Back to Projects

Gait AI: Smart Running Insole

An AI-powered smart insole that detects foot pressure in real time and flags form issues before they cause injury.

🏆 Winner, The 2026 Spring Generator Buildathon @ Babson College

ESP32ArduinoForce Sensors (FSR)Next.jsHardwareHealthTech

Overview

Gait AI is a smart insole system that detects pressure distribution across your feet in real time and flags form issues before they cause injury. We built a working hardware prototype, a fully deployed web app, complete market research, a GTM strategy, and unit economics projected months out, all in a single ten-hour sprint.

Favourite question from the judges: "Did you really build this in a day?"

Technical Architecture

The system is a full hardware-to-software pipeline built from scratch during the hackathon:

  1. Force-Sensitive Resistors (FSRs): Multiple FSR sensors are mounted at key pressure points on the insole (heel, ball of foot, and toe) to capture the full pressure distribution during a stride.
  2. Protoboard & ESP32 Microcontroller: The FSRs are soldered onto a custom protoboard wired to an ESP32 module. The ESP32 reads analog voltage from each sensor through its ADC channels, converting pressure into digital values.
  3. Arduino Backend Processing: Custom Arduino firmware on the ESP32 handles sensor calibration, data smoothing, and packages pressure readings into structured data transmitted over Bluetooth Low Energy (BLE).
  4. Next.js App Frontend: A mobile-first web application built in Next.js receives real-time pressure data from the ESP32 via BLE, visualizes foot strike patterns through a live coaching interface, and flags gait asymmetry to reduce injury risk.

Hardware Design

The physical prototype was designed to be wearable and functional for live demonstration:

  • FSR sensors at three contact points on the insole: heel, ball of foot, and toe
  • Custom protoboard layout to shorten wire runs and keep signal integrity from the analog sensors
  • ESP32 mounted in a 3D-printed enclosure with an adjustable strap for ankle attachment
  • Multi-conductor ribbon cable connects the insole sensors to the ankle-mounted processing unit
  • The full assembly fits inside a standard shoe

Tech Stack

  • Microcontroller: ESP32 (Wi-Fi + Bluetooth LE capable)
  • Sensors: Force-Sensitive Resistors (FSRs) at heel, midfoot, and toe zones
  • Firmware: Arduino (C++), sensor calibration, data smoothing, BLE transmission
  • Mobile App: Next.js, BLE data receiver, real-time visualization, gait analysis
  • Enclosure: 3D-printed ankle-mount housing with adjustable strap
  • Prototyping: Custom protoboard, soldered connections, ribbon cable harness

Results & Awards

🏆 Winner

The Generator Buildathon

10 Hours

Concept → Working Prototype

500+

Students from 30+ Colleges

Delivered a complete end-to-end product in a single day: working hardware prototype with live sensor data, a deployed app, market research, GTM strategy, and projected unit economics. Competed against 500+ students from over 30 colleges for $15k in prizes, sponsored by Anthropic, OpenAI, WHOOP, HubSpot, ElevenLabs, Cursor, and more.

Project Media

Gait AI team presenting on stage at The Generator Buildathon at Babson College
Presenting Gait AI on stage at The Generator Buildathon. "An AI running coach in your shoe."
The Gait AI team holding the prototype insole and ESP32 enclosure after winning the buildathon
The Gait AI team after the win, holding the smart insole prototype and ESP32 processing unit.
Gait AI smart insole with FSR force sensors soldered to protoboard and connected to ESP32 microcontroller
The Gait AI smart insole: three FSR force sensors mounted at heel, midfoot, and toe pressure points, wired through a ribbon cable to the ESP32 protoboard.
Gait AI complete assembly, shoe with embedded sensor insole connected to 3D-printed ESP32 ankle enclosure
The complete Gait AI wearable assembly: sensor insole installed in the shoe with wiring routed to the 3D-printed ESP32 enclosure and adjustable ankle strap.

Links

Pitch Deck (PDF)

Tools & Methods

ESP32Arduino (C++)Force-Sensitive ResistorsBluetooth LENext.js3D PrintingProtoboard / Soldering

Built in 10 hours at The Generator Buildathon @ Babson College.