Research in motion

Lean Angle started with a question rather than a product: how much of a ride can a compact device describe honestly?

The first finding is also the most inconvenient one. In a steady-state corner a motorcycle banks until gravity and cornering force resolve along its own vertical axis. An accelerometer riding along with it therefore reads straight down through the machine and reports almost no lean at all. The information is not noisy — at that moment it is simply not present in that sensor.

That single result shapes the rest of the project. It means lean angle cannot be read off one sensor, that rate of rotation has to be tracked over time, and that anything integrated over time will drift unless something else bounds it. It is the reason the architecture work ahead is about combination rather than component choice.

We are currently mapping which measurement approaches survive contact with vibration, mounting position and changing speed, and which claims we would have to drop as a result.