Java-Gaming.org Hi !
 Featured games (90) games approved by the League of Dukes Games in Showcase (683) Games in Android Showcase (196) games submitted by our members Games in WIP (750) games currently in development
 News: Read the Java Gaming Resources, or peek at the official Java tutorials
Pages: [1]
 ignore  |  Print
 Faulty circle-circle collision handling  (Read 1446 times) 0 Members and 1 Guest are viewing this topic.
Aaron

Junior Newbie

Java games rock!

 « Posted 2005-05-23 20:55:44 »

Hello.  I'm designing a game, and I'm trying to work out the collision detection and handling.

I made a test application with two balls; one that you move around with the arrow keys.  I copied the algorithm described in this article.  Generally when you try to move one ball over the other it's made to only go far enough to touch it, preventing overlap.  However, there's still some overlap at the left and right sides of the stationary ball, and sometimes when coming from just off the top or bottom the movable ball can be blocked even when a small gap is shown between the balls.  Only when you try to collide from the top or bottom do you get good results.

This is my method for handling collisions:
 1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60  61  62  63  64  65  66  67  68  69  70  71  72  73  74  75  76  77  78  79  80  81 `private boolean testForCollision(float Nx, float Ny) {        // Ball B is stationary.        float moveVecX = A.dx();        float moveVecY = A.dy();        float moveVecMag = (float)Math.sqrt((moveVecX * moveVecX)           + (moveVecY * moveVecY));                // Early Escape test: if the length of the movevec is less        // than distance between the centers of these circles minus        // their radii, there's no way they can hit.        double dist = Math.sqrt(((B.x() - A.x()) * (B.x() - A.x()))           + ((B.x() - A.x()) * (B.x() - A.x()))) - (A.radius() + B.radius());                if(moveVecMag < dist) {            return false;        }                // Normalize the movevec        Nx = moveVecX / moveVecMag;        Ny = moveVecY / moveVecMag;        // Find C, the vector from the center of the moving        // circle A to the center of B        float Cx = B.x() - A.x();        float Cy = B.y() - A.y();        // D = N . C = NxCx + NyCy        double D = (Nx * Cx) + (Ny * Cy);        // Another early escape: Make sure that A is moving        // towards B! If the dot product between the movevec and        // B.center - A.center is less that or equal to 0,        // A isn't isn't moving towards B        if(D <= 0) {            return false;        }                // Find the length of the vector C        double lengthC = Math.sqrt((Cx * Cx) + (Cy * Cy));        double F = (lengthC * lengthC) - (D * D);        // Escape test: if the closest that A will get to B        // is more than the sum of their radii, there's no        // way they are going collide        double sumRadiiSquared = (A.radius() + B.radius())           * (A.radius() + B.radius());        if(F >= sumRadiiSquared) {            return false;        }                // We now have F and sumRadii, two sides of a right triangle.        // Use these to find the third side, sqrt(T)        double T = sumRadiiSquared - F;        // If there is no such right triangle with sides length of        // sumRadii and sqrt(f), T will probably be less than 0.        // Better to check now than perform a square root of a        // negative number.        if(T < 0) {            return false;        }                // Therefore the distance the circle has to travel along        // movevec is D - sqrt(T)        double distance = D - Math.sqrt(T);        // Finally, make sure that the distance A has to move        // to touch B is not greater than the magnitude of the        // movement vector.        if(moveVecMag < distance) {            return false;        }                // Set the length of the movevec so that the circles        // will just touch        Nx *= distance;        Ny *= distance;                return true;    }`

If testForCollision() returns false, the ball is translated by its normal dx and dy.  Otherwise it's translated by Nx and Ny, which should bring it to the surface of the other ball.

Any help would be apprieciated.

