[ODE] hinge-2 and coordinate systems

[maddocks@metservice.com]

I never had any problems with the hinge2 joint. It was surprisingly easy 
to get working. Heres my set up code....


     bool Suspension::SetUp( graphics::Scene* scene, Body* body, dGeomID 
geomGroup)
     {
         assert( brake);
         assert( wheel);
         brake->SetUp( scene, geomGroup);
         wheel->SetUp( scene, geomGroup);

         // wheel hinge
         odeJoint = new dHinge2Joint( *scene->World(), 0);
         odeJoint->attach( *body->getBody(), *wheel->getBody());

         odeJoint->setAnchor( position.x, position.y - restLength, 
position.z); // hub position
         odeJoint->setAxis1( 0.0f, 1.0f, 0.0f);
         odeJoint->setAxis2( 1.0f, 0.0f, 0.0f);

         // set joint suspension
         // h = Simulation timestep    FIXME hardcoded
         Scalar h = 0.01;
         // k_p = Spring Constant
         // k_d = Spring Damping
         //
         // ERP = h k_p / (h k_p + k_d)
         // CFM = 1 / (h k_p + k_d)

         Scalar springERM = h * springConstant / ( h * springConstant + 
springDamper);
         Scalar springCFM = 1 / ( h * springConstant + springDamper);

         odeJoint->setParam( dParamSuspensionERP, springERM);
         odeJoint->setParam( dParamSuspensionCFM, springCFM);

         if( !steering)
         {
             odeJoint->setParam( dParamLoStop, 0);
             odeJoint->setParam( dParamHiStop, 0);
         }

         return true;
     }


This function is called after the car is oriented  and the suspension 
position is in world coordinates but from memory I don't think it made 
any difference.
I just ignored the coordinate system and assumed everything was as per 
opengl (Z = up axis). One thing that could be a problem is your body may 
be too heavy for your suspension springs.

Any questions, email me.


Henry

On Thursday, July 25, 2002, at 09:09  AM, Timothy J. Wood wrote:

>
>
>   I'm new to ODE and I'm having trouble with joints.  I'm building a 
> basic 4-wheel car from a chassis block and 4 wheels (drawn as 
> cylinders, but simulated as spheres).
>
>   My code is based fairly closely off the buggy test, but when I use 
> hinge-2 joints, the chassis just falls to the ground.  When I use fixed 
> joints (just for debugging), the wheels stay attached, but they quickly 
> spin 90 degress CCW about the up vector.  This is maybe expected since 
> when I create my wheels, I rotate them and fixed joints apparently 
> don't support non-identity relative orientations.
>
>   One difference between my code and the buggy code is that I'm using 
> the standard OpenGL coordinate system instead of the coordinate system 
> that ODE seems to use (with Z being up instead of Y).  Both are 
> right-handed, so I would assume all should be OK.  The other difference 
> is that I'm not currently using a geometry group since that should just 
> optimize stuff, not actually change the results in this case.
>
>   I've positioned and oriented my objects before attaching them.  Are 
> there any other obvious things that would cause hinge-2 joints to 
> completely fail to keep the anchor points together?  A search on the 
> mailing list archive shows only that others are able to make hinge-2 
> work, not any failure modes of this variety that I could find.
>
>   Thanks!
>
> -tim
>
> _______________________________________________
> ODE mailing list
> ODE@q12.org
> http://q12.org/mailman/listinfo/ode
>
>



"Two is not equal to three, even for large values of two"