Car Physics Problem

Game Mechanics
#1

Hello. I have started developing a racing game, so I started by implementing the car physics by the guide of “Marco Monster”, as well as a little “debug screen”. It is partly german though, so here a little translation:
“Geschwindigkeit”=“Speed”
“LW-Kraft”=“Air resistance force”
“RW-Kraft”=“Rolling resistance force”
“Beschleunigung”=“Acceleration”

Anyways, I’m having a few problems here. First of all, I’m tracking the wheel speed separately, this works pretty well for high speeds, but at low speeds it just spazzes out. Also, if you’re not accelerating, the wheels decelerate wayyy to fast. The third problem is that the maximum speed is about 180 km/h (or 112mph) in 4th gear, because in 5th and 6th gear, the car continues to decelerate (This is pretty logical, considering the engine torque goes down, because of that the driving force goes down, so the air resistance is greater than the driving force, which decelerates the car, but that shouldn’t happen, correct?)

Here is the most important part of the source:

public void update(float delta, boolean accelerating) {

	float speed = v.length();

	Main.renderStringFormatted(29, "%d", (int) rpm);
	Main.renderStringFormatted(30, "%d", gear + 1);
	Main.renderStringFormatted(0, "Geschwindigkeit: %.1fkm/h", speed * 3.6);

	rpm = wheelSpeed * vehicle.getGearRatio(gear) * 30 / Constants.PI;
	if (rpm < vehicle.getMinimumRPM())
		rpm = vehicle.getMinimumRPM();
	if (rpm > vehicle.getMaximumRPM())
		if(shiftGearsUp())
			rpm = wheelSpeed * vehicle.getGearRatio(gear) * 30 / Constants.PI;
		else
			rpm = vehicle.getMaximumRPM();

	float Fdrive = vehicle.getEngineTorque(gear, rpm)
			/ vehicle.getTireDiameter();
	Vector2f Ftraction = accelerating ? dir.mul(Fdrive) : new Vector2f();
	Vector2f Fdrag = v.mul(-vehicle.getShapeDrag() * speed);
	Vector2f Frr = v.mul(-vehicle.getShapeDrag() * 30);

	Vector2f Flong = Ftraction.add(Fdrag).add(Frr);

	Main.renderStringFormatted(1, "LW-Kraft: %.1fN",
			Fdrag.lengthSigned(dir));
	Main.renderStringFormatted(2, "RW-Kraft: %.1fN", Frr.lengthSigned(dir));

	float Wr = vehicle.getBackTireLoad(Flong.lengthSigned(dir)
			/ vehicle.getMass())
			/ (vehicle.getMass() * 9.81f);
	Main.renderStringFormatted(5, "Load V: %.2f%s", Wr, "%");

	Flong = Ftraction.mul(Wr);

	Vector2f Ffriction = new Vector2f();

	if (speed > 0) {
		float slip = (wheelSpeed * vehicle.getTireDiameter() - speed)
				/ speed;
		Main.renderStringFormatted(4, "Slip: %.2f", slip);

		Ffriction = Ftraction.mul(Math.min(16.6666667f * slip, 1));
	}

	Flong = Flong.add(Frr);
	float Tfriction = Ffriction.lengthSigned(dir) * vehicle.getTireDiameter();
	float Tdrive = Flong.lengthSigned(dir) * vehicle.getTireDiameter();
	float Ttotal = Tdrive - Tfriction;
	float aa = Ttotal / vehicle.getRearTireInertia();
	wheelSpeed += aa * delta;
	Main.renderStringFormatted(6, "Tfriction: %.2fNm", Tfriction);
	Main.renderStringFormatted(7, "Tdrive: %.2fNm", Tdrive);
	Main.renderStringFormatted(8, "Ttotal: %.2fNm", Ttotal);
	Main.renderStringFormatted(9, "Angular acc.: %.2frad/s^2", aa);
	Main.renderStringFormatted(10, "Wheel vel.: %.2fkm/h", wheelSpeed * vehicle.getTireDiameter() * 3.6);
	
	Flong = Flong.add(Fdrag);
	Vector2f a = Flong.div(vehicle.getMass());

	Main.renderStringFormatted(3, "Beschl.: %.2fm/s^2", a.lengthSigned(dir));

	v = v.add(a.mul(delta));
	if (dir.dot(v) < 0)
		v = new Vector2f();

	pos = pos.add(v.mul(delta));
}

And here is a SSCE: download (Source code/debug scripts/LWJGL/runnable jar included, start the debug.bat for your system)

Thanks for your time.

#2

I want you to take a week off from programming this and come back to it. When you come back, make note of how little sense this code makes to you. From this lesson, realize what you can change about your code so that it is more readable after taking a break. Now imagine never having seen the code and this is what you’re presented with. I see why this has almost 100 views and no input.

So, I took the red pill and decided to see what this was all about. Some things impressed me - some things scared me.

Anyway, whenever there is weird behavior in a program, I immediately try to find out where all of the pertinent divisions are. Why? Because divisions are difficult and error prone. I do not know much about your domain here. I do however believe that when you getBackTireLoad, it should never be passed a value like this: -6.943892E-4. Let’s just look at how that code is done since you didn’t include it:


public float getBackTireLoad(float a) 
{
    return 4.905f * mass + (shape.getCenterOfMassHeight() / shape.getAxleDistance()) * mass * a;
}

What in the world is 4.905? Where did you get it? What does it mean? When we explore the Car Physics page you presented, it can be located as: a = 7350 N / 1500 kg = 4.9 m/s2 (=0.5 G) but why not just put it in a comment so I know what’s going on?

So, to get the Wr we do this:

4.905f * mass + (shape.getCenterOfMassHeight() / shape.getAxleDistance()) * mass * (Flong.lengthSigned(dir) / vehicle.getMass() / (vehicle.getMass() * 9.81f))

(I just combined the formulas)

You need to do error checking on your division results. I’m going to keep looking at this and toying around with it to see how it ticks but I suggest that if you want to move forward for now, you start there.