Modern robotic systems incorporate more and more sensors and as the number of sensors rises, it becomes difficult to use the classic approaches of sensor processing. The bulk of existing methods are designed to work with few sensor values and to integrate these values into a simplified system model. In the special case of image processing, which already deals with a high number of (pixel-)sensors, only specialised solutions for narrow problems exist. In order to fully utilise the growing number of sensors on modern robotic systems, new approaches for massive multi-sensor processing have to be devolped. Such an approach could be inspired by neuroscience, as the human brain seems to handle the concurrent input of millions of sensors quite well. Even more surprising, the brain, in principle, seems to use just one general mechanism to process the information of the different senses. A first approach inspired by these properties of the brain will be presented in this talk.
Vortragsdetails
First steps towards universal multi-sensor processing
Veranstaltungsort
Raum Seminarraum 117, Robert-Hooke-Str. 5 in BremenIn der Regel sind die Vorträge Teil von Lehrveranstaltungsreihen der Universität Bremen und nicht frei zugänglich. Bei Interesse wird um Rücksprache mit dem Sekretariat unter sek-ric(at)dfki.de gebeten.
