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

Attention: Answers.ros.org is deprecated as of August the 11th, 2023. Please visit robotics.stackexchange.com to ask a new question. This site will remain online in read-only mode during the transition and into the foreseeable future. Selected questions and answers have been migrated, and redirects have been put in place to direct users to the corresponding questions on Robotics Stack Exchange. Additional details are available here.

0

KDL iterative IK solver joint angles not normalized

asked 2013-02-07 03:29:11 -0600

Stefan Kohlbrecher

flag of Germany

24361 ●173 ●299 ●444 https://www.energy-rob...

I'm using the arm_kinematics_tools package for solving the IK of a 6DOF arm. Everything is working fine, with the little caveat that the returned joint angles are not normalized.

An example:

position: [577.1623005464028, 565.4397480457859, 43040.16494263808, 441.2180765483309, -17007.80488691719, 939.0145541995307]

If those are sent to the robot_state_publisher, the arm position is fine, implying that the angles are correct, just not normalized. I'm aware that this behavior is also related to the seed state, but I was wondering if this is normal/intended behavior of KDL and if others have experienced this too.

edit retag flag offensive close merge delete

1 Answer

Sort by » oldest newest most voted

2

answered 2013-02-07 04:47:18 -0600

dbworth
1103 ●28 ●44 ●52

Hi, I remember seeing that too. In the package you'll find there are 3 types of the Newton-Raphson solver you can test. I found only one of them had this behaviour. The "best" of the three, with joint limits, did not.

But, for a 6 DOF robot, I'd generate an IKFast plugin. ;-)

edit flag offensive delete link

Comments

Ah ok, thanks :) Yes, I'd do that too (re IKFast), unfortunately there are some kinematics for which closed form IK generation fails.

Stefan Kohlbrecher ( 2013-02-07 05:00:08 -0600 )edit

DRC arm or snake robot? ;-) The basic N.R. solver is very bad. Using KDL to make a Jacobian-based solver is one, better option.

dbworth ( 2013-02-07 05:19:34 -0600 )edit

DRC arm :) See also http://openrave-users-list.185357.n3.nabble.com/Generating-IK-for-6DOF-humanoid-appendages-prototype-GFE-robot-fails-td4025593.html . Is there a Jacobian solver for KDL available?

Stefan Kohlbrecher ( 2013-02-07 08:52:30 -0600 )edit

There's functions in the KDL library to do with jacobians. I think the NASA R2 uses one, as does REEM: https://github.com/pal-robotics/reem_kinematics/blob/master/reem_kinematics_constraint_aware/src/ik_solver.cpp

dbworth ( 2013-02-07 12:04:33 -0600 )edit

Question Tools

Stats

Asked: 2013-02-07 03:29:11 -0600

Seen: 553 times

Last updated: Feb 07 '13

Related questions

ROS Answers is licensed under Creative Commons Attribution 3.0 Content on this site is licensed under a Creative Commons Attribution Share Alike 3.0 license.

Powered by Askbot version 0.10.2