SolarGuard

Team Members:

Lee Hon Lam李翰林 (MAEGN/2)

Yip Tang Chun 葉騰駿 (FTECN/2)

Chan Ming Fung 陳銘峰 (CHLL/1)

Abstract:

"SolarGuard" is an AI vision module for solar panels, based on Yahboom K230, enabling detection, positioning, and data communication without retraining. First phase completes R&D; subsequent phases will enhance adaptability and enable advanced applications like robotic arms and UAVs, supporting SDGs 7 and 12.

Project Progress

The team moved from an initial drone concept to a modular AI Edge-Vision Module and filed a Chinese invention patent, then completed a full closed-loop pipeline on the K230 platform—covering image capture, PV-panel detection, anomaly recognition, screenshot archiving and report generation—all running locally without cloud or depth sensors. They further developed a monocular pixel-ratio distance algorithm with jitter filtering and real-time offset computation for auto-targeting, and progressed from sketch to a physical prototype featuring PS2-controlled Mecanum-wheel movement with the K230 camera integrated into the mainboard. This culminated in a successfully validated hybrid-control workflow: PS2 movement → mode switching → panel recognition → camera-guided alignment → automatic approach. On the commercial side, they consolidated their model into three business modules, began ESG report R&D, and studied smart-socket and electrical safety systems, while also gaining external recognition through public outreach (100+ visitors), a college sharing session, an AI-generated promotional video, and an interview for the Hong Kong ICT Awards 2026.