1

There is a game in which i have to program the AI and compete with other programers. I am playing that game. It consists in a race, your pod have to pass through some checkpoints in order.

You have some input arguments to use in your algorithm:

nextCheckpoint (x,y) -> nX,nY
lastCheckpoint (x,y) -> lnX,lnY
currentposition (x,y) -> x,y

and the output of the code is the directión and the thrust.

int pod::errorX(){
    int R[2];
    int X1;
    float b;
    int DeltaX;
    //Vector of the line between Checkpoints
    R[0] = nX-lnX;
    R[1] = nY-lnY;
    //calculation of the distance in the X axis between the ship and the line
    b =(R[0]*(lnX-x) + R[1]*(lnY-y))/((-pow(R[0],2))-(pow(R[1],2)));
    X1 = lnY + b*R[0];
    DeltaX = x-X1;
    cerr<<"Dx: "<<DeltaX<<endl;
    return DeltaX;

}

I also have the equivalent pod::errorY()

finally:

DirectionX = nX-errorX();
DirectionY = nY-errorY();

Here some pictures of what I am doing:

the theory of what should happen

I want to do this becouse most of the times the inertia make my ship orbitate the checkpoints, this consume a lot of time and makes me loose the race.

Well now let's face the problem, it works sometimes and others is not even close.

When the inputs are small. like 

nX= 2;
nY= 2;
lnX=0;
lnY=0;
x=0;
y=0;

The output is:

directionX = 3
directionY = 1

But when they are bigger:

nX= 6553;
nY= 7817;
lnX=13044;
lnY=1908;
x=13921;
y=3530;

the output:

DirectionX=6940;
DirectionY=1907;

I do not know what happen. Maybe becouse b is decimal? is there any error that i havent seen?

Maybe this code is more confortable to test becouse you dont need the game:

#include <iostream>
#include <string>
#include <cmath>
using namespace std;

int main()
{

    int lnX = 13044;
    int lnY = 7817;
    int nX = 13044;
    int nY = 1908;

    int x = 13921;
    int y = 3530;

    int R[2]={};

    //Vector de la recta entre Checkpoints
    R[0] = nX-lnX;
    R[1] = nY-lnY;
    //puro algebra, para calcular el desfase de X
    int X1,DeltaX;
    long double b;
    b =(R[0]*(lnX-x) + R[1]*(lnY-y))/((-pow(R[0],2))-(pow(R[1],2)));
    X1 = lnY + b*R[0];
    DeltaX = x-X1;
    //desfase de y;
    int Y1,DeltaY;
    Y1 = lnY+b*R[1];
    DeltaY = y-Y1;

    //dirección
    int DirectionX;
    int DirectionY;

    DirectionX = nX-DeltaX;
    DirectionY = nY-DeltaY;
    cout<<X1<<" "<<Y1<<endl;
    cout<<DeltaX<<" "<<DeltaY<<endl;
    cout<<b<<" "<<DirectionX<< " "<<DirectionY<<endl;
Jake
  • 4,829
  • 2
  • 33
  • 44
Gatoninja
  • 63
  • 4
  • you are welcome, but to be honest, after reading half i skipped the rest to see a question at the end. unfortunately I didnt find any ;). It is not quite clear what is your question – 463035818_is_not_an_ai Oct 13 '16 at 14:22
  • related/dupe: http://stackoverflow.com/questions/588004/is-floating-point-math-broken – NathanOliver Oct 13 '16 at 14:22
  • Looks like you could be doing integer math when you should be doing floating math, and then truncating or rounding if necessary. – AndyG Oct 13 '16 at 14:24

1 Answers1

1

Is this line right?: X1 = lnY + b*R[0]; Or should it be X1 = lnX + b*R[0];

JMA
  • 494
  • 2
  • 13