Lean angle tracking with a phone: the physics and the pitfalls

Lean angle is the most requested and most misreported number in motorcycle apps. Riders love it, and a lot of implementations quietly produce fiction.

Here is why it is hard, how it should be done, and why we deliberately keep it off the screen while you are riding.

The physics that catches everyone out

Here is the intuition trap. You are mid-corner, knee out, grinning, bike leaned right over. Your phone’s accelerometer, if you read it directly, reports that the phone is nearly vertical.

That is not a bug. It is correct.

In a steady-state corner, the bike leans until it aligns with the resultant of two forces: gravity pulling down and centripetal force pushing sideways. From the bike’s own frame of reference, that resultant points straight “down” through the tyres. That is the entire reason a leaned motorcycle feels planted rather than like it is falling.

So an accelerometer bolted to the bike, which measures exactly that resultant, reports “down is down” and hence very little lean.

Any app that reads phone tilt and calls it lean angle produces numbers that are far too small in steady corners and wildly noisy everywhere else.

What correct measurement needs

To get a real number you need to reconstruct the bike’s attitude in the world, not in its own frame:

Gyroscope integration. The gyro measures rate of rotation, which is frame-independent. Integrate roll rate over time and you get change in lean. The catch is drift: integrated gyro data wanders within seconds, so it cannot stand alone.

Accelerometer as a long-run reference. Over a long enough window, and while riding straight, the accelerometer does tell you where true vertical is. Fuse the two — gyro for fast changes, accelerometer for slow correction — and you get an attitude estimate that is both responsive and drift-free. This is standard sensor fusion, the same idea behind every drone and phone orientation system.

Speed, from GPS. Lean angle in a steady corner relates to speed and corner radius. Speed lets you sanity-check the estimate and detect the corners in the first place.

Mount calibration. A phone sitting at 8 degrees in its cradle biases every reading by 8 degrees. There is no way to infer this from the data alone, so it has to be measured once with the bike upright. Skip it, and your left-handers and right-handers will not match — which is the tell-tale sign of an uncalibrated app.

Rejection of dirty data. Hard braking, hard acceleration, road camber, kerbs and speed bumps all corrupt the estimate. Readings taken during those moments must be discarded rather than averaged in, or your peak lean angle becomes a record of the worst pothole you hit.

Why the numbers you see elsewhere are too big

Once you understand the sources of error, the pattern is easy to spot:

• Camber. A road that falls away adds apparent lean.
• Pitch leak. Braking and accelerating rotate the phone about a different axis, and a naive model bleeds some of that into roll.
• Vibration. High-frequency buzz shows up as short spikes. If the app records maximum lean as the highest single sample, one buzz becomes your personal best.
• Optimism as a feature. Bigger numbers feel better and get shared. There is no incentive for an app to be conservative.

A useful sanity check: on public roads, on cold-ish tyres, in normal riding, peak lean lives in the thirties to mid-forties. Sustained fifties is track riding. If a Sunday ride to the bakery reports 61 degrees, the maths is wrong, not you.

Why we do not show it live

This is a product decision, not a technical one, and we hold it firmly.

A live lean angle meter on a phone mounted to the bars is a target. Riders will chase it. Chasing a number on a screen means looking at a screen and asking for more lean on a public road with gravel, damp patches, oncoming traffic and an unknown radius — and each of those is a much better reason to be upright than any number is a reason to lean further.

So lean angle is recorded during the ride and shown afterwards, in the post-ride review, next to the corner it happened in. There you can look at it properly, compare it to what the road did, and learn something. The bike is parked, which is the only sensible place to be interested in your maximum lean angle.

What it is actually good for

Consistency, not maximum. The interesting question is not your peak. It is whether your left-handers match your right-handers. Nearly every rider is meaningfully better on one side, and a consistent gap of several degrees is a real, fixable thing — usually a body-position habit or a confidence asymmetry.

Smoothness. How quickly you go from upright to leaned, and whether the transition is one clean input or three corrections, says more about your riding than the number at the apex.

Seeing a road you rode. Overlaying lean against the route shows you which corners you actually used and which you straight-lined without noticing.

Not for comparing with mates. Different phones, different mounts, different calibration, different roads. Cross-rider comparison of lean angle is noise dressed as competition.

If you want the number to mean something, calibrate it once, ride normally, and look at it when you get home.