PVector addition resulting in a NAN result ... can't figure out why.

Programming Questions

almost_sc..

PVector addition resulting in a NAN result ... can't figure out why.

in Programming Questions • 2 years ago

Hi Folks,

I've been trying to figure this out for a few hours and have hit a wall.

I'm trying to update drawing position based on some data from a 6DOM accelerometer/gyro combo.

I've got some code that combines the accelerometer and gyro data (using a Kalman filter) into a the best estimated acceleration vector.

I'm then simply trying to add that vector to a velocity vector and then add that to a position vector.

But when I try to add the incoming acceleration vector to the velocity vector I get NaNs across the board.

Here is the relevant code:

float [] RwEst = new float[3];
//RwEst is the acceleration vector produced by the kalman filter
PVector location;
PVector velocity;
PVector accelertaion;
void setup()
{
size(VIEW_SIZE_X, VIEW_SIZE_Y, P3D);
smooth();
location = new PVector(width/2, height/2,0);
velocity = new PVector(0.0, 0,0);
accelertaion = new PVector(0.0, 0.0,0);
}

void draw() {

getInclination();

update();

ellipse(location.x, location.y, 10, 10);

}

void update() {
//RwEst is the acceleration vector produced by the kalman filter
println("RwEst= "+RwEst[0]+" "+RwEst[1]+" "+RwEst[2]+" " );
accelertaion.set(RwEst);
println("acc= "+accelertaion);
velocity.add(accelertaion);
println("velo= "+velocity);
location.add(velocity);
println("location= "+location);
println(" " );
}

Here is a sample of the output:

RwEst= -0.80083185 0.2533032 -9.964663

acc= [ -0.80083185, 0.2533032, -9.964663 ]

velo= [ NaN, NaN, NaN ]

location= [ NaN, NaN, NaN ]

RwEst= -0.8005298 0.25594395 -9.96462

acc= [ -0.8005298, 0.25594395, -9.96462 ]

velo= [ NaN, NaN, NaN ]

location= [ NaN, NaN, NaN ]

Any thoughts would be great.

Thanks!

Replies(2)

thomas.di..

Re: PVector addition resulting in a NAN result ... can't figure out why.

2 years ago

i think the cause for the NaN result (Not a Number) is somewhere else in your code.
for useful debugging, you should print the values of "velo" before you add the velocity.
i think (but i could be wrong, cause i dont know the rest of the code) that "velo" is used somewhere else before adding the accerleration and already has invalid values then, when it comes to adding the acceleration.

the following shows, how NaN can be produced:

PVector p = new PVector(0, 0, 0);
PVector p_dir = new PVector(1.0, 1.0, 1.0);
// the following causes the NaN results
p.mult(1.0/0.0); //or try: p.div(0.0)
p.add(p_dir);
println(p);

here's another useful page:
http://www.concentric.net/~ttwang/tech/javafloat.htm

almost_sc..

Re: PVector addition resulting in a NAN result ... can't figure out why.

2 years ago

Thanks,

I found the problem. The first time my acceleration data comes in it has a nan in it. I put a conditional to skip those and all is well now. Thanks for your post, it pointed me in the right direction.

Top Reply