Thermal comfort measurement using thermal-depth images for robotic monitoring

Abstract This paper describes an application of thermal-depth images to human thermal comfort measurement. A mobile monitoring of the elderly and residents of care houses is one of the promising applications of mobile assistive robots. Monitoring if a person feels comfortable is an important task of such robots. We rely on an established comfort measure in the architecture domain, namely, predicted mean vote (PMV). PMV is calculated mainly by six factors and one of which is the clothing insulation or clo-value. Clo-values are usually measured by a thermal mannequin, a specially-designed apparatus for the purpose. We apply human recognition techniques in thermal-depth images to efficiently measure clo-values, thereby enabling on-line assessment of thermal comfort. We evaluate the method and develop a mobile robot system for experimental testing.