Simulation timeline
│ Policy calls│ Supervisor decisions Click a marker or event to seek.
Video runs in simulation time; inference waits are omitted. Action times between recorded camera snapshots are derived from the control frequency.
Recorded camera views
These are the nearest preceding saved snapshots, not new feedback for every action.
Inference and recorded state
Policy latency · seconds per call
Supervisor latency · seconds per call
End-effector position · metres · x blue / y orange / z green
Gripper closure · recorded snapshots
State curves connect saved samples. They do not measure motion between snapshots.
Selected event
Active language instruction
Latest supervisor rationale
Selected native action
Native normalized delta OSC commands. Gripper: −1 open, +1 close. These values are commands, not measured displacement.
Latest predicted action chunk
Recorded robot state
Native outcome and run totals
Run configuration and checkpoint
Recorded supervisor request settings
Trace events
← / → step through filtered events. Space plays or pauses video.