The development of new multispectral camera sensors is in the focus of the new research project “MISEL” (Multispectral Intelligent Vision System with embedded Low-Power Neural Computing), that the Chair of smart sensor systems at Bergische Universität is involved in. The research team from Wuppertal, led by Prof. Dr. Daniel Neumaier, and his international colleagues will receive a research grant of 4.96 Million Euro from the European Union (Horizon 2020), 203,000 Euro of which will go to the Bergische Universität Wuppertal.
For an energy-efficient and fast pattern recognition, e.g. for the automatic differentiation between drones and birds, neuromorphic components are to be implemented into the camera sensors. “These are components that are inspired by the neurobiological structures of the nervous system”, explains Prof. Neumaier. This makes it possible to train a chip to recognize different patterns through targeted learning processes. Possible areas of application for these camera-sensors are aviation, autonomous driving, autonomous robots or production technology.
The camera-sensors are initially based on classic Silicon-CMOS-electronics. These are extended by newly developed neuromorphic elements as well as graphene-based photodetectors. “For project MISEL the research team in Wuppertal will work on the development and characterization of the graphene-based photodetectors for visible and infrared light“, says Prof. Neumaier. The goal of this activity is to increase the spectral range as much as possible using one technology platform.
The project involves nine European partners, with VTT, the Technical Research Centre of Finland, as project-coordinator. Other members of the consortium are: AMO GmbH (Aachen), Bergische Universität Wuppertal, Fraunhofer Research division Center for Applied Nanotechnology (Hamburg), Kovilta Oy (Finland), Lund University (Sweden), Lodz University of Technology (Poland), University of Santiago de Compostela (Spain) as well as the Laboratoire national de métrologie et d’Essais (France).