Sensorial Perception, Modeling and Representation of the World Model for Driving Assistance Systems

The goal of driving assistance systems is to assist the driver in the driving process, in order to increase the traffic security and relieve the driver from a series of repetitive and boring activities by warning the driver and even taking direct control over of the vehicle. Currently the following systems are researched and experimented: longitudinal and lateral control systems, driving lane following systems, lane departure warning systems, automatic cruise control, pedestrian protection systems, intersection management systems etc.


The research carried out during this project is multidisciplinary and covers various domains, from sensorial processing, a priori and in situ knowledge representation, situation evaluators, behavior generation to effective navigation control with the goal of implementing complex functions specific to autonomous behavior, such as: perception, knowledge, imagination, reasoning. The multitude of existing approaches and solutions cannot be integrated into a generic solution because of the lack of a unified sensorial and a priori and in situ knowledge representation platform.


The unified platform will deal with the problematic of sensorial perception, road estimation and modeling, generic obstacle estimation, modeling and classification and also of building a world model representation for the driving assistance systems


The main research topics will focus on the research and development of new techniques of modeling for all the entities, their evolution and of the relationships between them, of new techniques of model parameters estimation from incomplete, noise corrupted and sequentially acquired data, of robust meta-classifiers for real-time obstacle identification and of efficient world representation methods.