Refine
Labor/Institute
Keywords
- ITS (Intelligent Transportation Systems) (4) (remove)
Document Type
- Article (4) (remove)
Language
- English (4)
Has Fulltext
- no (4)
This paper presents a novel application of artificial cognitive systems to traffic scene understanding and early recognition of highway maneuvers. This is achieved by use of Bayesian networks for knowledge representation, to mimic the human reasoning on situation analysis and to manage inherited uncertainties in the automotive domain, that requires efficient and effective analysis of high volume and frequency data streams. The maneuver recognition uses features, analyzing the observed vehicles behavior and available free space on the target lane. Dynamic Bayesian networks (DBN) capture the motion of the own and surrounding vehicles as a dynamic process, following the trend development of lateral motion features. The static and dynamic models for maneuver recognition are statistically evaluated with real highway driving data sequences. The DBN demonstrates earlier recognition (∼ 1.1 seconds) and higher accuracy (1.12% error rate) with more stable performance than the static models. The system is deployed on an experimental vehicle, where a Divide-and-conquer approach to inference in object-oriented Bayesian networks is introduced and its implementation is shown to require computation time of 0.15 milliseconds and ROM memory ∼ 10 2 kilobytes. The promising performance evaluation results are confirmed by test drives in real highway traffic.
Cooperative perception makes it possible – in addition to emergency warnings – to provide drivers with early advisory warnings about potentially dangerous driving situations. Based on research results pertaining to imminent crash warnings, it was expected that the effectiveness of such advisory warnings depends on situation-specific anticipations by the driver. During a simulator study, N = 20 drivers went through a wide range of longitudinal traffic and intersection scenarios. The scenarios varied in the possibility to anticipate traffic conflicts (Anticipation: high vs. low) and were completed under different visibility conditions (Visibility: obstructed vs. visible), with and without driver assistance based on cooperative perception (i.e., visual-auditory advisory warnings two seconds prior to the last-possible warning moment; assistance: no assistance vs. with assistance). The warning concept was based on empirical pre-studies and previously validated on a public test intersection. During non-assisted driving, critical situations were mainly experienced when the possibility to anticipate traffic conflicts was low. Visual obstructions lead to a further increase in the frequency of critical situations. Furthermore, the results indicate a clear mitigation of critical encounters when providing early advisory warnings which is independent from sight obstructions. This applies particularly to surprising and unexpected scenarios and thus illustrates the potential of cooperative perception to enhance active traffic safety.
This article introduces a novel approach towards the recognition of typical driving maneuvers in structured highway scenarios and shows some key benefits of traffic scene modeling with object-oriented Bayesian networks (OOBNs). The approach exploits the advantages of an introduced lane-related coordinate system together with individual occupancy schedule grids for all modeled vehicles. This combination allows an efficient classification of the existing vehicle-lane and vehicle-vehicle relations in traffic scenes and thus substantially improves the understanding of complex traffic scenes. Probabilities and variances within the network are propagated systematically which results in probabilistic sets of the modeled driving maneuvers. Using this generic approach, the network is able to classify a total of 27 driving maneuvers including merging and object following.
In this paper we introduce a novel approach towards the recognition of typical driving maneuvers in structured highway scenarios and identify some of the key benefits of traffic scene modeling with object-oriented Bayesian networks (OOBNs). The approach exploits the advantages of an introduced lane-related coordinate system together with individual occupancy grids for all vehicles. This combination allows for an efficient classification of the existing vehiclelane and vehicle-vehicle relations in a traffic scene and thus substantially improves the understanding of complex traffic scenes. We systematically propagate probabilities and variances within our network which results in probabilistic sets of the modeled driving maneuvers. Using this generic approach, we are able to classify a total of 27 driving maneuvers including merging and object following.