Evaluating Pedestrian Risks in Shared Spaces Through Autonomous Vehicle Experiments on a Fixed Track

TL;DR

This study investigates pedestrian safety in unregulated shared spaces where autonomous vehicles operate on fixed tracks, such as trams. It uses Surrogate Safety Measures PET and TTC to quantify pedestrian responses to warnings under various distraction conditions, employing a paired, non-parametric analysis. Key findings show that warnings increase safety distance primarily for pedestrians wearing headphones, while effects for non-headphone users remain inconclusive; TTC results are limited by small stopping-event samples. The work highlights the potential for safety cues to adapt from road contexts to rail-like fixed-track environments and emphasizes the need for further data to guide safety standards for tram-pedestrian interactions in shared spaces.

Abstract

The majority of research on safety in autonomous vehicles has been conducted in structured and controlled environments. However, there is a scarcity of research on safety in unregulated pedestrian areas, especially when interacting with public transport vehicles like trams. This study investigates pedestrian responses to an alert system in this context by replicating this real-world scenario in an environment using an autonomous vehicle. The results show that safety measures from other contexts can be adapted to shared spaces with trams, where fixed tracks heighten risks in unregulated crossings.

Figures (9)

• Figure 1: Experiment procedure: Participants walk between signs A–G, following a path similar to the blue route, while the vehicle moves along the red path.
• Figure 2: The figure shows a participant crossing the track of the vehicle. The participant is warned about the approaching vehicle when they enter the area between the two blue highlighted lines, parallel to the fixed track of the vehicle marked in red.
• Figure 3: Visualization of the PET calculation for a crossing scenario. The participant moves along the black curve in the reference frame of the vehicle, and exits the vehicle's path at a relative distance $d_{PET}$. Dividing the distance by the speed $v_{vehicle}$ of the vehicle we obtain the PET.
• Figure 4: The potential collision between the pedestrian and the vehicle is predicted based on the pedestrian's continued movement with velocity $v_{pedestrian}$ and the vehicle's continued movement with velocity $v_{vehicle}$, reaching the point defined by the distance to the vehicle, $d_{TTC}$. This calculation is conducted at each timestep as the pedestrian progresses along the black curve. The minimum TTC value is considered to evaluate the safety of the interaction.
• Figure 5: Visualization of the distributions of TTC values for all stopping scenarios, split into warning and no warning interactions. The center line of each box indicates the mean values. The boxed area contains 50% of the points. Above each configuration is the number of interactions of that specific configuration. Note that in three stopping interactions with a warning no TTC could be determined due to the absence of a collision path.