@phdthesis{Schmidt2012,
  author    = {Schmidt, Gerald},
  title     = {The Influence of Anticipation and Warnings on Collision Avoidance Behavior of Attentive Drivers},
  url       = {http://nbn-resolving.de/urn:nbn:de:bvb:20-opus-73789},
  school      = {Universit{\"a}t W{\"u}rzburg},
  year      = {2012},
  abstract  = {This thesis deals with collision avoidance. Focus is on the question of under which conditions collision avoidance works well for humans and if drivers can be supported by a Forward Collision Warning (FCW) System when they do not react appropriately. Forward Collision Warning systems work in a way that tries to focus the driver's attention in the direction of the hazard and evoke an avoidance reaction by some sort of alert (e.g., tone or light). Research on these warning systems generally focuses on inattention and distraction as the cause for crashes. If the driver is inattentive, the results of a crash are thought to be worse as the driver's reaction is belated or might not mitigate the crash at all. To ensure effectiveness in the worst case, most of the experiments studying FCW systems have been conducted with visually distracted drivers. Research on the cause and possible countermeasures for crashes of attentive drivers are hardly available, although crash databases and field operational test data show that 40-60\% of the drivers look at the forward scene shortly before they crash. Hence, only a few studies elaborated on ideas about the reasons for crashes with attentive drivers. On the basis of the literature, it is worked out that one reason for delayed avoidance behavior can be an incorrect allocation of attention. It is further elaborated that high level attention processes are strongly influenced by interpretation of the situation and the anticipation of future status. Therefore, it is hypothesized that alert drivers react later when they can not foresee a potential threat or even when they misinterpret the situation. If the lack of threat anticipation or incorrect anticipation is a reason for crashes, a FCW system could be a great help, when the FCW is easily comprehensible. It is hypothesized that a FCW can compensate for missing threat anticipation in the driver. The results of the experiments show that the level of threat anticipation has the largest influence on driver behavior in an imminent crash situation. The results further suggest that FCW systems - especially warnings of audible or haptic modality - can help attentive drivers who do not anticipate a threat or misinterpret a situation. The negative influence of missing or mislead threat anticipation on objective measures was small when the threat appeared suddenly. This is thought to be due to the visual appearance of the introduced threat. It is assumed that this type of stimulus triggers a lower level attentional process, as opposed to a top-down attention process controlled by an anticipatory process. In the other scenario types such a lower level process may not be triggered. An important result of the second study is that (Forward) Collision Warnings have to be learned. Participants with warnings reacted slower than participants without any FCW in the first critical event. Participants with a visual warning reacted particularly slow. Later in the experiment, the probands with warnings were constantly faster than their counterparts without them. Hence, the results of this study suggest that a haptic or audible modality should be used as a primary warning to the driver. The characteristic of visual warnings to draw the visual attention is both a blessing and a curse. It is suggested to use the visual warning component for only a short period of time to attract the driver's attention to the forward scene, but then end the display to not further distract him. Car manufacturers try to avoid as many unnecessary alarms as possible. If driver monitoring would be available, it is often planned to suppress warnings when the driver is looking through the windshield. The results suggest not to do so. If a driver reaches a critical situation represented by a low Time-to-collision (TTC) or a high need to decelerate, he should always get a warning, unless he is already braking or steering. The most important arguments for this are: - Looking at the street does not mean that the driver has the correct situational awareness. - The driver has to learn the meaning of the warning. - The driver will not be annoyed by a warning when the situation is considered critical.},
  subject      = {Zusammenstoss},
  language  = {en}
}