Hi!
I saw a statemment that it is difficult to compute to compute angular velocity from angular acceleration. Instead it is said to be better to compute ang. momentum by integrating torque and then compute ang. velocity from momentum.
Anybody knows why?
belive me it is not really dificult ,how to make it you will find here, the full explenation:
http://www.d6.com/users/checker/pdfs/gdmphys2.pdf
and all physics articles, very intresting if you want extand your physics knowledge::
http://www.d6.com/users/checker/dynamics.htm
in physics lcp and constraints arwe really dificult…
I’ve used Chris Decker’s articles to create the embedded physics engine in our particle system. I suspect the statement might have to do with the induced change in angular velocity due to collisions. One approach to collision response is to calculate an impulse force due to the collision. An impulse force is a simplification of the various phases a collision really goes through. The impulse force can be used to calculate the torque and then the change in angular velocity of the colliding objects.
Mike
This doen’t have to do with collisions, but with pure math I think.
While it is easy to calculate e.g. speed from acceleration, it seems to be difficult to calculate rot.speed from rot.acceleration. Why?
Transposed to the ‘translational’ world the statement claims its easier to calculate momentum (m*v) from force and later speed from momentum than speed from acceleration.
Can’t I just divide torque by intertia giving ang. acceleration and integrate that to ang.velocity? Thats what I do in translations …
Background: I’m looking for the basic operational dataset to send over the network for physics. One idea is to omit any physical data (mass, intertia) and just pass dynamics data (pos., speed, acc., orientation, ang.vel., ang.acc). The statement claimed that I need the inertia tensor to ang.vel. from torque and ang.momentum.
It’s been a while since I dug into the math, so this is hopefully close.
If you are referring to rotational movement about the center of mass of an object (a plane doing a barrel roll for example) then Newton’s 2nd Law of rotation comes in (X is cross product):
torque = r X F = I X a
Where I is the moment of inertia, and a is the angular acceleration. In the general case, the moment of intertia is a 3 x 3 matrix (a tensor). If the object is treated as uniformly dense along all axis (like a cube or sphere) then it can be a scalar value.
So I would guess that the author may have assumed the general case where one would need to take the cross product of the inverse of the moment of interia tensor and the torque to find the angular acceleration (which would be used to find the new angular velocity).
Since the angular momentum is:
L = I X w
Where L is the angular momentum and w is the angular velocity, and
torque = dL/dt
the new momentum would be:
L' = L + torque = I X w'
and the cross product of the inverse of the moment of inveria tensor would still need to be calculated to find the new angular velocity.
Even in the general case, I don’t understand the difference either.
Is this an online reference you can share?
Could it have to do with the representation of the velocity and acceleration? For example using Euler representation versus Quanternions?