Gazebo / ROS problem in multiple Monte Carlo simulations

asked 2020-06-25 07:16:00 -0600

this post is marked as community wiki

This post is a wiki. Anyone with karma >75 is welcome to improve it.

Hello everyone, I'm trying to test an obstacle avoidance method for multiple robots using Monte Carlo simulations, I setup a mechanism to repeat the simulations by setting the states of the models (Turtlebot3s and some boxes) for each test in Gazebo. Here is my issue , when the robots start to move, there is some kind of drifting or slipping (definitely not from the controller, because sometimes it slopes to the sides and it is not possible for the Turtlebot), plus when I ran multiple simple tests, I noticed that the robots at start points and even during the travel lean to the sides randomly (left side or right side), and obviously the result would be they travel different paths in fixed time. (something like the Turtlebots slipping when there is no input inside Gazebo) maybe you can notice the problem from figures below:

I controlled everything multiple times, my controllers are stable and they are working with almost 100hz, the simulation mechanism is working well, but the problem was not solved, I guess that it comes from the possible inaccuracy or uncertainty from the ROS or Gazebo. Any advice, solution, help?

edit retag flag offensive close merge delete