Robotics / Digital twin / Check against prediction

Digital twin · prediction check

Overload against Held load

Observed run

Prediction

• base: position is 0.040 rad rms off the prediction
• base: current is 560% off the prediction (1.098 A rms)
• base: winding peaks 38.5 C above the prediction
• shoulder: position strays 1.358 rad from the prediction, past the 0.1 rad fault limit
• shoulder: position is 0.701 rad rms off the prediction
• shoulder: current is 194% off the prediction (3.465 A rms)
• shoulder: winding peaks 58.5 C above the prediction
• elbow: position strays 0.977 rad from the prediction, past the 0.1 rad fault limit
• elbow: position is 0.429 rad rms off the prediction
• elbow: current is 154% off the prediction (1.244 A rms)
• elbow: winding peaks 51.0 C above the prediction
• wrist: position strays 0.566 rad from the prediction, past the 0.1 rad fault limit
• wrist: position is 0.070 rad rms off the prediction
• wrist: current is 340% off the prediction (0.815 A rms)
• wrist: winding peaks 30.7 C above the prediction

Both runs are simulations: the observed run stands in for a real joint until a device posts one with source device:<id>.

The observed run carries 3 kg on tool; the prediction carries 0.3 kg on tool.

The observed scenario is "min-jerk reach and hold, payload past the shoulder's peak torque" and the prediction's is "min-jerk reach and hold".

Observed

Printed arm, stand test fault

Prediction

Printed arm, stand test ok

Rules

command within 0.01 rad rms; position watch 0.02 rad rms, fault 0.1 rad max; current watch 25%; winding watch 5 °C, fault 15 °C

Joints