Interact or counteract? Behavioural observation of interactions between vulnerable road users and autonomous shuttles in Oslo, Norway

De Ceunynck, Tim; Pelssers, Brecht; Bjørnskau, Torkel; Aasvik, Ole; Fyhri, Aslak; Laureshyn, Aliaksei; Johnsson, Carl; Hagenzieker, Marjan; Martensen, Heike · 2022 · OpenAlex-citations

DOI: 10.55329/fbhr3456

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 behavioral interactions between vulnerable road users (VRUs)—specifically pedestrians and cyclists—and automated shuttles in real-world traffic conditions. Motivated by the lack of empirical research on how humans negotiate with autonomous vehicles in mixed traffic, the authors sought to determine if VRUs exhibit "bullying" or aggressive behavior toward the shuttles’ defensive driving style, as predicted by game-theoretic models. The research also examined whether interaction patterns changed over time as road users became familiar with the technology. The methodology involved a field experiment in Oslo, Norway, where two SAE Level 3 automated shuttles operated on a 1.2 km route with varying traffic environments, including intersections and shared spaces. Video observations were conducted at five fixed locations from May to November 2019. Researchers used a predefined codebook to quantitatively analyze thousands of interactions captured by temporary cameras. The study focused on yielding behaviors at crossings and intersections, as well as overtaking maneuvers in same-direction scenarios. The results revealed that automated shuttles frequently failed to adhere to traffic rules. At pedestrian crossings, shuttles failed to yield to pedestrians in 26% to 50% of interactions, depending on the location. At a right-turn intersection, shuttles failed to yield to cyclists going straight in 38% of cases, with yielding performance significantly worsening over time. In same-direction interactions, cyclists predominantly overtook the slow-moving shuttles, usually on the left side. Contrary to game-theoretic predictions, the study found little evidence of VRUs intentionally bullying or exploiting the shuttles’ defensive nature. Most interactions were cooperative, with pedestrians and cyclists generally yielding or overtaking without aggressive intent. Furthermore, there was very little indication of temporal effects suggesting that interaction patterns evolved significantly over the observation period. The significance of this research lies in its challenge to the assumption that human road users will inevitably exploit the predictable, defensive behavior of autonomous vehicles. Instead, the findings suggest that VRUs generally interact cooperatively, though the shuttles themselves pose safety risks due to inconsistent rule compliance. The study highlights the importance of real-world field experiments over simulations for understanding complex traffic negotiations. It implies that while the "bullying" hypothesis may not hold true in practice, the technical limitations of current automated shuttles in yielding scenarios remain a critical safety concern that requires optimization before wider deployment.

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
• automation surprise