MEDUSA: Multi Sensor Data Fusion Grid for Urban Situational Awareness, EDA. The MEDUSA project studies a combination of different types of sensors supporting operations within urban environments. The essential improvement that the MEDUSA project is implementing is the provision of a means to address major existing capability shortfalls: (i) in the fusion of data from multiple diverse types of sensors and (ii) in the data representation, by means of a consolidated and integrated view, including overlaying across 3D GIS (Geographical Information System) models to facilitate decision-making by commanders of control and support operations.
The MEDUSA project analyzed a diverse cross-section of existing imaging sensor technologies including those in the field of video, with specific reference to images on visible and infrared bands. The fusion of the sensor data in particular with Digital Terrain Elevation information provided by the GIS is currently an acknowledged key technology/capability shortfall and thus a salient benefit of the MEDUSA project outcome – both for increasing overall knowledge in the domain and as a means of accelerating downstream exploitation of enhanced approaches. Another group of sensors that is considered for data fusion in generating an Operational Ground Picture, are sensors in very different categories, inclusive of acoustic sensors, motion detectors, chemical sniffers, through the wall radars and the like.
MEDUSA was funded by the European Defense Agency (EDA).
The project's final review was successfully held between 9-10 February 2012 in the premises of Bundeswehr University (UNIBW) in Munich, Germany.
Multimedia Group's contribution to MEDUSA was two-fold, in visual processing algorithms and in the High Level Fusion (HLF) engine of MEDUSA with the development of the reasoning engine. Multimedia Group researched and developed real time algorithms that work on streaming video data for smoke detection and unusual motion detection (see them in action in the videos below). These algorithms were successfully integrated in the MEDUSA sensor fusion grid and were used during the final demonstration. Additionally, the Group developed the HLF engine which collected data from all sensor processing modules and reason on them to infer about the threat level of situations. The development of the engine was based on Semantic Web technologies.
|Smoke detection from 00:12||Unusual motion detection in 00:14|
During the project, the following publications were produced by the Group:
Articles about MEDUSA and CERTH/ITI