SUPERCARS · DATA & STRATEGY LAB Telemetry intelligence · auto-retraining on every stint
Consuming from shared bus: 0 laps available Model: untrained Last event: —

LAP-TIME REGRESSION MODEL (ORACLE)

—

PREDICT LAP TIME

—

AUTO ML INSIGHT (ORACLE SYNTHESIS)

—

LAP STREAM

#TrackLap

ANOMALY KPIS

DISTRIBUTION

LOCATION

UNSUPERVISED ANOMALY SCORE

Two engines over the same telemetry channels (speed, G, tyre temps, throttle/brake, rpm): Isolation Forest (pure-Python) isolates outlier samples; the PyTorch Autoencoder (GPU when available) learns the normal channel correlations and flags samples it reconstructs poorly — catching combinations no clean lap produces. Red points exceed the threshold.

DRIVER STYLE MAP · LSTM SEQUENCE EMBEDDING

NEAREST DRIVING STYLE

Run the model to map every lap's driving style.

A shared PyTorch LSTM autoencoder learns each lap's control sequence (throttle/brake/steer/speed/lat-G/lon-G) into an 8-D style embedding, projected here to 2-D. Laps sitting close together were driven in a similar style; the highlighted point is the selected lap. First run trains the encoder (a few seconds); later runs are cached.

CONFIGURATION A (BASELINE)

CONFIGURATION B (EVOLUTION)

spacer
CAR-TO-PIT TELEMETRY · DISTANCE SYNCHRONISED

DIGITAL PIT WALL 2026

GPU-ready track intelligence, multi-lap delta and driver coaching in one synchronized engineering workspace.

ANALYSIS LINKSTANDBY
LAP DELTA · B−A—finish line
THEORETICAL BEST—best microsectors
BIGGEST OPPORTUNITY—distance window
REFERENCE—selected baseline
TRACK INTELLIGENCE · 3D · TEAM A

SELECT TWO LAPS

FRIOSUPERAQUECIDO
A
B
Move across telemetry to inspect the lap
00:00.000
PERFORMANCE ENGINEER

COACHING SYNTHESIS

AUTO

Select a circuit and load two laps. If only one lap exists, generate a reference pair.

SYNCHRONIZED TELEMETRY

DISTANCE DOMAIN OVERLAY

A  B
20 MICROSECTORS

OPPORTUNITY MATRIX

negative delta = B ahead

MANUAL PUSH INSIGHT

Custom summary if you want to override the auto-generated one.