THE VERDICT
What the stopwatch is actually saying
Evidence → interpretation → the next driving action.
INTERACTIVE LAP LAB
Replay every lap on the clock
Play advances by measured elapsed time: the kart visibly covers less track while slowing and more while accelerating. The numbered ghosts align each separately recorded lap at t=0 for a virtual time-relative comparison; the laps were not driven simultaneously.
TIME-ALIGNED LIVE REPLAY
GNSS ± metresSelected kart + all-lap ghosts
Large lime kart = selected lap. Numbered circles = separately recorded laps aligned to the same elapsed time, not simultaneous karts. Paths remain measured and unsnapped.
PACE TRACE
GNSS speed validatedSpeed and time delta by distance
LEARNING CURVE
estimated timingLap time progression
SECTOR DNA
4 fixed gatesWhere each lap won or lost
RACING-LINE & CORNER MODEL
A model opponent—without calling it ground truth
Suggested line, corner-speed envelope, inferred braking and low-confidence rotation cues. Measured telemetry remains primary; every coaching number changes with the assumptions.
Driver scenario: 72 kg · accelerator flat whenever acceleratingComparison, not proof: braking is inferred from deceleration; these sensors do not measure tire slip, so drift and skid cannot be confirmed.
Selected model lap——
Kart + driver—scenario input
Fitted effective grip—not a measured tire μ
Near-term benchmark—your own sector stack · stronger evidence
SUGGESTED LINE
scenario onlyAssumed corridor, model line & measured path
The translucent band is OSM centerline ±4.5 m—not surveyed asphalt. The dashed blue GNSS path has metre-scale uncertainty, so lateral separation is illustrative rather than an apex verdict.
SPEED ENVELOPE
QSS modelMeasured lap versus simulated full-throttle / brake solution
Orange bands are model braking demand. Actual braking is inferred from coherent speed loss; there is no brake-pedal channel. Hover either chart to move the shared track cursor.
CORNER-BY-CORNER
Evidence → interpretation → one safe experiment
Observed / conservative-model minimum speed. Brake distances are coarse 5 m bins with practical uncertainty around 15–25 m.
What “skid”, “drift” and grip mean here
Grip proxy combines smoothed phone/body lateral acceleration and inferred deceleration inside a fitted friction circle. Possible over-rotation requires excess yaw, elevated grip use and local time loss. Neither proves tire slip: phone movement, mounting, bumps, GNSS heading and the unrecorded steering angle can create the same pattern. There is no defensible tire slip angle.
Line, speed and braking assumptions
The line minimizes curvature inside a 9 m OSM-centered corridor, leaving the 1.35 m kart footprint and 0.55 m centerline safety margin. Speed is a cyclic point-mass simulation with an effective grip limit, friction-circle braking, power-limited full-throttle acceleration, rolling resistance, aerodynamic drag and a fitted top-speed cap. Elevation is omitted because the pressure-derived profile is experimental.
How to use the model safely
Start with Conservative coaching. Change only one corner and one behavior at a time. A slower minimum plus early speed-loss cue suggests testing a shorter/later main stop; a late/near-limit cue means adding margin. Do not use a nominal or high-grip ceiling as a braking instruction, and never prioritize an exact metre marker over visibility, flags, traffic, tire temperature or track conditions.
MODEL SOURCES
Facts, priors and method references
PHYSICS LAB
Grip, energy and rotation—without pretending assumptions are measurements
Mass-independent quantities stay measured. Mass, drag and rolling resistance remain explicit controls.
G–G ENVELOPE
phone/body proxyMeasured motion demand
MOTION TRACE
10 Hz filteredAcceleration and rotation by distance
ASSUMPTION-AWARE MODEL
editable estimateEnergy & tractive-power sandbox
Speed-loss energy / lap—specific until assumptions unlock
Peak kinetic energy—at selected lap top speed
P95 required engine power—acceleration + rolling + aero
Grip-demand proxy—p99 phone/body horizontal g
Model: P ≈ [m·a + m·g·Crr + ½ρCdAv²]·v / η. This estimates demand at the assumed drivetrain; it does not identify engine output or kart model.
SPATIAL MODEL
Drive the recorded lap in 3D
A camera-led replay built from the same measured-time playhead. Position and speed come from the selected lap; heading is a smoothed trajectory tangent and the road width is only a visual aid.
LIVE 3D KART REPLAY
approximate camera poseChase camera or first-person cockpit
LAP 4 · T+0.0 S0.0km/h
THIRD-PERSON CHASE0 mvisual z ×4
GNSS ≈1 Hz · heading smoothed · visual road edges are not surveyed
What is measured: elapsed time, speed and the selected GNSS line. What is visualized: interpolated motion, a smoothed look direction, experimental pressure-derived relative height and a nominal-width road ribbon centered on that line. This is not surveyed track geometry or exact steering.
3D TRAJECTORY
Speed over experimental relative elevation
Elevation ×8
— modelled range— cross-lap repeatability9 m venue-stated climbs/fallsExperimental vertical axis
KEY COMPLEXES
Five speed minima worth revisiting
Automatically detected braking/apex regions. The spread shows where execution changed most; exact racing-line claims are constrained by GNSS accuracy.
AUDIT TRAIL
Every conclusion has a source and a confidence boundary
The original archive remains untouched. Derived outputs are reproducible from hashed inputs.
DATA QUALITY
What can be trusted
METHODS
How each layer was produced
PROVENANCE
Open source folder ↗