Tracking and classification of overtaking vehicles on Autobahnen
- 24 August 2005
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 314-319
- https://doi.org/10.1109/ivs.1994.639535
Abstract
An object recognition module is presented, which is part of a multifocal computer vision system for autonomously driving on Autobahnen that was developed at the Universitat der Bundeswehr Munchen. Based on the 4D approach developed at UniBwM the system functionalities include road following and obstacle avoidance. Obstacle detection as implemented actually is capable of dealing with objects under aspect conditions where the objects can be described by a 2D shape model. This assumption is not valid for objects in the closer environment of the ego-vehicle just when the situation is most critical. Filling this gap is the intention of the module described here. It attempts to recognise the overtaking vehicles as these are approaching the ego-vehicle from behind and track them until they passed the ego-vehicle. In order to master this task the objects are described by a 3D shape model as polyhedron. After instantiating different object hypotheses these are tested and verified in parallel by spatio-temporal reasoning techniques in conjunction with domain dependent fuzzy knowledge.Keywords
This publication has 5 references indexed in Scilit:
- A camera platform for intelligent vehiclesPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2005
- Multiple object recognition and scene interpretation for autonomous road vehicle guidancePublished by Institute of Electrical and Electronics Engineers (IEEE) ,2005
- Road recognition from multifocal visionPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2005
- Simultaneous estimation of 3D shape and motion of objects by computer visionPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2002
- “Neural” computation of decisions in optimization problemsBiological Cybernetics, 1985