The proposed people detection and tracking method is based on a multi-modal sensor fusion approach that utilizes 2D laser range and camera data. The data points in the laser scans are clustered and classified using an SVM based AdaBoost classifier trained with a set of geometrical features of these clusters. Image detection is obtained using an extension of Implicit Shape Model (ISM) that learns an appearance codebook of local descriptors from a set of hand-labeled images of pedestrians and then votes for centers of detected people. Each detected person is then tracked using an advanced multiple motion model tracker method. Qualitative and quantitative results are shown. Pedestrians are a particular class of obstacles whose configuration information is treated separately for a type of autonomous vehicle navigation where the pedestrians have a higher priority of avoidance than other generic environment obstacles. The autonomous navigation scheme presented depends on the structure of the environment as well as the form of global connectivity information. Prediction of future motion of the moving obstacles is taken into account in order to generate a feasible set of vehicle trajectories. Qualitative results are shown in simulation and experimental setup.
Autonomous vehicle navigation and pedestrian detection in urban environments
Autonome Fahrzeugnavigation und Fußgängererkennung im städtischen Bereich
2009
10 Seiten, 8 Bilder, 18 Quellen
Conference paper
English
Autonomous navigation for mobile service robots in urban pedestrian environments
BASE | 2011
|Locating Autonomous Vehicle in Urban Environments
British Library Conference Proceedings | 2016
|