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.

1

IK with KDL

kdl
ik

asked 2013-12-10 15:30:40 -0500

85 ●13 ●15 ●20 https://www.linkedin.c...

updated 2013-12-10 23:05:59 -0500

Hi, everyone! I have a similar problem and I solved it like described in this topic ros-users. Program works fine, but when I try calculate joints values more different than inital values, IK solver not found. Source files: leg_kinematics.cpp hp_chainiksolverpos_nr_jl.cpp

For example. Inital frame of end-effector : 0.14208; -0.14555; -0.11145 Inital joints : 0, 0, 0

Desired frame : 0.16574, -0.1868, -0.08033

(delta_twist from hp_chainiksolverpos_nr_jl)

[ INFO] [1386694221.094697005]: delta_twist.vel 1: 0.023664 delta_twist.vel 2: -0.041252 delta_twist.vel 3: 0.031120  
[ INFO] [1386694221.094781119]: delta_twist.vel 1: 0.003682 delta_twist.vel 2: -0.006420 delta_twist.vel 3: -0.009216  
[ INFO] [1386694221.094833633]: 
Joint 1: -0.000007 
Joint 2: 0.245227 
Joint 3: 0.211637

Desired frame : 0.098897, -0.25073, 0.053264 ( joints: 0.46, 0.73, 0.84)

[ INFO] [1386694869.786644209]: 
delta_twist.vel 1: -0.043179 delta_twist.vel 2: -0.105182 delta_twist.vel 3: 0.164714 [0m 
[0m[ INFO] [1386694869.786831916]: 
delta_twist.vel 1: 0.024520 delta_twist.vel 2: -0.157306 delta_twist.vel 3: 0.083648 [0m 
[0m[ INFO] [1386694869.786888482]: 
delta_twist.vel 1: -0.016815 delta_twist.vel 2: -0.191048 delta_twist.vel 3: 0.084714 [0m 
[0m[ INFO] [1386694869.786932308]: 
delta_twist.vel 1: 0.036921 delta_twist.vel 2: -0.185430 delta_twist.vel 3: 0.084714 [0m 
[0m[ INFO] [1386694869.786992199]: 
delta_twist.vel 1: -0.014347 delta_twist.vel 2: -0.214840 delta_twist.vel 3: 0.084714 [0m 
[0m[ INFO] [1386694869.787067596]: 
delta_twist.vel 1: 0.036921 delta_twist.vel 2: -0.185430 delta_twist.vel 3: 0.084714 [0m 
[0m[ INFO] [1386694869.787110975]: 
delta_twist.vel 1: -0.014347 delta_twist.vel 2: -0.214840 delta_twist.vel 3: 0.084714 [0m 
[0m[ INFO] [1386694869.787151938]: 
delta_twist.vel 1: 0.036921 delta_twist.vel 2: -0.185430 delta_twist.vel 3: 0.084714 [0m 
[0m[ INFO] [1386694869.787192158]: 
delta_twist.vel 1: -0.014347 delta_twist.vel 2: -0.214840 delta_twist.vel 3: 0.084714 [0m 
[0m[ INFO] [1386694869.787235519]: 
delta_twist.vel 1: 0.036921 delta_twist.vel 2: -0.185430 delta_twist.vel 3: 0.084714 
[31m[ERROR] [1386694869.787273642]: IK not found

I can not understand what the problem is...

edit retag flag offensive close merge delete

2 Answers

Sort by » oldest newest most voted

3

answered 2013-12-16 10:14:08 -0500

Pablo Iñigo Blasco

2982 ●74 ●121 ●128 http://www.linkedin.co...

Your URDF describes a 3 DOF manipulator. However, your solver sets 6 constraints on the end effector (3 position + 3 orientation). In general, it is not possible to satisfy the 6 constraints only with 3DOF. The jacobbian you are using is 6x3. That means that it tries to satisfy the 6 constraints. That is because you get those incorrect and strange solutions.

And that is why I developed the solution described in http://ros-users.122217.n3.nabble.com/IK-with-KDL-td1867716.html?

It worked well for me reaching a 3D position with a 3DOF manipulator.

Another choice is adding 3 "fake revolute artificial joints" in the end effector.

However, the best solution should be change the jacobbian matrix to a 3x3 matrix that only considers the final position of the end effector (ignoring its final orientation). If you (or someother) develop this solution, please, share it with us :-).

edit flag offensive delete link

Comments

But I have same solution in hp_chainiksolverpos_nr_jl.cpp: if(Equal(delta_twist.vel,Vector::Zero(),eps)) break; delta_twist.rot ignored. And I do: ik_solver_vel.setWeightTS(matrix_Mx); Last three diagonal values are 0 in matrix_Mx. Is this like your solution?)

tuuzdu ( 2013-12-17 04:57:42 -0500 )edit

The weight matrix was designed to break the numeric search if a partial solution (in position but not in the orientation)was found. It doesn't improve the numeric search itself. It still may be very hard to find even a "partial solution" if the kinematic characterization or the goal is not suitable.

Pablo Iñigo Blasco ( 2013-12-17 23:51:39 -0500 )edit

1

answered 2013-12-10 20:20:30 -0500

Ruben Smits
543 ●4 ●10 ●16 http://www.intermodalics.eu/

Hello,

Are you sure the end position can be reached? Also your delta_twist.vel 1 value seems to jump around, could it be that your solution is near a singular position of the Chain. Could you provide us the Chain you are using?

edit flag offensive delete link

Comments

Yes, I'm sure, because I use rviz (with joint_state_publisher, robot_state_publisher and tf) to get values of joints and cartesian coordinates. I attached urdf file to above.

tuuzdu ( 2013-12-10 23:13:30 -0500 )edit

Anyone have ideas?

tuuzdu ( 2013-12-12 17:16:52 -0500 )edit

Question Tools

1 follower

Stats

Asked: 2013-12-10 15:30:40 -0500

Seen: 1,265 times

Last updated: Dec 16 '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