Optimal Charge Security Camera

Versatile & Intelligent security camera with energy-aware model selection.

Abstract

This project develops an intelligent security camera system that dynamically selects YOLO models based on battery level and clean energy availability. The system balances accuracy, efficiency, and environmental sustainability through three controller approaches: Custom (weighted scoring), Oracle (MILP optimization), and Benchmark (performance-at-all-costs). Using historical LDWP energy data and comprehensive simulation framework, we evaluate controller effectiveness across seasonal variations to optimize real-time decision making for battery-powered surveillance systems.

Contributions

Samyak Kakatur: GitHub Code, Controller Training + Neural Network, Simulation Pipeline, Result Graphs, Report

Jasper Lin: Battery + Energy modeling, Carbon Data Processing, Visuals, Slides, Report

Special Thanks: Prof. Mani Shivastava, Kang Yang