BUG: Velocity compute and scale fix
So, getting relative velocity to behave turned out to be tricky. The main issue comes down to error caused by sampling rates between the RBD/PBD update rate and the haptic update rate. When mixing the two strangeTM things happened, and non-realistic forces ended up being rendered to the device. Also, it seems that the rigid PBD solver is more sensitive to this than the RBD solver, and I suspect this is because the numerical error in the rigid PBD solver is larger than the RBD solver.
That being said, there were multiple bug fixes along the way, so this MR is to get those in. They are:
- Updating scale in OpenHapticDeviceClient for velocity
- Fixing angular velocity calculation to avoid quaternion flipping error
- Applying user defined translation scaling to velocity in TrackingDeviceControl
Edited by Andrew Wilson