligned and sufficiently spaced), fixing each collision in turn will suffice - just don't stop after the first collision.