You are here

June 28, 2013: Best Paper Award at SEIT 2013 for ACTLab/TUE team.

June 28, 2013: Luis Lopera, Marc Troost and Oliver Amft received the best paper award for the manuscript entitled: "Using a thermopile matrix sensor to recognize energy-related activities in offices" at the 3rd International conference on Sustainable Energy Information Technology (SEIT 2013). The work presented in this paper discusses an approach to recognise user profiles related to energy consumption using a thermopile matrix sensor. The work is related to the EU GreenerBuildings project, coordinated by ACTLab.

Full citation information:

Lopera Gonzalez, L. I.; Troost, M. & Amft, O.: Using a thermopile matrix sensor to recognize energy-related activities in offices.
SEIT 2013: Proceedings of the 3rd International Conference on Sustainable Energy Information Technology, Elsevier.

Abstract: Various installations and appliances used by building occupants are manually operated, including office devices, kitchen appliances, washing basins, etc. By monitoring appliances usage and thus energy consumption, office occupants could received feedback on their energy needs, which is considered vital to spur energy conservation.  In this work, we investigate a novel generation of 2D-matrix thermopile sensors for recognising objects and object-occupant interactions from their heat patterns for a total of 21 activities using a single sensor installation. The activities were chosen according to their relevance for appliance energy consumption. We present a processing concept adapted for thermopile matrix sensors to detect and track objects. Furthermore, detected objects were classified according to object state and occupant interaction categories.  In scripted and real-life datasets using a ceiling mounted matrix sensor, we demonstrate that a single sensor installation can provide information on various activities, rather than instrumenting many devices and appliances with individual sensors. We show that activities with a clear thermal signature can be recognized with more than 96% accuracy. We also show experimental results for activities that have a thermal signature closer to the ambient temperature.

Details and full text: GoogleScholar