[ODE] bad wheels

[Jon Watte]

> My vehicle is a box with 4 spheres as wheels (hinge2-jointed), it's
> driving on a trimesh. My problem is steering. Sometimes the left wheel
> locks steering full left, sometimes the wheels steer in different
> direction, also problem increases when I drive faster. 

Do you set the constraints twice? If not, you have to, because of an 
ordering problem (ODE won't let lostop and histop ever conflict). 
I submitted a patch to fix this, but it seemed to have been dropped 
on the floor.

Add the lines with "+":

> if(m_isSteering == false) 
> { 
>   m_steer = 0.0f; 
>   dJointSetHinge2Param (wheel1id, dParamLoStop, 0); 
>   dJointSetHinge2Param (wheel1id, dParamHiStop, 0); 
+   dJointSetHinge2Param (wheel1id, dParamLoStop, 0); 
>   dJointSetHinge2Param (wheel2id, dParamLoStop, 0); 
>   dJointSetHinge2Param (wheel2id, dParamHiStop, 0); 
+   dJointSetHinge2Param (wheel2id, dParamLoStop, 0); 
> } 


>   dJointSetHinge2Param (wheel1id, dParamVel, v); 
>   dJointSetHinge2Param (wheel1id, dParamFMax, 0.2f); 
>   dJointSetHinge2Param (wheel1id, dParamLoStop, -0.5f); 
>   dJointSetHinge2Param (wheel1id, dParamHiStop, 0.5f); 
>   dJointSetHinge2Param (wheel1id, dParamFudgeFactor, 0.1f); 


> Shouldn't this work? 
> Anyone who got any better steering-"algoritm" or suggestion?


That's the wrong way of doing it. You want to set LoStop and 
HiStop to be the same, and to be the angle at which you want to 
turn. Further, a force of 0.2 is usually quite small -- for a 
body of mass 1, that'll only accelerate it to 0.2 units per 
second per second. (But, as I said, you don't want to be doing 
this anyway).

There's some steering code in the Carworld sample, which you 
can likely crib. It also does the thing where the inner wheel 
turns more than the outer wheel, which is crucial.

http://www.mindcontrol.org/~hplus/carworld/


Cheers,

			/ h+