Physical accuracy of a simulation using bullet physics.

Post Reply
dhruvesh.patel
Posts: 2
Joined: Fri Jul 31, 2015 8:11 am

Physical accuracy of a simulation using bullet physics.

Post by dhruvesh.patel »

I need to simulate a hanging rope for my project but I need the simulation to be physically correct. Can I use bullet to do this ? Has there been any study which matches bullets simulation results with actual experiments ?

Thank you.
d3x0r
Posts: 51
Joined: Tue Dec 11, 2012 9:59 pm

Re: Physical accuracy of a simulation using bullet physics.

Post by d3x0r »

From reviews I searched through it looked like bullet was fairly accurate for simulation results.
A rope though... depends on exactly how real you want it. If it's a rope of fair length suspending an object, the rope will untwist; there's nothing native that's 'rope'. But I'd imagine it could be simulated using a lot of line segments attached with a bt6DofConstraint... which will allow a slight stretching ability, but will be unconstrained in rotation... so if you had a hanging spinning object it would spin forever. I'd imagine one could add additional constraint information that resembled a spring in the twist direction...
Basroil
Posts: 463
Joined: Fri Nov 30, 2012 4:50 am

Re: Physical accuracy of a simulation using bullet physics.

Post by Basroil »

Rope is one of those softbody problems that doesn't really play nicely with rigid body physics. If you only care about the statics or have a heavy rope (frequencies produced by the simple wave equation for the length of the rope are much smaller than the sampling time of your simulation), you will get fairly good results with the above joint chain (though you should consider the multibody approach using featherstone to improve performance). If you want the exact dynamics including mode shapes, it's far better to use the wave equations or an implicit FEM type package for the solving stage.

For now, try the joint chain at high framerates to see if that works for you.
Post Reply