Thread: physics

i'm working on vehicle physics in OpenGL, and all seems well, except for one problem. the glTranslatef(); function dosn't seem to be working. at all. either that, or theres somthing wrong with the physics code itself which is very, very possible.

vehicle.cpp (most of it. dosn't show includes and variable declarations):

Code:

void vphys::setup(float engine, float mass, float pos[2], float sdir[2])
{
	vd.epower = engine;
	vd.m = mass;
	//vd.p[2] = pos[2];
	vd.dir[2] = sdir[2];
}

void vphys::mainroutine()
{
	//traction
	vd.tractionF[0] = vd.dir[0]*vd.epower;
	vd.tractionF[1] = vd.dir[1]*vd.epower;

	//drag
	vd.speed = sqrt(vd.v[0]*vd.v[0]+vd.v[1]*vd.v[1]);
	vd.dragF[0] = -dragC*vd.v[0]*vd.speed;
	vd.dragF[1] = -dragC*vd.v[1]*vd.speed;

	//roll resistance
	vd.rrF[0] = -rrC*vd.v[0];
	vd.rrF[1] = -rrC*vd.v[1];

	//total longtitudinal force
	float longF[2];
	longF[0] = vd.tractionF[0]+vd.dragF[0]+vd.rrF[0];
	longF[1] = vd.tractionF[1]+vd.dragF[1]+vd.rrF[1];

	//acceleration
	vd.a[0] = longF[0]/vd.m;
	vd.a[1] = longF[1]/vd.m;

	//velocity
	vd.v[0] = vd.v[0]+dt*vd.a[0];
	vd.v[1] = vd.v[1]+dt*vd.a[1];

	//final position
	vd.p[0] = vd.p[0]+dt*vd.v[0];
	vd.p[1] = vd.p[1]+dt*vd.v[1];
	glTranslatef(0, 0, vd.p[1]);
}

from my render(); function:

Code:

	glScalef(0.01, 0.01, 0.01);
	_peVphys->setup(900, 1000, startposition, startdirection);
	_peVphys->mainroutine();
	pModel->draw();

I really have no idea whats wrong. Although I havn't coded in 3.5 weeks, so i'm a little rusty. (more than I was to begin with :P).

-psychopath

Observations:

You have the following code:

Code:

	//vd.p[2] = pos[2];
	vd.dir[2] = sdir[2];

You probably want a memcpy here instead of assignment. Even if you did want just an assignment, you are assigning element #2 out of a 2 element array. Remember, zero based. So you are walking off the end of that array, bad news.

You may want to create your own Vector2D class. Then you can overload some useful math operators and have them behave like built in types. That would be much clearer to read.

Also when you call translate you use the 3rd parameter which is for Z translation usually, are you sure you want that?

Maybe more errors but I just glanced at it, I'll take another look later when I have more time. Good luck.

>>You probably want a memcpy here instead of assignment. Even if you did want just an assignment, you are assigning element #2 out of a 2 element array. Remember, zero based. So you are walking off the end of that array, bad news.<<

That crossed my mind, but I wasn't sure, so I left it like it was.

>>You may want to create your own Vector2D class. Then you can overload some useful math operators and have them behave like built in types. That would be much clearer to read.<<
That crossed my mind as well, but again, I wasn't sure, so stuck with arrays for the time being.

>>Also when you call translate you use the 3rd parameter which is for Z translation usually, are you sure you want that?<<
yup. Y is up in my engine, and you start looking down the negative Z.

Anyway thanx! I'll get started on the fixes asap and i'll post if I have any other questions.

-psychopath

arg! still nothing. heres the revised code:

vehicle.cpp

Code:

void vphys::setup(float engine, float mass, float pos[2], float sdir[2])
{
	vd.epower = engine;
	vd.m = mass*gravity;
	vd.p.x = pos[0]; vd.p.z = pos[1];
	vd.dir.x = sdir[0]; vd.dir.z = sdir[1];
}

void vphys::mainroutine()
{
	//traction
	vd.tractionF.x = vd.dir.x*vd.epower;
	vd.tractionF.z = vd.dir.z*vd.epower;

	//air-drag
	vd.speed = sqrt(vd.v.x*vd.v.x + vd.v.z*vd.v.z);
	vd.dragF.x = -dragC*vd.v.x*vd.speed;
	vd.dragF.z = -dragC*vd.v.x*vd.speed;

	//rolling resistance
	vd.rrF.x = -rrC*vd.v.x;
	vd.rrF.z = -rrC*vd.v.z;

	//total longtitudinal force
	vd.longF.x = vd.tractionF.x + vd.dragF.x + vd.rrF.x;
	vd.longF.z = vd.tractionF.z + vd.dragF.z + vd.rrF.z;

	//acceleration
	vd.a.x = vd.longF.x/vd.m;
	vd.a.z = vd.longF.z/vd.m;

	//velocity
	vd.v.x += dt*vd.a.x;
	vd.v.z += dt*vd.a.z;

	//position
	vd.p.x += dt*vd.v.x;
	vd.p.z += dt*vd.v.z;

	glTranslatef(0, 0, vd.p.z);
}

IMO everything seems fine. i'm completely stumped. any ideas?

-psychopath

To be honest, I'd really have to see more of the program to see what is really going on. If you want to attach a zip of all the source I would be happy to look at it later tonight when I get home. If you don't feel comfortable with that, that's fine too. I don't see anything jumping out at me with the code you posted. What you might try, is commenting out the complicated stuff (drag, traction) and just get the bare bones positional updates working first. Step through in a debugger, make sure there aren't any zero valued variables throwing off your calculations. That's where I would start.

i finally got it to work, but only by accident. I was trying to translate a vehicle scaled to 0.02 into a different position (not via the physics code), and that didn't work either. So, I tryed translating it by a larger number, and it moved. From that I found that I had to divide the about I wanted to move it by the scale value. (In that case, 0.5/0.02=25), so then i tryed multiplying my physics final position by a very large number, and it began to move, but slower than it should have. Then eventually I was led to a variable (dt), which I had set to 0.01. I changed it to 1.0 and it worked perfectly at normal speed.

Thanx anyway though, and thanks again for the help before, as it was working even less before then .

-psychopath