The project aims to develop a system that is capable of gathering, fusing, interpreting and producing information about parts of a risky industrial environment through recognition and description of objects and activities. The proposed behavior analysis surveillance system is able to monitor human activities in high-risk areas and to deduce about the normality of the situations that individuals are getting involved. The representation and interpretation of behaviors is possible through a harmonic combination of low-level features (geometric), which will be extracted, from visual data and high-level knowledge data (semantics).
In order to fulfill this aim, a wireless sensor network based on Ultra WideBand (UWB) technology has been developed. Two kinds of network nodes have been deployed: a) nodes equipped with cameras and b) nodes with range scanners. Cameras can be deployed at high-risk areas and activated to gather video data as long as people exist in this area. Afterwards, the local sensor process unit process the data and then it provide them to a central processing station which integrate information from various areas, in order to decide in real time, when this individual acts illegally and poses a threat or is involved in an accident.
Range scanners are used in the assembly lines to recognize the objects lying on them. The acquired range data can be processed in real-time in order to verify if the product that crosses the assembly line is the one that was supposed to be.