Overview of CI research in automotive ITS

Abstract

Illustrative examples of applications of the CI methods in ITS are overviewed. They include off-board and on-board applications. On-board object recognition is further discussed from the standpoint of making a more reliable system for ITS applications. I. INTRODUCTIONComputational intelligence (CI) traditionally includes neural networks (NN), fuzzy logic systems (FS) and evolutionary computations (EC) as three main areas distinguishing CI in terms of employed methods from classical AI. The difference between CI and AI is blurred because both the IEEE Computational Intelligence Society and the Association for the Advancement of Artificial Intelligence have sometimes been employing essentially the same methods to attack same challenges.Instead of categorizing by methods, the division by application is accepted which is more useful for applied research. In this paper ITS applications of CI are separated in two large categories:1) On-board: inward (driver or vehicle) focused and outward (environment) focused, 2) Off-board: infrastructure based including communications and Vehicle-Infrastructure Integration (VII).The on-board applications are exemplified in [1], [2], which include but are not limited to driver behavior assessment (drowsiness, intentions, etc.), object recognition, vehicle health monitoring and intelligent control.The infrastructure applications are exemplified in [3], [4], [5], which include but are not limited to remote counting of road objects, recognizing license plates and other vehicle properties.In this paper traditional CI applications in ITS still in the research phase are illustrated. The state of art in object recognition is then summarized. It is also discussed what combining multiple sensors, or sensor fusion, could bring to the table and what perhaps is not so important for the ITS area. This paper is organized as follows. The next section briefly overviews illustrative off-board applications of CI in ITS and system architectures. Section III briefly overviews on-board applications. Section IV discusses in more details object recognition applications and their specifics for ITS, as well as overviews some experimental results, followed by Conclusion.