Detecting and Signalling Potentially Unnoticed Objects

The project aims at detecting and signalling potentially unnoticed objects in the vehicle sur-round for a driver’s assistance scenario. To support such function results derived from proc-essing two different sensor streams need to be combined. First, the gross view direction of the driver inside the vehicle needs to be monitored on a coarse scale, and, second, the car environment will be analyzed for salient events such as, e.g., detecting the presence of inde-pendently moving objects. The approach should enable the characterization of suspicious features and objects in the environment based on feature-specific and spatial attention mechanisms. Also different objects (driver inside and objects outside the vehicle) need to be tracked over time. The combination of both information streams will be analyzed to supply an alarm signal in the event of unmatched view direction and position of salient object appear-ance in the vehicle’s environment. Such an information fusion will help in the future to supply proper information displays of peripheral information utilizing proper device technology (tan-gible interfaces) and sensory channels (sonification of visual data).