in development · native iPhone app
Lean With It.
A native iPhone app that records the motion and route of a rigidly mounted motorcycle, estimates lean angle, and turns the ride into synchronized post-ride analysis.
Why I built it
I wanted useful motorcycle telemetry without adding hardware or asking the rider to stare at a screen. An iPhone already has a capable IMU and GPS receiver, but converting phone attitude into believable motorcycle lean is a calibration and signal-analysis problem—not a matter of displaying roll. Lean With It is the attempt to solve that problem honestly, preserve the evidence, and make the result useful after the bike is parked.
Highlights
Records through a locked screen
Core Motion captures mounted-phone observations at 50 Hz while Core Location records the route and keeps the background session active.
Learns the riding zero
The motorcycle can begin on its side stand. Sustained GPS-confirmed straight riding establishes the neutral bank reference used by live and post-ride analysis.
Keeps source data immutable
Motion and route observations remain untouched while versioned estimators and analyzers can rebuild ride summaries as the math improves.
Replays credible corners
Lean, speed, and route position share one timeline. Conservative corner cards expose useful events while omitting candidates that lack enough evidence.
Validated so far
- Five road recordings have produced continuous 50 Hz motion capture.
- Locked-screen recording was explicitly established on the first three rides.
- The current riding-zero estimator has repeatedly stayed close to an independent full-ride baseline.
- GPS and IMU direction agree strongly after accounting for course-reporting latency.
Still under test
The current estimator is heading-invariant and automatically corrects the starting posture, but the reported angles are still estimates. Vibration behavior, absolute lean accuracy, road camber, and broader physical left/right validation need more controlled evidence.
The app deliberately avoids traction-margin claims, lean scores, and live coaching. It is a recorder first: source observations stay local, questionable corner candidates are omitted, and analysis can be rebuilt as the estimator improves.
Distribution remains personal while the sensor model and confidence rules are being validated on real rides.
Stack
- Swift 6
- SwiftUI
- Core Motion
- Core Location
- MapKit
- Swift Testing
- XcodeGen