You Shall Not Pass: Warning Drivers of Unsafe Overtaking Maneuvers on Country Roads by Predicting Safe Sight Distance
Adrian Bauske, Arthur Fleig · 2025
Read You Shall Not Pass: Warning Drivers of Unsafe Overtaking Maneuvers on Country Roads by Predicting Safe Sight Distance by Adrian Bauske, Arthur Fleig (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-18163.txt.
Introduction
Adrian Bauske and Arthur Fleig build a warning tool for passing cars on country roads, in a 2025 arXiv paper. Passing with traffic coming the other way kills many drivers each year. Older aids lean on car-to-car radio links and sensors most cars lack. This tool instead works with plain in-car sensors and 3D map data to judge how much clear view a pass would need.
The team tested two screen designs in a virtual driving rig with twenty-five drivers. Both designs made drivers wait for safer gaps. One, built around watching the road state, scored higher on ease of use and stole less focus. Taste still split between the two. The authors also check where their sight forecasts miss real driving, and name the fixes that help most.
Catalog record
| CREATOR | Adrian Bauske, Arthur Fleig |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 116 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-18163
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
You Shall Not Pass: Warning Drivers of Unsafe Overtaking Maneuvers on Country Roads by Predicting Safe Sight Distance Adrian Bauske, Arthur Fleig Abstract . Overtaking on country roads with possible opposing traffic is a dangerous maneuver and many proposed assistant systems assume car-to-car communication and sensors currently unavailable in cars. To overcome this limitation, we develop an assistant that uses simple in-car sensors to predict the required sight distance for safe overtaking. Our models predict this from vehicle speeds, accelerations, and 3D map data. In a user study with a Virtual Reality driving simulator (N=25), we compare two UI variants (monitoring-focused vs scheduling-focused). The results reveal that both UIs enable more patient driving and thus increase overall driving safety. While the monitoring-focused UI achieves higher System Usability Score and distracts drivers less, the preferred UI depends on personal preference. Driving data shows predictions were off at times. We investigate and discuss this in a comparison of our models to actual driving behavior and identify crucial model parameters and assumptions that significantly improve model predictions. Conference: CHI Conference on Human Factors in Computing Systems; April 26-May 1, 2025; Yokohama, Japan CHI Conference on Human Factors in Computing Systems (CHI ’25), April 26-May 1, 2025, Yokohama, Japan DOI: 10.1145/3706598.3713768 ISBN: 979-8-4007-1394-1/25/04 CCS: Human-centered computing Empirical studies in HCI CCS: Human-centered computing User studies Figure 1 . (A) Collisions with oncoming traffic when overtaking on rural roads can be fatal. (B) We enhance the driver’s decision-making by computing the necessary sight distances for safe overtaking maneuvers on country roads, taking into account vehicle speeds, accelerations, and 3D map data. We compare two UI variants that use these predictions: (C) Our monitoring-focused UI displays a warning icon on the heads-up display whenever the sight distance is insufficient for safe overtaking. (D) Our scheduling-focused UI additionally displays the distance (in meters) until the next overtaking opportunity. This figure is divided into four parts: A, B, C and D. Part A shows a car in the process of overtaking another car on a two lane rural road. The overtaking car is roughly next to the overtaken car. On the opposing lane there is a car close to the overtaking car, forming a very dangerous overtaking maneuver. Part B shows a schematic of a curved road with a tree in the center of the curve. A typical overtaking maneuver is depicted on the road with an overtaking car, an overtaken car and an opposing car. Furthermore, sight lines to the opposing lanes are drawn, some of which the tree blocks. Part C shows the monitoring-focused UI. An icon that warns of overtaking is displayed behind the steering wheel. The car is driving on a regular country road with cars in front and cars on the opposing lane. Part D shows conceptionally the same situation as Part C, a car driving on a country road with an active overtaking warning, but the UI additionally displays the information 451m in red on