Home | Tutorials | Wiki | Issues
Ask Your Question
1

Nondeterministic ball bounce[SOLVED]

asked 2016-09-22 09:17:48 -0500

nzlz gravatar image

updated 2016-09-26 05:09:23 -0500

Hello,

I am getting different behaviors with exactly the same execution and world settings. As you can see in this video , I reset the world every time and launch the ball with exactly the same velocity. This happens when launching the ball against any kind of surface (not only this board).

Update: more videos: vid2, vid3.

Ball model.sdf file:

<?xml version="1.0" ?>
<sdf version="1.5">
  <model name="basketball">
    <static>false</static>
    <self_collide>true</self_collide>
    <link name="ball">
      <inertial>
        <mass>0.25</mass>
        <!-- inertia based on solid sphere 2/5 mr^2 -->
        <inertia>
          <ixx>0.00169</ixx>
          <iyy>0.00169</iyy>
          <izz>0.00169</izz>
          <ixy>0</ixy>
          <ixz>0</ixz>
          <iyz>0</iyz>
        </inertia>
      </inertial>
      <visual name="visual">
        <geometry>
          <sphere>
            <radius>0.13</radius>
          </sphere>
        </geometry>
      </visual>
      <collision name="collision">
        <geometry>
          <sphere>
            <radius>0.13</radius>
          </sphere>
        </geometry>
        <surface>
          <bounce>
            <restitution_coefficient>0.5</restitution_coefficient>
            <threshold>0.1</threshold>
          </bounce>
          <contact>
            <ode>
              <max_vel>5</max_vel>
              <min_depth>0.0001</min_depth>
            </ode>
          </contact>
        </surface>
      </collision>
    </link>
  </model>
</sdf>
edit retag flag offensive close merge delete

2 Answers

Sort by ยป oldest newest most voted
1

answered 2016-09-23 09:51:14 -0500

nzlz gravatar image

After testing different parameters changing the following seems to be a decent fix:

Gazebo->Physics->solver->iterations = 100 (default is 50)

edit flag offensive delete link more
0

answered 2016-09-22 13:54:59 -0500

This is a pretty lame answer, but--numerical instability. My guess is that this is an issue with the underlying computations done in gazebo, and more importantly in whichever physics engine you're using. It's also possible the ball is moving slightly around the launcher, and the timing of when you fire matters.

edit flag offensive delete link more

Comments

The timing is the same (or almost) since the ball is automatically launched when I reset the simulation. About the physics engine, I wasn't able to make the ball bounce using bullet.

nzlz gravatar imagenzlz ( 2016-09-23 08:52:46 -0500 )edit
Login/Signup to Answer

Question Tools

1 follower

Stats

Asked: 2016-09-22 09:17:48 -0500

Seen: 260 times

Last updated: Sep 26 '16