We present ThermalSense, a method for dynamically detecting and predicting thermal comfort by using thermographic imaging to look for the physiological markers of vasodilation or vasoconstriction. We describe how ThermalSense can be used to infer how to control heating and cooling systems and reduce energy use while maintaining comfort.

We evaluate ThermalSense using a study involving thirty individuals over five weeks in an office building. Our study shows that, on around 40% of occasions, the HVAC system could have expended less energy to achieve comfort. It further demonstrates that thermographic imaging can be used to infer whether heating or cooling must be activated to maintain comfort, with an accuracy of 94-95%.