Doubled Centrifugal Term of Joint Reaction Forces

Official Python bindings with a focus on reinforcement learning and robotics.
Post Reply
Posts: 1
Joined: Sun Nov 21, 2021 2:47 pm

Doubled Centrifugal Term of Joint Reaction Forces

Post by wlsrbs82 »


Firstly, thanks for your PyBullet engine. I am happy to use your engine for my works.

For using PyBullet as a mechanical system design tool, I am on the verification phase currently.
I tested 1 DOF rotating bar with some mass for very simple case study.
I observed that centrifugal force from PyBullet is twice result from commercial CAE software.
(HyperWorks MotionSolve and Autodesk Inventor Built-in Dynamic Simulation)
More clearly, my setting is like this, for example.
mass = 1kg, COM from center of rotation = 2m, constant rotating speed = 1 rad/s
Although 2N is expected from centrifugal force, I got the 4 N as a related reaction force from getJointState of PyBullet.
For another very complex case of 6DOF, most of the results was quite similar to those from commercial software.
But centrifugal force case has definitely double value when PyBullet is used.

My system is Windows 10 and is currently up to date with pybullet-3.1.7. I have attached the urdf and motion input in question.
For my simulation, I used POSITION CONTROL and zero dampings.
Could you give some advices of what might be wrong? or is there any isssue of reaction force calculation?
1 DOF System URDF
(1.11 KiB) Downloaded 24 times
1DOF GDT.csv
Filtered Trapezoidal Velocity Profile
(94.67 KiB) Downloaded 20 times
Post Reply