Ship the AI
Smart Elevator using Machine Learning Techniques
Designed and prototyped a contactless smart elevator system that predicts a user's destination floor using RFID-based location tracking and machine learning, aiming to minimize button presses and waiting time.
- client
- SASTRA University (Prototype)
- when
- 2019-2020
- headline
- Contactless floor prediction prototype
01 · Problem
Traditional elevators require manual floor selection, increasing touch points and wait times. ML-driven prediction using RFID and camera data enables contactless operation, faster service, and improved hygiene—critical for high-traffic buildings and post-pandemic environments.
02 · What I built
- Built hardware stack using Raspberry Pi, Arduino, NodeMCU (ESP8266), RFID readers, and cameras
- Implemented firmware and control logic for RFID tag reading and sensor input handling
- Collected movement data from RFID and camera streams to create user travel pattern dataset
- Applied image processing and ML models to predict destination floors based on user identity and history
- Developed central Raspberry Pi controller to fuse sensor data, run prediction model, and send elevator commands
- Achieved working lab-scale prototype demonstrating automatic floor selection and contactless operation
03 · Result
- Working prototype with automatic floor prediction
- Contactless elevator operation using RFID and ML
- Reduced button presses and wait times in lab testing
- Foundation for building-scale deployment