Update GearBox Joint Definition

Issue #1058 new
John Hsu
created an issue

Current model definition for gearbox joint is not flexible enough.

We should switch from needing parent link, child link, reference link, axis1 and axis2 to specifying 2 single dof revolute/prismatic joints.

Inspired by trying to model pr2 like 4-joint parallel gripper.

Comments (9)

  1. John Hsu reporter

    @Dave Coleman Not yet, current definition has a limitation that the two joints constrained by a gearbox constraint must share a common body. Because of timing, getting ready for 3.0 release, I don't have time to generalize gearbox in a way that one can constrain any pairs of joints with a mimic tag yet. Will look into this after the 3.0 release.

    Please vote on this issue if this is a useful feature. Thanks.

  2. Stefan Kohlbrecher

    We have a hand model that has a mimic tag and I just ran into a similar issue as described here on Gazebo Q/A, namely the hand violently shaking as soon as grasping an object is attempted (using Gazebo2). Are mimic joints supposed to be supported in Gazebo now, or are issues with them expected? The conversation in this ticket sounds like the mimicimplementation was never finished.

  3. Hamza Merzić

    Any pointers on how to properly implement joint mimicking for any of the physics solvers would be much appreciated. If it works out, I would love to try to generalize it and contribute back. Thanks!

  4. Log in to comment