Feature #1715
Provide better interface for creation of wrench
| Status: | Resolved | Start date: | 01/06/2014 | |
|---|---|---|---|---|
| Priority: | High | Due date: | ||
| Assignee: | - | % Done: | 100% | |
| Category: | Software | |||
| Target version: | rci0.5 |
Description
In order to make it easier to create a rci::Wrench, please provide a better interface like you did e.g. for rci::Poses.
Related issues
Associated revisions
History
#1 Updated by Anonymous over 10 years ago
- Assignee changed from Anonymous to C. Emmerich
#2 Updated by Anonymous over 10 years ago
- Category set to Software
- Target version set to rci0.5
#3 Updated by C. Emmerich over 10 years ago
- Assignee changed from C. Emmerich to Anonymous
Sorry, Arne. But I will not write a converter for this type unless the type and the interface is conceptually clear to me. On the one hand there is rci:Wrench which stores cartesian torques somehow as quaternions (and I dont even know whether it is possible to do so; at least unit quaternions dont make sense here in my opinion), on the other hand there is rst.wrench storing cartesian torques as 3-dim vector, which is kind of interpretable and goes inline with ROS etc and the LWR (whatever representation that is by the way). I dont know the connection between these two!
So, you have 3 options:- Give me an extended and complete course of how you came to the quaternions representation and how to convert this into a 3-dim-vector representation. Then I will do the coding.
- Change the representation of rci::Wrench into a 3-dim vector and let's discuss what exact representation/meaning that could be/have ... and I will do the coding.
- Write the converter by yourself.
So, which door is yours? Number 1, 2 or 3?
:D
#4 Updated by C. Emmerich over 10 years ago
#5 Updated by Anonymous over 10 years ago
It seems to indeed be unconventional to use quaternions for torques, yet possible (AFAIR), see this (especially first answer and its links). So it seems that the most natural way to represent cartesian torques is indeed a 3D vector with torques about x, y, z.
As you mentioned, this would be in line with ROS, Orocos and rst.
This would be door 2, Sir.
However, all of the above formats don't seem over-specified to me. But even after some searching, I did not find a real best-practice for representing torques. I think the best way is to have a look at the torque representations in the well-established physics engines (bullet, etc.).
#6 Updated by Anonymous over 10 years ago
We should consider the publication Geometric Relations between Rigid Bodies: Semantics for Standardization, which is a recent publication on this topic, including force and torque representations (see Appendix, Section VII-D, Table V).
#7 Updated by C. Emmerich over 10 years ago
Do you know where I can find the appendix? It is not included in this pdf (and also not in part2 - another part of this tutorial)...
#8 Updated by Anonymous over 10 years ago
Sorry, wrong link. Here it is with appendix (can't uplaod because of IEEE copyright stuff).
#9 Updated by J. Moringen over 10 years ago
In r719, is the second parameter of the Wrench constructor by-value (as opposed to by-reference) on purpose?
Wrench(const Forces& f, const Torques t);
#10 Updated by C. Emmerich over 10 years ago
No, this was not on purpose. Changed in r720. Thx for the hint.
#11 Updated by C. Emmerich over 10 years ago
- Status changed from New to Resolved
- % Done changed from 0 to 100
I consider this issue as closed. For final review of (new) rci::Wrench interface/modelization, see #1043.