Dr. Christina Hirschl is head of research division Sensor Systems at Silicon Austrian Labs Gmbh (SAL), an Austrian industry-oriented research and development center for electronic based systems. Acting as link between science and industry SAL uses latest research results achieved in the area of sensors, RF and power electronics to make products, technologies and processes simpler, safer and more effective. She studied physics at the University of Vienna and during her PhD worked at the investigation of chaotic properties of fluids. After the PhD she moved to industry and worked for Ruag Space as thermal engineer for several years. In 2010 she became a project manager for renewable energy projects at CTR Carinthian Tech Research AG, 2014 area manager for Smart Systems and 2019 head of research division Sensor Systems.
Technical Vision Talk: “Combining hardware and data science to change our tomorrows world – research of electronic based systems”
(joint work of Christina Hirschl & Federico Pittino)
Due to the ever increasing availability of cheap and low-power electronic hardware, in recent years it has become possible to deploy large sensors networks for the fine monitoring of environments, machines and all kinds of complex systems. The optimization of such sensors networks is, however, of paramount importance, in order to reduce the cost and energy consumption, by minimizing the number of deployed sensors and the amount of transferred data. The optimization can be performed using Data Science, from the individual sensors to the network levels. Especially examples in the field of renewable energy, like the monitoring of PV power plants or the indoor monitoring for better air quality for examples show these possibilities.
Optimization of transferred data can be achieved at the sensor level by the means of Embedded AI. These techniques involve the processing of raw data on the individual sensors’ hardware, thereby allowing to transfer only the relevant information. Some of the most prominent applications are in the realm of Computer Vision, in particular embedded cameras, but similar techniques can be applied to any kind of sensor.
Another method for reducing the transferred data involves Virtual Sensing techniques. Such techniques allow, on the one hand, to reduce the number of deployed sensors by virtualizing some of them, i.e., estimating the measurement that would be obtained by this subset of sensors from the reading of all the other sensors. On the other hand, Virtual Sensing allows the replacement of expensive sensors with cheaper ones or the monitoring of remotely accessible locations, by deriving models on the data collected by the available sensors.
These techniques derived from Data Science have the potential for enabling an efficient and accurate deployment of sensor networks, integrating their hardware characteristics with accurate models derived on large amounts of collected data.
__________________
Thu. May 19 | 9:30 am– Technical Vision Talk: “Combining hardware and data science to change our tomorrows world – research of electronic based systems”