UPDATE
Well, I could never figure out what was wrong, so I ended up redoing it.  It's based a lot on what I tried to do before, and I'm not sure I can pin-point what's different that made the fix.  What I did change was that this time I find the time when the balls collide, then translate them accordingly.  This is what I came up with:
 1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60  61  62  63  64 `private double calculateTranslation(Ball A, Ball B) {            // Relative position.          double dPx = B.x() - A.x();            double dPy = B.y() - A.y();            double dP = Math.sqrt((dPx * dPx) + (dPy * dPy));                        // Relative velocity.            double dVx = B.dx() - A.dx();            double dVy = B.dy() - A.dy();            double dV = Math.sqrt((dVx * dVx) + (dVy * dVy));                      double minDist = (B.radius() + A.radius());                        // Distance between A and B.            double dist = dP - minDist;          if(dV < dist)                  return 1.0;          // Normalize dV.          double dVNx = dVx / dV;            double dVNy = dVy / dV;                        // Are A and B moving away from each other?            // Find dot product of dVN and dP.          double dotProd = (dVNx * dPx) + (dVNy * dPy);          if (dotProd >= 0)            return 1.0;                        // dotProd is also the component of dP parallel to the normal of dV.            // This can be used to find the shortest line between the center of B            // and dV, which we'll call F.            // dP is the hypotenuse.          double Fsq = (dP * dP) - (dotProd * dotProd);                        double minDistSq = minDist * minDist;                      // Escape test: if the closest that A will get to B          // is more than the sum of their radii, there's no          // way they are going collide          if (Fsq >= minDistSq)                  return 1.0;          // F and minDist form a right triangle (minDist the hypotenuse) with a            // third line we'll call T.  Tsq is the square of T.          double Tsq = minDistSq - Fsq;                        // Check that Tsq isn't negative.            if (Tsq < 0)                  return 1.0;                      // Therefore the distance the circle has to travel along          // movevec is dotProd - T          double distFinal = Math.abs(dotProd) - Math.sqrt(Tsq);          // Finally, make sure that the distance A has to move          // to touch B is not greater than the magnitude of the          // movement vector.          if (dV <= distFinal)                  return 1.0;                      // Find the time it will take for A & B to make contact.            double time = distFinal / dV;          return time;}`

I'm going to keep this since I haven't encountered any problems like I did with the last method, and I prefer getting the time value anyway.
Pages: [1]
 ignore  |  Print

You cannot reply to this message, because it is very, very old.

 orrenravid (188 views) 2016-07-16 03:57:23 theagentd (259 views) 2016-07-11 14:28:54 Hydroque (341 views) 2016-07-06 05:56:57 Hydroque (492 views) 2016-07-03 08:52:54 GrandCastle (375 views) 2016-07-01 09:13:47 GrandCastle (366 views) 2016-07-01 09:09:45 CopyableCougar4 (421 views) 2016-06-25 16:56:52 Hydroque (401 views) 2016-06-22 02:17:53 SwampChicken (365 views) 2016-06-20 13:22:57 SwampChicken (298 views) 2016-06-20 13:22:49
 Archive 25x Brynn 24x EgonOlsen 22x Slyth2727 22x orangepascal 22x TritonDreyja 21x Spasi 21x orange451 20x DavidBVal 19x theagentd 16x KaiHH 12x LiquidNitrogen 11x Opiop 11x princec 11x Phased 10x CoDi^R 10x
 Making a Dynamic Plugin Systemby Hydroque2016-06-25 00:13:25Java Data structures2016-06-13 21:22:09Java Data structures2016-06-13 21:20:42FPS Camera Tutorialby Hydroque2016-05-22 05:40:58Website offering 3D Models specifically for games for freeby vusman2016-05-18 17:23:09Website offering 3D Models specifically for games for freeby vusman2016-05-09 08:50:56Website offering 3D Models specifically for games for freeby vusman2016-05-06 11:10:21Website offering 3D Models specifically for games for freeby vusman2016-04-29 12:56:17
 java-gaming.org is not responsible for the content posted by its members, including references to external websites, and other references that may or may not have a relation with our primarily gaming and game production oriented community. inquiries and complaints can be sent via email to the info‑account of the company managing the website of java‑gaming.org