How to accurately simulate motor reduction effect on inertia

asked 2017-08-27 01:17:22 -0500

Marcos Maximo gravatar image

I am researching humanoid robot walking and I am trying to use Gazebo to accurately simulate the DARwIn-OP robot in terms of dynamics. To have a accurate dynamic model, I am trying to accurately simulate the MX-28 servomotors dynamics. However, I have doubts about how I should include the motor reduction effect on inertia. I tried to simply add N^2 * Jmotor to the inertia of each driven link in the axis rotation direction. Despite not perfect, I have found that this is an approximation usually done in manipulator dynamics simulation. For a discussion about this, see: https://ntrs.nasa.gov/archive/nasa/ca...

However, since I do not know how exactly Gazebo simulates dynamics, I am not sure that I have implemented this correctly in Gazebo. Which is the correct way of implementing motor reduction effect on inertia in Gazebo?

edit retag flag offensive close merge delete

Comments

Hello , I am interested in that topic as well. Did you get any results?

val gravatar imageval ( 2020-07-23 05:38:33 -0500 )edit

Hi, unfortunately no. I am pretty sure today that the way I was simulating was wrong. Just adding N^2 * Jmotor to the inertia of each driven link has not the same effect has summing N^2 * Jmotor to each element in the diagonal of the inertia matrix. There is a gearbox implementation in Gazebo. I tried using it and I remember that the results for 2 links agreed with what was expected, but I had trouble getting correct results with 3 links. Still looking for an answer.

Marcos Maximo gravatar imageMarcos Maximo ( 2020-10-11 11:59:35 -0500 )edit