External Force Sensing Reliablity

[RLi43]

I would like to share some observations here:

• KUKA Sunrise provides API to fetch estimated external force and torque applied on the flange or a fixed framed attached to flange, as well as the inaccuracy of the measurement.
• With KUKA Sunrise FRI 3.0, external force is available via FRI.
• Both measurements from KUKA and lbr-stack FT estimator subject to drift and zero drift. Data from KUKA looks more stable (not yet systematically and quantitively compared).
• Measurement in position control mode (rather than impedance control modes) looks more stable.

If it can be confirmed that estimation from KUKA internally is more reliable, I suggest to implement External Force in LBRState. (Well, you also need FRI 3.0, or an extra network channel to fetch it). Still, installing a force sensor on the flange is a more reliable option.