Image Recognition System for Automated Lighting Retrofit Assessment

Abstract

Buildings are responsible for approximately 40% of all US energy use and carbon emissions. Lighting technologies continue to evolve, leading to potential energy savings through retrofits of lighting systems. Building lighting systems is typically the first item evaluated by commercial and industrial energy auditors. This paper presents the first phase of a project to develop unmanned aerial and ground vehicles capable of conducting autonomous energy audits of commercial buildings. The paper presents a prototype system that can enumerate and classify the lighting in a building using an optical camera, accelerometer, spectrometer, and distance sensor. As the aerial vehicle navigates throughout a room, the prototype system captures images and collects frequency data of lighting. The system employs image recognition techniques to quantify lighting in each room. Using the unique frequency spectrum of each lighting type, the prototype system classifies the different types of lighting with the spectrometer. An accompanying software program then analyzes the quantity and type of lighting to recommend economical alternatives, or lighting retrofits.