the right side next to the warning icon. † † cc-license: by 1. Introduction Imagine driving on a two-lane country road in the car marked by the red arrow in Figure 1 (A). How to avoid such potentially fatal situations and make overtaking safer? Technical assistance systems can significantly improve safety and help reaching the UN General Assembly’s adopted resolution 74/299 Improving global road safety ( Assembly, 2000 ) to halve the number of road traffic deaths and injuries by 2030. As we are transitioning towards vehicle automation, there is extensive literature on overtaking on motorways and multi-lane roads, see e.g., ( Yu et al., 2014 ; Fank et al., 2021 ; Ortega et al., 2023 ) . Commercially available driving assistants are able to automatically overtake on multi-lane freeways ( Mercedes-Benz USA, 2023 ) , where there is no oncoming traffic. However, UNECE statistics ( for Europe, 2021 ) show that road traffic accidents are far more common on country roads than on motorways. These roads are often narrow, with blind curves and junctions, offering few safe overtaking spots. Traffic is diverse, including cyclists (vulnerable road users ( Council of European Union, 2017 ) ) and slow farm vehicles. Some drivers feel "compelled" to overtake and frequently search for overtaking opportunities ( Kinnear et al., 2015 ) . As many dashcam videos on platforms like YouTube show, some drivers take greater risks with shorter sight distances, endangering themselves and others. Yet, to the best of our knowledge, no commercially available assistance system for overtaking on rural roads exists. This is because a big challenge is sensing or confidently predicting oncoming traffic in due time. Sensors installed in cars today are unsuitable for this task, e.g., because of their limited range. As an alternative, many concepts require vehicle-to-vehicle (V2V) and/or vehicle-to-interface (V2I) communication ( Tan and Huang, 2006 ; Toledo-Moreo et al., 2009 ) , which is not yet (widely) available. In this paper, we develop an assistant for country roads that operates without these concepts and works within the range of current in-car sensors. Our assistant does not overtake autonomously, and instead enhances the driver’s decision-making process before initiating any overtaking maneuver. Our main idea is to predict the required sight distance for safe overtaking and warn drivers of dangerous overtaking maneuvers, such as in Figure 1 (A). We base our prediction on vehicle speeds, accelerations, and 3D map data (Figure 1 (B)). We introduce two UI variants: the monitoring-focused UI, which shows a warning icon when sight distance is insufficient (Figure 1 (C)), and the scheduling-focused UI, which provides additional information like the distance to the next overtaking opportunity. We refer to these instances as opportunities as our system identifies situations where visibility is insufficient. If no warning is displayed, it is still the driver’s responsibility to check for oncoming traffic and execute the overtaking maneuver. We assess the potential of our assistant and the two UIs in a user study ( N=25 ) on a Virtual Reality (VR) driving simulator. The following questions, for which we derive hypotheses in Section 5.1 , guide our research: RQ.1 How do drivers perceive and accept our assistant (in terms of workload and usability) across the UIs? RQ.2 Which UI helps drivers (more) to avoid dangerous overtaking maneuvers? RQ.3 How accurate are the sight distance prediction models used by our assistant? In summary, we contribute a driving assistant with two UI variants, enabled by new models for sight distance prediction. We evaluate the assistant and compare the UIs and models in a user study, and extensively discuss the results. We open-source the sight distance prediction models at https://github.com/abauske/YouShallNotPass Moreover, with TOWARDS ( Bauske and Fleig, 2025 ) , we make the driving data from the user study on the VR driving simulator – with 1240 minutes of driving (1363 minutes simulation time including breaks) and 576 completed overtaking maneuvers – freely available. 2. Background and Related Work 2.1. Available Sensors and Data, and Constraints Ahangar et al. ( Ahangar et al., 2021 ) provide a comprehensive list of currently available in-car sensors and their capabilities. This includes ultrasonic sensors (distant measurement to objects up to 3m away), radio detection and ranging (RADAR) (presence detection of and distance measurement