IoT Gym Equipment Usage Tracker

A gym floor is full of expensive equipment and thin data. Owners know the membership count but not the machine count that matters: how many minutes each treadmill, cycle and bench station actually works per day, which hours saturate the floor, and which machines sit idle while members queue elsewhere. Manual observation does not scale, and commercial gym-management software tracks billing, not equipment. This project instruments the machines themselves: compact sensor nodes (vibration/motion sensing on an ESP32-class board) mount on each piece of equipment and classify it as in-use or idle from its vibration signature, reporting state changes over Wi-Fi to a gateway that aggregates active minutes per machine. The cloud dashboard renders a live floor map with IN USE / FREE badges, per-machine utilization bars, a peak-hours heat view and daily totals — the exact numbers behind "we need another squat rack". The student delivers the sensing hardware, the gateway logic and a genuinely working dashboard demo.

1. A sensor node is mounted on each machine (treadmill frame, cycle body, bench station); it samples its accelerometer/vibration sensor continuously at low duty cycle.
2. Firmware classifies the machine as active when vibration energy exceeds the tuned threshold for the debounce period, and idle after sustained quiet — state changes are transmitted immediately.
3. Nodes report over Wi-Fi to the gateway, which timestamps each state change and accumulates active minutes per machine in the cloud backend.
4. The dashboard polls the backend to render the live floor map, per-machine utilization bars, the hourly peak pattern and the event log.
5. A periodic heartbeat from each node confirms it is alive; a missed heartbeat raises a "node offline" flag on the dashboard.
6. The student commissions the floor with the included procedure: mount nodes, tune the activity threshold per machine type with the tap-test, and log a supervised workout to validate classification.

Tech stack:

• ESP32-class sensor nodes (one per machine)
• MPU6050 / vibration sensor modules
• Wi-Fi telemetry to gateway + cloud backend
• Web dashboard (HTML/CSS/JS demo included)
• Arduino IDE (C/C++ firmware)
• Li-ion battery packs with charging modules

• [Per-machine usage sensing] Compact wireless nodes detect real mechanical activity from vibration/motion signatures — a machine counts as used only when it is actually being worked, not when someone leans on it.
• [Live floor-status dashboard] Every instrumented machine shows IN USE or FREE in real time, so the dashboard doubles as a live floor map for staff and members.
• [Utilization analytics] Per-machine active minutes, utilization percentage of opening hours, and day-wise totals reveal which equipment earns its floor space.
• [Peak-hour patterns] An hourly usage view shows exactly when the floor saturates, supporting decisions on batch timings, staffing and new purchases.
• [Idle-vs-active classification] Firmware uses vibration thresholds with debounce timing so a bumped machine does not log a workout — the classification logic is documented and tunable.
• [Battery-friendly node design] Nodes sleep between checks and report on state change plus heartbeat, with the power budget analyzed in the report.
• [Working demo dashboard] A single-file web dashboard with realistic simulated floor telemetry is included for presentation and viva.

• Sensor node design (one node per machine in the demo set)
• Gateway firmware with aggregation logic
• Working single-file dashboard demo with simulated floor telemetry
• Mounting and threshold-tuning commissioning guide
• Project report PDF (theory, sensing, classification, dashboard, results)
• PPT presentation and viva Q&A document

• Classification is threshold-based on vibration energy, not rep-counting — it measures active minutes, not sets, reps or user identity.
• Thresholds must be tuned per machine type; a setting that works on a treadmill may misread a light dumbbell rack — the tuning procedure is mandatory, not optional.
• The dashboard demo ships with simulated telemetry; live data appears only after nodes are mounted and connected.
• Wi-Fi must cover the whole floor; concrete basements may need an access point reposition.
• Battery life depends on activity levels; the report gives the power budget honestly rather than promising a fixed number.
• It tracks equipment, not members — no cameras, no identity, no personal data by design.