[ODE] Composite object vs. tri-mesh question

[Brett Hemes]

Hello,

I am learning ODE for use in robotics simulations for research and  
have a couple of questions regarding geoms and collisions.  I have to  
represent a trapezoidal prism-shaped mobile robot that can rotate on  
all three of its axes.  I recall reading somewhere that dynamic tri- 
meshs are less desirable than composite objects made from  
primitives.  If I try and build a nice trapezoidal prism from boxes  
however, I have geometries that share surfaces that commonly collide  
with other geoms (another thing I read is bad).  Is there any  
generally accepted answer for this problem?  I am currently using the  
composite object approach and have yet to see any stability problems  
in version 0.9 with double precision but I would like to know if this  
is a hack.

Additionally, is it common/possible to have geoms in different spaces  
attached to the same body where one space is used for collision  
detection and the other for drawing the output?

Thanks,
Brett




-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://ode.org/pipermail/ode/attachments/20071114/f1c34bb7/attachment.htm