“Game over” for autonomous shuttles in mixed traffic? Results from field surveys among pedestrians and cyclists on how they interact with autonomous shuttles in real-life traffic in Norway

Bjørnskau, Torkel; Aasvik, Ole; De Ceunynck, Tim; Fyhri, Aslak; Hagenzieker, Marjan; Johnsson, Carl; Laureshyn, Aliaksei · 2023 · Crossref

DOI: 10.1016/j.trip.2023.100781

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study investigates the interaction dynamics between autonomous vehicle (AV) shuttles and vulnerable road users—specifically pedestrians and cyclists—in mixed traffic environments. Motivated by game-theoretic predictions that AVs will be severely obstructed or "bullied" by human road users who exploit the vehicles' defensive programming, the research tests whether these theoretical outcomes manifest in real-world settings. The authors hypothesize that interactions evolve through three phases: initial caution, followed by aggressive exploitation of the AV’s yielding behavior, and potentially a later phase of social consideration. The research employs field surveys conducted during two pilot projects in Norway: one in Kongsberg (2018–2019) and one in Oslo (2019). Data were collected via tablet-based interviews with pedestrians and cyclists along the shuttle routes at multiple time points, including before shuttle deployment and at intervals during operation. The Kongsberg pilot involved EasyMile EZ10 shuttles on a route including pedestrian streets, while the Oslo pilot used Navya Arma shuttles on a waterfront route with shared spaces and cycle lanes. The study excluded car drivers due to practical interview constraints and low motor traffic volume. Statistical analysis using SPSS examined changes in self-reported behavior and perceptions over time, focusing on yielding behaviors, crossing decisions, and overtaking frequency. The findings indicate that road users increasingly exploited the AV shuttles’ defensive driving style. Survey results showed that as familiarity with the shuttles grew, pedestrians and cyclists reported crossing in front of the shuttles more frequently, even when the shuttle had the right of way or outside designated crossings. Respondents acknowledged knowing the shuttle would stop, which encouraged them to proceed first. In Oslo, cyclists frequently overtook the shuttles when riding behind them. The data supported the game-theoretic hypothesis that human road users adapt to the predictable, non-competitive behavior of AVs by prioritizing their own flow, effectively treating the AV as a weak actor that will yield. There was little evidence of the predicted third phase where social norms might encourage more considerate behavior toward the AVs within the study period. The significance of this research lies in its empirical validation of game-theoretic models regarding AV integration. It suggests that without changes to AV decision-making algorithms or traffic infrastructure, autonomous shuttles may face severe operational delays and reduced efficiency in mixed traffic due to human exploitation of their safety protocols. The study highlights the critical need for AVs to communicate intentions more effectively or adopt more assertive driving strategies to maintain traffic flow, challenging the assumption that purely defensive driving ensures successful integration into existing road networks.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• driver vru interaction
• acceptance adoption
• ehmi external hmi