Reinforcement Learning for Energy Optimization in Smart Home Automation
Project Overview
Developed an AI-powered energy optimization system for my thesis titled “Optimizing Energy Efficiency in Smart Home Automation through Reinforcement Learning and IoT”. The goal was to address the limitations of conventional automation by designing a smart system that dynamically adapted to user behavior.
Technologies Used
- Programming Languages: Python, C
- Frameworks & Libraries: Flask, TensorFlow
- Hardware: ATmega328P Microcontrollers, PIR Motion Sensors, Temperature Sensors
- Algorithms: Reinforcement Learning, Penalty-Based Feedback Mechanism
Key Challenges
- Overcame the challenge of ensuring reliable real-time data collection with ATmega328P microcontrollers, which initially caused delays in data processing for the reinforcement learning model.
- Faced difficulties in aligning the AI model with unpredictable user overrides that led to inefficient energy use; resolved this by implementing a penalty-based feedback mechanism to enhance the model’s adaptability.
Results
- Achieved a 5% reduction in energy consumption compared to conventional automation methods.
- The project was published in the Asian Journal of Research in Computer Science, showcasing its real-world impact and contribution to energy-efficient smart homes.