Amsterdam-based Quby has received three patents for the next generation of its Toon smart thermostat (link in Dutch). The patents pertain to two novel hardware features and to the data science approach for identifying appliance-specific energy usage. They have been awarded to the subsidiary of utility company Eneco by the Netherlands Enterprise Agency (RVO).
Toon 2 features a now patented solution for dealing with disturbances in the temperature readings, caused by the heat generated by the electronic components in the device itself. The new design allows for more accurate measurements while using less computing power. Toon 2’s ultrasonic movement detection has also been patented. Using this functionality, the display can detect if there are people nearby and how close they are, without actually ‘listening’ in on what is being said in the home – something other electronic assistants have become infamous for.
To assist consumers in reducing energy waste, Toon 2 offers a waste checker service. This has been built on a method that uses machine learning at scale to identify every device within a household: the power consumption patterns of a large number of appliance types were introduced as ‘fingerprints’ to a self-learning algorithm, along with data on efficient and inefficient usage. As a result, the algorithm is now able to recognize about 681,000 appliance cycles a day. The measurements are translated into personalized advice. The method invented by Quby’s data scientists Kaustav Basu and Stephen Galsworthy has been patented by RVO as well.