Different result when I using different physics engine to simulate a rotating cube imposed by a torque

Typically: "How do I... ", "How can I... " questions
Post Reply
21207_iHome
Posts: 8
Joined: 25 Apr 2017, 03:13

Different result when I using different physics engine to simulate a rotating cube imposed by a torque

Post by 21207_iHome » 10 Nov 2017, 04:03

Hi everyone,
I want to simulate an object, whose mass is 1Kg and the diagonal elements of inertia tensor is {2, 1, 1}, imposed by a torque T={1, 1, 1} using the function simAddForceAndTorque. Then I wish to retrieve the angular velocities described in world frame. However, I found that different physics engines gave me different results. For example, the results generated by bullet 2.7 and newton are same. On another hand , bullet 2.8 and ODE are identical.
bullet 2.7 and newton: Image https://d.pr/FREE/VeB2c7
bullet 2.8 and ODE: Image https://d.pr/FREE/lpRNck
scene file: https://d.pr/FREE/pE1bz2
I am confused with the problem. Which one is correct?

coppelia
Site Admin
Posts: 6218
Joined: 14 Dec 2012, 00:25

Re: Different result when I using different physics engine to simulate a rotating cube imposed by a torque

Post by coppelia » 13 Nov 2017, 22:17

Hello,

a physics engine makes many approximations in order to be able to run at competing speeds. Some approximations do not influence results of simulation A, but results of simulation B, others will just do the opposite. For this reason there is not just one physics engine in V-REP, so that the user can make a more informed decision on how realistic results are (which does not mean that if 3 engines provide the same results, that they are right!).
Physics engines are known to have problems with rapidly rotating bodies in general. In case of doubt, I tend to believe the Vortex engine more than the others, but that's a personal opinion based on experience.

Cheers

Post Reply