[ODE] Re: I have buckling joints with 4 wheeled vehicle

[Denes Nagymathe]

Hi,

I faced the same problem when i played around with ODE. The problem is that 
ODE measures the hinge angles in the current time frame (the start of the 
step), but when the wheels are spinning fast enough, the resulting 
correction torque points to a wrong direction (because it's based on the 
angles at the start of the frame, but the wheel has already turned away).
 My solution was to 'hack' (or let's call it 'improve' :) the hinge2 joint 
to extrapolate the wheel orientation, and use this 'future' orientation 
instead of the current to measure hinge angles. It looks like this:

...
//  HINGE2_GET_AXIS_INFO (q,s,c);
//by Denes: use _extrapolated_ orientation
  dVector3 ax1,ax2;
 {
  dReal b2q[4], dq[4];
  dMatrix3 b2R;
  dWtoDQ (joint->node[1].body->avel,joint->node[1].body->q,dq);
  for (int j=0; j<4; j++) b2q[j] = joint->node[1].body->q[j] + 
0.5*dq[j]/info->fps;
  dNormalize4 (b2q);
  dQtoR (b2q,b2R);
  dMULTIPLY0_331 (ax1,joint->node[0].body->R,joint->axis1);
  dMULTIPLY0_331 (ax2,b2R,joint->axis2);
  dCROSS (q,=,ax1,ax2);
  s = dSqrt (q[0]*q[0] + q[1]*q[1] + q[2]*q[2]);
  c = dDOT (ax1,ax2);
 }
...

(The original HINGE2_GET_AXIS_INFO call in hinge2GetInfo2() is commented 
out, and my code is inserted.)
Using a large ERP also helps a bit, so using the two together gave the best 
results for me.

Denes