[ODE] On torque ..

Sergey Kurdakov k15c20 at zelmail.ru
Sat May 8 15:45:37 MST 2004


Hello,

so today promised hints not sure that they are better that yesterdays ;) but .

first as you stuck with rotations :) the suggestion is to discuss this just in few more details :)

so - rotation of different vectors.

read something related

for ex you might compute quaternion of rotation like it is done for example here
http://cal3d.sourceforge.net/docs/api/html/classCalQuaternion.html
see
     CAL3D_API CalQuaternion  shortestArc (const CalVector &from, const CalVector &to)

 Computes the shortest arc quaternion that will rotate one vector to another. This function finds the shortest arc quaternion. Based
on equations from "Game Programming Gems" - chapter 2.10


also see some code here - this will give 'immediate rotation matrix" but anyway it might who you more facts on rotation
http://www.acm.org/jgt/papers/MollerHughes99/

search maybe more google for rotate one vector to another :)
as there are no specially difficult  things here...

now when you have quaternion which describes desired rotation you might either 'animate' your body using slerp - but seems you
already dropped this idea but why not to combine simulation with 'hard written results' as anyway as a result of applied torque you
will get desired rotation??? so maybe if you give more hints why this approach ails then maybe others might give some hints...

if that actually will not be what you need then

 you might as I suggested - develop 3 -4 controllers for 3 or even 4 for quaternion components of the vector and keep track of
applied torques - momentums.

this might look somewhat 'unworkable' but practise shows- that if coeffs of feedback loop are chosen after several 'try and tune'
attempts  applied torque generated by feedback loop will   well fit into  producing desired rotations. though for  an estimate of
deisred torque  I think that you might use  things which were suggested by Jon Watte - compute torque using  integrals  or maybe
even to take that russian article and have the torque they suggest as basis for feedback loop. .. I'm not sure but :) you might try
babelfish to translate .... and finally if you really would need - I might translate an article for you as :) currently I have
plenty of time and think to spend it on some things which I like to do ...



please let know how things are progressing :)

Regards
Sergey Kurdakov





More information about the ODE mailing list