[ODE] Degree of liberty of Slider joint
gcarlton at iinet.net.au
Thu Jun 15 17:41:57 MST 2006
I don't know of anybody implementing such a feature. If anybody did do
this, it would be a worthwhile addition to ODE.
It seems possible to allow the slider joint to only use the 3 or so
constraints to allow for a free rotation mode. A further mode could be
to take off more constraints and just have a 1-dimensional joint that
allows free movement along the other axis. That would be a very cheap
way of simulating a hoverbike, where its rotational stability is
attempted using other methods (e.g. an AMotor to try to keep it upright,
independant of its ground repusive force). Although this may need a new
joint type again, one that doesn't have an upper bound to allow a hover
to shoot up over a jump.
Extra slider params would need to switch between these modes.
Another idea would be to extend the ball joint type to allow for slider
params, in the same way as the wheel joint type does (called
universal2?? I forget) - that joint is essentially a rotational joint
type, but with the ability to relax the "slider" dimension. However,
this doesn't work exactly like a slider, as it always has effective
limits at 0, and the only thing you can do is play around with the
slider ERP/CFM. The real slider allows a "free range" to be specified.
Lastly, when I played around with the slider joint type I found it
substantially less stable than the slider of the wheel joint type. If
you search around for slider2 on this list you will see I made my own
alternative which acts like the wheel joint type effect, with the limits
always at 0. I found the behaviour of this alternative to be superior.
Remi Ricard wrote:
> I was looking at Slider Joint on the wiki page :
> and was thinking "Hey this joint can rotate around the sliding axis".
> Then I looked at code in joint.cpp and there is 5 constraints or 6 if
> fmax exists so the rotation around the sliding axis is not possible.
> So my questions are:
> Is this true ?
> If this is true maybe the figure in the wiki should have a rectangular
> shaft to explicitly state that there is no rotation around the sliding
> axis ?
> Do you know if someone implemented a Slider which has no constraint on
> rotation ?
> ODE mailing list
> ODE at q12.org
More information about the ODE