ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question
1

katana teleop using phantom

asked 2013-07-18 00:21:56 -0500

rem870 gravatar image

updated 2013-07-18 01:44:49 -0500

Hi, is it possible to teleop katana arm robot with a phantom using ros packages?

thanks

edit retag flag offensive close merge delete

2 Answers

Sort by » oldest newest most voted
1

answered 2013-07-18 01:40:47 -0500

updated 2013-07-18 01:42:44 -0500

There's no reason why you couldn't do that. The phantom_omni package gives you a 6D pose (the pose that you want the end effector to be in). You'd have to create an arm navigation configuration package for your complete robot (like kurtana_arm_navigation), then you can write a node that sends the pose from the phantom as a goal to the move_arm action. The motion planner would then plan a path to the desired pose and execute it on the robot.

Maybe even better would be to implement an inverse kinematics controller instead of using motion planning; perhaps this would allow you to send goals more frequently, allowing for a smoother motion of the arm. The package katana_ikfast_kinematics_plugin provides a 5D IK for the Katana arm.

You can also have a look at the pr2_omni_teleop package, where they developed something similar for the PR2.

edit flag offensive delete link more
0

answered 2013-08-06 04:00:14 -0500

rem870 gravatar image

I could read the desired pose from the phantom calculate the desired positions via an inverse kinematics algorithm on matlab or c++ and then subscribe to the /katana_arm_controller/joint_movement_action/goal.

is that right???

thanks!

edit flag offensive delete link more

Comments

1

The katana stack already provides an inverse kinematics package; you can simply send the desired pose to move_arm. NB: please use the comment function for feedback, don't create a new answer.

Martin Günther gravatar image Martin Günther  ( 2013-08-12 04:02:55 -0500 )edit

i think move_arm is extremely slow, so its better to do it via inverse kinematics

rem870 gravatar image rem870  ( 2013-08-16 05:02:35 -0500 )edit

I wouldn't say move_arm is "extremely slow". But if you want more reactive behavior, you should indeed use inverse kinematics, as I already suggested.

Martin Günther gravatar image Martin Günther  ( 2013-08-25 23:11:20 -0500 )edit

Question Tools

Stats

Asked: 2013-07-18 00:21:56 -0500

Seen: 366 times

Last updated: Aug 06 '13