-7

The idea is simple, it is to calculate the velocity of two objects after constraints.

Constraint code:

const VECTOR3<T> vRadiusObject1 = param.vObj1ContactPoint - param.Obj1.vBarycenter;
const VECTOR3<T> vRadiusObject2 = param.vObj2ContactPoint - param.Obj2.vBarycenter;

const VECTOR3<T> vRelativeVelocity1 = param.Obj1.vVelocity + cross(param.Obj1.vAngularVelocity, vRadiusObject1);
const VECTOR3<T> vRelativeVelocity2 = param.Obj2.vVelocity + cross(param.Obj2.vAngularVelocity, vRadiusObject2);

T fImpulse = (-dot(param.vDirection, vRelativeVelocity1 - vRelativeVelocity2) * (1 + param.RestitutionCoefficient)) / (param.Obj1.fInverseMass + param.Obj2.fInverseMass +
    dot(cross(cross(vRadiusObject1, param.vDirection), vRadiusObject1) * param.Obj1.mtInverseInertia +
        cross(cross(vRadiusObject2, param.vDirection), vRadiusObject2) * param.Obj2.mtInverseInertia, param.vDirection ));

param.pOutVelocity1->vVelocity = param.Obj1.vVelocity + (param.vDirection * fImpulse * param.Obj1.fInverseMass);                    
param.pOutVelocity1->vAngularVelocity = param.Obj1.vAngularVelocity + (cross(vRadiusObject1, param.vDirection) * fImpulse * param.Obj1.mtInverseInertia);

param.pOutVelocity2->vVelocity = param.Obj2.vVelocity - (param.vDirection * fImpulse * param.Obj2.fInverseMass);                                        
param.pOutVelocity2->vAngularVelocity = param.Obj2.vAngularVelocity - (cross(vRadiusObject2, param.vDirection) * fImpulse * param.Obj2.mtInverseInertia);

return fImpulse;

Use the formula: constraint formula

But it was found that the calculated point velocities of the two objects were not the same.

Calculate Point Velocity Code:

const VECTOR3<T> vRadiusObject = vPoint - vCenter;
return motion.vVelocity + cross(motion.vAngularVelocity, vRadiusObject);

The Test Data:

float fMass1 = 100;
float fMass2 = 230;
const VECTOR3<float> vCenter[] = { VECTOR3<float>(0, 9, 0), VECTOR3<float>(0, 0, 0) };
const VECTOR3<float> ContactPoint[] = { VECTOR3<float>(-3, 4.5, -3), VECTOR3<float>(-3, 4.5, 3), VECTOR3<float>(3, 4.5, -3), VECTOR3<float>(3, 4.5, 3) };
const VECTOR3<float> vSize = VECTOR3<float>(6, 9, 6);

Motion<> motion1, motion2;

MATRIX44<float> mtOBB1 = MATRIX44<float>::Identity;
MATRIX44<float> mtOBB2 = MATRIX44<float>::Identity;

motion1.vVelocity = VECTOR3<float>(0, 200, 0);
motion1.vAngularVelocity = VECTOR3<float>(0, 2, 0);

motion2.vVelocity = VECTOR3<float>(0, 100, 0);
motion2.vAngularVelocity = VECTOR3<float>(10, 0, 0);

I calculated "ContactPoint[0]"(-3, 4.5, -3) The velocity of the constraint point after constraint calculation：

OBJECT1 ContactPoint[0] ： {x=3.92508936 y=160.614685 z=46.1242218}

OBJECT2 ContactPoint[0] ： {x=8.37378883 y=138.909134 z=40.2669983}

But according to the derivation of the formula they should be equal. WHY????

Here is my test project,full source code: https://github.com/noodlerun/impulse.git

I expect that the relative velocities of the constraint points of the two objects should be equal.

noodle_run
  • 1
  • 1
  • 1
    See [What is a debugger and how can it help me diagnose problems?](https://stackoverflow.com/questions/25385173/what-is-a-debugger-and-how-can-it-help-me-diagnose-problems) and [How to debug small programs](https://ericlippert.com/2014/03/05/how-to-debug-small-programs/). Also remember to include a [mre] and put *all* relevant details *in* the question, *not* behind external links (or as images). This post *may* also be relevant: [Is floating point math broken?](https://stackoverflow.com/questions/588004/is-floating-point-math-broken) – Jesper Juhl Sep 02 '23 at 13:30
  • Can you pick out the few lines of code that actually show the problem, no one here is going to spent time going over a code base that large. – Pepijn Kramer Sep 02 '23 at 13:43
  • 1
    Please make your question self-contained. People should be able to understand what you're asking without visiting external sites. [Why should I not upload images of code/data/errors?](https://meta.stackoverflow.com/questions/285551/why-should-i-not-upload-images-of-code-data-errors) – Evg Sep 02 '23 at 14:14

0 Answers0