Video observation of encounters between the automated shuttles and other traffic participants along an approach to right-hand priority T-intersection

Pokorny, Petr; Skender, Belma; Bjørnskau, Torkel; Hagenzieker, Marjan P. · 2021 · OpenAlex-citations

DOI: 10.1186/s12544-021-00518-x

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 safety and behavioral interactions between automated shuttles and other traffic participants in regular traffic conditions, addressing a critical gap in observational research on autonomous vehicles. As automated shuttle deployments increase globally, understanding their interactions with vulnerable road users (VRUs) is essential for safety. The researchers focused on encounters along an approach to a right-hand priority T-intersection in Oslo, Norway, where Navya Arma shuttles (SAE Level 3) operate. The study specifically examined three types of encounters: interactions with cyclists or e-scooterists in parallel bicycle lanes, interactions with pedestrians on a raised zebra crossing, and overtaking maneuvers by motorized vehicles behind the shuttle. The methodology relied on external video recordings collected over 20 days, totaling 220 hours of footage. The site featured a one-directional bicycle lane and a zebra crossing. Researchers used RUBA software to detect shuttle presence and T-Analyst software, calibrated with aerial imagery, to measure numerical variables such as speed, acceleration, and position. Manual observation supplemented these measurements to assess behavioral patterns and rule compliance. The analysis covered 318 unique shuttle maneuvers, from which 83 specific encounters with other traffic participants were identified and analyzed. Reference speeds for uninfluenced shuttles, passenger cars, and cyclists were established to contextualize the shuttle's behavior during encounters. The results revealed distinct behavioral patterns and risks. In encounters with VRUs in the bicycle lane (60 cases), the shuttle often reacted with "hard stops" (17 cases) or slowing without stopping (22 cases). Hard stops, characterized by decelerations of approximately −1.61 m/s², occurred when VRUs behaved unpredictably or rode close to the shuttle. These abrupt stops pose risks for passenger safety and potential rear-end collisions. Two "stalemate" situations occurred where both the shuttle and VRU stopped, requiring manual intervention by the onboard operator. In pedestrian encounters (4 cases), the shuttle failed to yield correctly in two instances, forcing pedestrians to adjust their behavior. Regarding overtaking (19 cases), 16 vehicles overtook the shuttle by crossing into the opposite lane, exploiting the shuttle’s defensive driving style and slow speed. The study concludes that introducing automated shuttles into mixed traffic creates new interaction dynamics and safety challenges. The shuttles' strict adherence to rules and defensive programming can lead to misuse by other road users, such as overtaking, and cause frequent hard stops that endanger passengers and increase rear-end collision risks. Furthermore, the shuttles struggled with complex situations involving unpredictable VRU behavior, sometimes requiring human takeover. These findings highlight the need for improved intention estimation algorithms and better integration strategies to manage the complex, non-standard interactions that arise when automated vehicles share roads with human-driven traffic.

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