Occupant thermal comfort is a pillar of smart buildings. However, the research community lacks a sufficient updated dataset and often has to apply the obsolete ones with inaccurate comfort modeling performance. In this paper, we present our latest urban tropical dataset for thermal comfort research. We follow a non-intrusive approach by using the emerging wearables to collect the data about the occupant profile, vital signs, thermal sensation, and ambient environment. In total, we obtain a unique dataset with 8,972 data items in two years. Our dataset will be publicly available and owns the high potential to propel the smart building researches like thermal comfort modeling and energy management.