berkeleyopenarms / blue_moveit Goto Github PK

Creating a analytic IK solver for Blue will reduce computation time for MoveIt!
http://docs.ros.org/kinetic/api/moveit_tutorials/html/doc/ikfast/ikfast_tutorial.html

Hi there!

I've been working on integrating the Blue arm with MoveIt on Gazebo. It's working right now, but because I had to change a bunch of other packages, I'm not too sure on how I should be contributing?

Some changes had to be made, and I'm not sure if it can be integrated without breaking any other real-life implementations, so any advice regarding this would be helpful.

Some changes:

• Split the joint_torque_controllers into two (one for each arm)
• Redefined some URDFs with mimic joints (I got them to work with a mimic_joint_plugin, but grasp simulations cause the simulation to fail on collision with the grasped object.)
• Redefined some controller gains and changed the joint_trajectory_controller controller to work with a effort_controllers/JointTrajectoryController type controller to expose an action interface.

Issue

For commit: 093ae01
In file: /blue_moveit/blue_moveit_config/launch/blue_moveit_controller_manager.launch.xml

The arg for moveit_controller_manager is not filled, which causes the arm to not start as it substitutes in the default argument, targeting a non-existant .yaml file, which breaks the control stack.

Fix

trajectory_execution.launch.xml calls blue_moveit_controller_manager, and is passed the blue_side argument from move_group.launch.

Ensure that trajectory_execution.launch.xml passes the blue_side argument to blue_moveit_controller_manager.launch.xml