Demonstration timing: telemetry to latency-corrected action label Labels are constructed on the same physical decision-time grid used online; lead and aggregation change the label, while offset is an open-loop diagnostic. Recorded frame_start stream · irregular telemetry timestamps Decision-grid sample construction Evaluation paths frame_start held new sample false true diagnose phase candidate offset feedback policy Telemetry F_j @ τ_j position · velocity ·race timer Control U_j held from τ_j Telemetry F_{j+1} next recordedtimestamp Control U_{j+1} held from τ_{j+1} … later frames monotonic race-timeaxis Decision tick t_k next tick ≥ t_k +decision_interval_ms Observation x_k feature pipeline onselected telemetry Shifted label time τlabel = t_k +action_lead_ms Aggregatecontrols? config flag Point label first recorded actionat or after τlabel Window label integrate windowthen quantize once BC training sample x_k + discrete labelonline cadence Open-loop replay action selected bycurrent race timer --action-offset-ms sweep replay timestampsto diagnose latency Trajectory tracking state-based feedbackcorrects lateral error Closed-loop benchmark finish · pace ·interventions · failures Aggregation requires control_alignment=frame_start and integrates [t_k +lead, t_{k+1} + lead). Gas/steering use duration-weighted means; brake tapsuse overlap duration before quantization. --action-offset-ms shifts only open-loop replay timestamps for diagnosis. Itdoes not rewrite the demonstration or replace closed-loop bc-benchmark.