‹
Back
09:00 - 10:30
Parallel sessions 7
›
09:00 - 10:30
Room: HSZ - 7E02
Chair/s:
Arnd Engeln
Although the amount of traffic deaths in Germany decreases since the early 70s, still nowadays, more than 7 persons lose their lives, and more than 1000 persons get injured in road traffic in average each day. The main reasons are seen in human behavior errors as there are distraction, disregard of priority and inappropriate distances or speeding. That for, the symposium focuses research on human behavior related to traffic safety. The talks focus on research on pedestrians’ prediction of other road users’ behavior, the perspective of different road user groups on cyclists’ behavior as well as the impact of technical innovations on individual behavior – as there are camera-monitor systems in cars and sound reduced cars on the street.

With a literature review on existing UX models to inform the measurement of acceptance of automated driving, we then transition to the topic of “Automated driving - acceptance and interaction in traffic”:
In the talk session immediately following we will present current research on automated driving. The integration of automated cars into mixed traffic with manual motor vehicle drivers, pedestrians and cyclists generates open questions on how to communicate and interact with them. Solutions should ensure traffic safety as well as acceptance by the traffic participants involved. If acceptance is low, the improvement of traffic safety may fail.
The first two talks address research on the impact of external communication on the behavior of manual drivers as well as on other road users. In the following talk the user of the automated car and how to avoid usage errors is focused on.
Submission 563
Cautious at the Curb: Conservative Bias in Pedestrian Interpretation of Vehicle Yielding
SymposiumTalk-01
Presented by: Daniel Eisele
Daniel Eisele 1, Anja Katharina Huemer 1, Tibor Petzoldt 2
1 Bundeswehr University Munich, Germany
2 Dresden University of Technology, Germany
Understanding whether vehicles will yield is critical for safe and smooth interactions with both human-driven and automated vehicles (AVs). This study explored (a) how accurately pedestrians can identify whether vehicles will yield (b) the influence of road design, vehicle motion, vehicle type, and explicit communication on accuracy, and (c) the correspondence between subjective certainty and actual accuracy.

Participants viewed short video clips depicting approaching vehicles that either yielded or maintained their speed. The vehicles varied in type and explicit communication: a conventional vehicle without eHMIs, an AV with a mode eHMI indicating automation, or an AV additionally indicating whether the vehicle is currently decelerating. Yielding behavior and road design were also manipulated. After each clip, participants judged whether the vehicle would yield and indicated their certainty.

Participants were accurate overall (86%). They exhibited a (presumably adaptive) conservative bias: false negatives (falsely assuming the vehicle would not yield) far outnumbered false alarms (falsely assuming the vehicle would yield). Mixed-effects analyses showed that accuracy was shaped primarily by braking behavior and road design, while explicit communication improved recognition only when vehicles truly intended to yield, leaving false alarm rates unaffected. Certainty was generally high but varied with cue clarity.

The uncovered interactions underscore the need for integrated vehicle and road design to jointly support accurate and confident intent recognition.