Review sensor frames

[PeterBowman]

In the context of roboticslab-uc3m/yarp-devices#263, I observed that sensor axes do not match the TCP frame described in our manual. I believe said frame matches the orientation of force/torque raw data streamed by the legacy Jr3Pci implementation (in --dextrorotary mode). However, the proper sensor axes should be obtained by rotating about the depicted X axis by 180 degrees.

In other words, given the above drawing and focusing on the arm TCPs:

• Axis X follows the normal direction "inwards" (as depicted).
• Axis Z points up.
• Axis Y would point left.

Possible solutions:

1. Keep the current TCP frame as-is, define an additional sensor frame matching the new implementation.
2. Change the TCP frame to match the new implementation, update scripts and demos.

[PeterBowman]

3. Since there is an ongoing redesign of several joints and links and therefore (I presume) the sensors will be unmounted anyway, just mount them on their opposite side.

cc @smcdiaz

[jgvictores]

@PeterBowman Is https://robots.uc3m.es/teo-developer-manual/diagrams.html#ft-sensors-ports-channels-and-coordinate-systems accurate?

[PeterBowman]

@PeterBowman Is https://robots.uc3m.es/teo-developer-manual/diagrams.html#ft-sensors-ports-channels-and-coordinate-systems accurate?

It would seem so, indeed, at least for the wrist sensors (I haven't checked the ankles yet).