[Orca Hand] Retargeting latency and glove mapping for builders researchers (intermediate)
Orca Hand retargeting discussion on glove mapping, motion latency, operator feel, and practical checks for teleoperation responsiveness.
Orca Hand demos often look acceptable in video while still feeling too slow or too confusing for the operator to produce consistent data.
How do you measure and improve retargeting latency and glove-to-hand mapping before running longer teleoperation sessions?
Please share practical checks for operator feel, finger correspondence, update-rate bottlenecks, and what level of lag is still usable.
If you reply, include one exact latency symptom and one exact change that made the hand feel more predictable.
Searchers usually want to know whether the bottleneck was sensing, retargeting math, transport, or actuator response. Please name the layer.
If you have a quick benchmark for teleop feel, such as a repeated pinch or reach-and-close sequence, post that too.
It is especially helpful to explain when latency becomes a dataset-quality problem rather than just a user-experience annoyance.