We built a leg-fidget counter. Here's what that actually took.

The problem nobody is measuring

Non-Exercise Activity Thermogenesis — NEAT — is the energy your body burns doing everything that isn't deliberate exercise. Fidgeting is a big part of it. Levine's research puts the gap between high and low fidgeters at 350 kcal/day. That's a real difference.

The problem: nobody knows how much they actually fidget. There's no baseline. We decided to build one.

What BinBot does

BinBot runs in the background on iPhone and Apple Watch and counts your leg movements across the day — three modes: shakes, jumps, rotations. It writes everything to HealthKit so the count lives in your health timeline alongside steps and sleep. A Live Activity and home screen widget keep the running total visible without opening the app.

The part that took the most time

We sample at 100Hz via CoreMotion and run the signal through an IIR bandpass filter at 3–9Hz. That frequency band is where leg fidgeting lives; walking and car vibration sit mostly outside it. In theory.

In practice, the edge cases are the hard part. Fast walking, rough roads, and certain seated movements can leak into the detection range. We use a five-layer rhythm classifier on top of the filter to reduce that noise, but false positives still happen occasionally. We document this honestly.

The other known limitation: Watch-to-iPhone sync is currently one-directional. The Watch logs independently and pushes to iPhone, but not the reverse.

Takeaway

Chasing perfect accuracy in motion sensing is a trap. We ship it as a trend tool — the day-over-day pattern matters more than any single count being precise.

Free, ad-supported. No paywall.

https://apps.apple.com/jp/app/id6760091302