Multimodal Sensor Integration

The main focus of this study was how to combine the output of specific intelligent sensors, which are able to decide on their own if a situation is alarming or not. We presented a fusion approach that constructs a situational awareness from different sensors and uses a Bayesian network for final reasoning. We built a real-life prototype of an entry control point with several intelligent sensors such as fingerprint reader, wireless card reader, camera and electronic lock.

Relevant papers:

  • B. Kaluža, E. Dovgan, T. Tušar, M. Tambe, and M. Gams. A Probabilistic Risk Analysis for Multimodal Entry Control. Expert Systems with Applications, 2011. doi:10.1016/j.eswa.2010.11.065 [pdf]
  • E. Dovgan, B. Kaluža, T. Tušar, and M. Gams. Improving user verification by implementing an agent-based security system. Journal of Ambient Intelligence and Smart Environments, 2(1):21-30, 2010. [pdf]