Open fmauch opened 10 months ago
What about simulated robots? Would it be possible to generate more precise inertia values from CAD files?
Well yes, this is about robot simulation. The robots contain config files as e.g. the following for a UR5e:
[Link]
mass = [3.761, 8.058, 2.846, 1.37, 1.3, 0.365]
center_of_mass = [[0, -0.02561, 0.00193], [0.2125, 0, 0.11336], [0.15, 0.0, 0.0265], [0, -0.0018, 0.01634], [0, 0.0018, 0.01634], [0, 0, -0.001159]]
inertia_matrix = [[0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0.0002]]
[Geometry]
radius_wrist3 = 0.0375
radius_lower_arm = 0.0375
offset_x3_lower_arm = 0.0066
Those values are also referenced on a UR Forum post: https://forum.universal-robots.com/t/moment-of-inertia-for-newton-euler-dynamic/14844/2.
I'm a bit confused. Where do those values come from? Is that something stored in the robot controller?
I'm a bit confused. Where do those values come from? Is that something stored in the robot controller?
yes. It is part of /root/.urcontrol/urcontrol.conf.UR5
. On URSim you will find it in /ursim/.urcontrol/urcontrol.conf.UR5
Currently, we use a cylinder approximation for the inertia. However, we can read proper inertia values from the robot. We should probably use this on the description.