[ros-users] cartesian trajectories

Antoine Rennuit antoinerennuit at hotmail.com
Mon Nov 10 10:54:14 UTC 2014

Hi guys,

Willing to reopen this thread... And I have 2 remarks to start with:

Should we consider that REP-i0003 is a direct answer to last year's API review on cartesian trajectories?I really believe that the trajectory definition and the way this trajectory is achieved (e.g. via impedance control) are 2 different subjects which should be treated in isolation: let us first make the cartesian trajectory definition good and then later we can build on it for impedance control. But really cartesian trajectory is path and impedance is control: complementary but different subjects.



From: konradb3 at gmail.com
Date: Fri, 7 Jun 2013 09:12:37 +0200
To: ros-users at code.ros.org
CC: ros-sig-robot-control at googlegroups.com
Subject: Re: [ros-users] cartesian trajectories

After considering some of possible use-cases i came up with yet another proposal (more detailed) :
[1] http://ros.org/wiki/trajectory_msgs/Reviews/Cartesian%20Trajectories_API_Review_2013_06_05

Konrad Banachowicz

2013/6/6 Konrad Banachowicz <konradb3 at gmail.com>

Yes it is definitely out of scope of simple cartesian trajectory but it contains sub-message CartesianTrajectory which is simple and quiet self contained. I think that such complex goal message gives us good basis for discussion on it's building blocks like : CartesianTrajectory, CartesianImpedance ....

Konrad Banachowicz

2013/6/6 Georg Bartels <georg.bartels at cs.uni-bremen.de>

    On 06/05/2013 09:40 PM, Jonathan Bohren


          On Wed, Jun 5, 2013 at 1:18 PM,
            Konrad Banachowicz <konradb3 at gmail.com> wrote:

Header header  # A stamp of 0 means "execute now"

string[] effector_names
CartesianTrajectory[] trajectory
  PoseStamped tool  # The frame which is being controlled
  CartesianTrajectoryPoint[] points
    duration time_from_start
    Pose pose
    Twist twist
CartesianImpedance[] impedance
    TBD stiffness % cartesian stiffness 
    TBD damping % damping ratio
CartesianTolerance[] path_tolerance  # Tolerance for aborting the path
  float64 position
  float64 orientation  # Permitted angular error
  float64 velocity
  float64 angular_velocity
CartesianTolerance[] goal_tolerance  # Tolerance for when reaching the goal is considered successful
JointTrajectory posture  # For determining the redundancy
JointImpedance nullspace_impedance # TBD


        I like where this version is
    Konrad's proposition looks indeed promising. However, I'd like to
    point out that this is more than a Cartesian Trajectory. This looks
    more like a goal message to a very sophisticated controller, e.g.
    whole-body motion framework or iTaSC. Shouldn't the Cartesian
    Trajectory msg be somewhat smaller in scope and then be reused?
    Maybe more like in the spirit of std_msgs which are mainly used as
    building blocks for bigger and semantically annotated messages?


        Meanwhile, between comments in
          different mailing lists, and comments on the wiki in the old
          robot_mechanism_controllers, it'd be great if we could
          aggregate this discussion. Really it seems like this is an
          extension to the trajectory_msgs package, so maybe we can do
          an online review where people can propose and comment on
          different options and subjects. I've started one [1] with the
          initial proposal from the old wg review, and added Konrad's
          proposal above (feel free to remove/edit that, Konrad). I
          think we can use the robot control SIG to coordinate, and ping
          ros-users for high-level notices. Then when we've at least
          gotten the scope down, we can initiate an REP request. How
          does that sound?

        [1] http://ros.org/wiki/trajectory_msgs/Reviews/Cartesian%20Trajectories_API_Review_2013_06_05

    Good idea. From now on moving to the REP?





          Jonathan Bohren

          Laboratory for Computational Sensing and Robotics





ros-users mailing list
ros-users at code.ros.org




ros-users mailing list

ros-users at code.ros.org


ros-users mailing list
ros-users at code.ros.org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ros.org/pipermail/ros-users/attachments/20141110/38bc259c/attachment.html>

More information about the ros-users mailing list