All projects
Capstone8 months · 2025–2026
Automated Rotating Cat Litter Box
Designed and validated a multi-sensor embedded system with calibration, noise analysis, and threshold tuning to achieve reliable real-time detection.

Role
Embedded Systems & Test Lead
Duration
8 months · 2025–2026
Team
5 Students (Mechanical & Mechatronics)
/Problem
Existing automated litter boxes either misfire (rotating while occupied) or fail to detect smaller animals. The system required a deterministic detection stack robust to vibration, lighting, and litter dust — without relying on expensive vision systems.
/Approach
- 01Specified a multi-sensor fusion stack: HX711 load cells for mass, an IR break-beam array for occupancy, and Hall-effect sensors for drum position.
- 02Implemented an ESP32-based firmware state machine (IDLE → OCCUPIED → SETTLE → CLEAN → DUMP) with debounced transitions and watchdog safety timers.
- 03Developed a calibration procedure using reference weights and statistical noise characterization (σ, drift, hysteresis) for each sensor.
- 04Tuned detection thresholds with logged real-world data to achieve high true-positive and near-zero false-trigger rates.
/Results
- False-trigger rate driven to <1% across 200+ test cycles.
- Detection latency < 400 ms from entry to OCCUPIED state.
- Achieved BOM within target budget; design documented for manufacturability.