Environmental Perception for Automated Vehicles

This thesis addresses the issues of the environmental perception for automated vehicles. The environment which the vehicle has to perceive can be generally classified into two groups: stationary and movable objects. The perception is to represent and understand these environments by organizing, identifying, and interpreting sensory information. Firstly, the sensory information is obtained based on the vehicle pose (position and orientation), and therefore the environmental perception necessarily includes a localization task. Then, the stationary objects are generally represented with a map, and the movable objects are tracked to represent their temporal behaviors. These environmental perception tasks can be summarized into two well known technical problems: simultaneous localization and mapping (SLAM), and detection and tracking of movable objects (DATMO). In this thesis, a novel framework to solve these problems by using Velodyne laser scanner is presented.