[ros-users] [Discourse.ros.org] [Humanoids] Humanoid_msgs Definitions

[Olivier Stasse]

Hello, from another humanoid user group perspective (LAAS, Gepetto)

IMHO I would be in favor of having a message close to what JSK is using because it is going in the direction of multi contact planning.
A multi-contact planner would provide at least a list of contact stances with the duration between each contact stance.

An extension of the limb values to the URDF body identifier would probably work if someone wants to do multicontact with other bodies.

At least one field would be missing removing the contact or going to contact.

If there is no simultaneous contact event that would be enough.

With respect to the issue raised by the end effector trajectory we have tried:
- Half step ( y, z position)
- Polynomials
- Bezier curves
- Joint trajectories

Because the connection between the planner and a real-time walking pattern generator often involves 
generating a CoM trajectory after the planning the last solution is likely to be modify by the real time controller.
A field indicating the type of trajectories published in a different topic would be maybe more practical ?





---
[Visit Topic](https://discourse.ros.org/t/humanoid-msgs-definitions/2119/11) or reply to this email to respond.