GearSense is a retrofit eBike controller that detects what gear you're in and automatically adjusts motor assist to hold your target speed — uphill, into headwinds, under load. No buttons. No assist levels. Just ride.
Think automatic vs manual transmission — except it's your existing bike and your existing gears. The motor becomes an invisible co-rider that always knows exactly how much help you need.
Every eBike controller on the market reacts to torque or speed — but none of them know what gear you're in. That means the assist level you need on a hill in 3rd gear is completely different from the assist you need at the same speed in 7th, and the controller has no idea. You manage it manually. You shouldn't have to.
GearSense reads your gear ratio in real time from a hall effect sensor on the front wheel. No cable tap, no derailleur sensor, no cassette modification. It sees the ratio between wheel revolutions and pedal strokes, identifies which gear you're in from the clusters in that data, and adjusts motor behaviour accordingly — automatically.
Each gear produces a characteristic wheel-to-pedal ratio that clusters tightly around a fixed value. GearSense learns your specific bike's gear map on a calibration ride, then detects your current gear in real time on every subsequent ride.
The control model is simple: each detected gear has a target speed. When the bike falls below that target — because of a hill, headwind, or load — the motor increases power to compensate. When the bike is above target, the motor backs off. The rider never touches an assist setting. Shift up, the target speed rises. Shift down, it drops. Hills are handled automatically.
First real-world ride data from a commute in Shawnigan Lake, BC. A 7-speed bike, 8.5 minutes of data, algorithm running on raw sensor output.
The unknown going into that ride was whether the gear ratio would cluster cleanly enough to be useful on real roads with real vibration and interference. It does. The challenges remaining are engineering challenges — magnet retention, calibration protocol, control firmware. The concept is proven.
So like an automatic vs a standard car?
$#@%&! DO it. I'd buy one.
Neither rider had encountered an eBike that offered anything like this. The gap GearSense addresses is real and felt.
Phase 1 runs on an ESP32 dev board, a KY-003 hall effect sensor zip-tied to the front fork, and a spoke magnet. Total parts cost: under $30. Everything already in a typical maker's parts bin.
The long-term vision is a self-powered wireless sensor node — the spoke magnet that GearSense is already sensing also induces current in a harvesting coil on the fork, charging a supercapacitor that powers the sensor. Fit and forget. No charging, ever.
GearSense is developed openly. Every field test, firmware decision, and design rethink is documented as it happens.
Read the development log →