The perception and the interpretation of the vehicle's surrounding are the core modules in any Advanced Driving Assistance Systems and the first two levels in any autonomous driving vehicle. The task of the perception level is to provide objects assessment (objects' position, form, classification, orientation, speed) using the primary sensorial information. The task of the interpretation level is to use the objects assessment in order to achieve situation assessment, i.e. identification of the

more »

... nstantaneous relations between the traffic environment elements as well as their evolution in time. The work presented in this paper lies within the interpretation layer, aiming to identify the ego-vehicle travelling lane by analyzing the relationships between the ego-vehicle and the surrounding traffic objects, which are detected by a stereovision based perception system. The proposed solution is to identify the ego-vehicle lane by matching the visually detected lane landmarks with the corresponding map landmarks, available through an original Extended Digital Map solution. Additionally, the visually detected vehicles are used in the lane identification process. Due to the proposed approach for lane identification, this task will be referred to as lane assessment in this paper. The used solution for the lane assessment is a probabilistic one, in the form of a Bayesian network, which uses as evidence the visual cues provided by the stereovision perception system. In order to incorporate the dynamic nature of the problem, the temporal dimension was introduced in the network, by proposing a dynamic Bayesian model. Experimental results illustrate both the efficiency of the proposed method even in complex situations, and also the improved efficiency of the dynamic approach over the static one.