REQUIREMENTS
What you need.
- A compatible LiDAR-equipped iPhone Pro running iOS 18 or later.
- A preferably straight, visible road section of about 20 metres or more.
- A safe observation point outside moving traffic.
- Fixed visual features such as kerbs, buildings, posts or road markings.
- Useful lighting and an unobstructed view of part of the vehicle path.
A fixed mount is not required. During measurement, keep the iPhone as steady and as close to the saved observation point as practical.
SETUP
Three separate steps.
Route profile
First, save the route name and local speed limit. The limit can be changed later and determines whether an estimate is displayed in green or red.
Observation point
The app saves the intended camera view of the road and fixed surroundings. This image supports later visual re-anchoring. An optional coarse GPS location only warns when the location is clearly wrong; it never blocks a run.
Route
Mark four road points directly in the camera view while the AR session stays active. LiDAR and stabilised raycasts provide their spatial positions.
GEOMETRY
The A–D rectangle.
The safe walk starts at A, continues to B on the right, crosses only when safe to C and returns to D on the left. The app derives length, width and the road plane from these points.
A–D provide the metric reference. The start and end edges are not hard gates that a vehicle must fully cross.
MEASUREMENT
Visual positions, not a stopwatch.
When measurement starts, the current camera image is registered against saved anchor images. Four corner indicators help with alignment. A plausible partial registration is enough; weak quality remains visible as a diagnostic but does not automatically stop the process.
A Core ML model detects vehicles locally. For each stable track, the lower centre of the vehicle box is projected onto the calibrated road plane. Multiple positions and timestamps form a robust linear estimate of distance travelled.
A regular estimate uses at least five image points with four statistical inliers. Three geometrically consistent points can produce an explicitly marked approximation. Brief occlusions of roughly two seconds may continue the same track.
RESULT
More than a single number.
Each locally saved record may contain:
- speed in km/h with optional mph display
- time, direction, visible partial distance and duration
- number of image points and inliers
- fit error, confidence and technical diagnostics
- the best locally captured vehicle image from the track
Estimates above the route profile’s speed limit are shown in red. Measurement archives remain local and can be deliberately shared as CSV or diagnostics.
ACCURACY
A technical estimate.
Accuracy depends on calibration, distance and perspective, lighting, visible duration, stable vehicle identity and spatial image registration. Cropped vehicle boxes, short visible paths and steep perspective increase uncertainty.
Vmax check is not calibrated or approved for official, legal or evidential speed measurements. Do not interpret individual estimates without checking their plausibility.
SAFETY
The road comes first.
- Only calibrate when traffic conditions are safe.
- Cross the road only at a safe location.
- Never stop on the carriageway or focus on the screen while crossing.
- Measure from a safe position outside moving traffic.
- Follow local rules and respect the rights of people in captured images.