Popular scientific abstract
Integrating automated vehicles (AVs) into the traffic system may contribute to substantial reductions in road-traffic fatalities and injuries while improving mobility. Even though traffic in the future may become fully automated, there will still be vulnerable road users (VRUs) like pedestrians and cyclists interacting with the AVs. We currently lack information on how mixed urban traffic will be affected by the interaction between the different the road users in terms of safety, trust and transparency. The aim of ESR 10: Human machine interfaces (HMIs) promoting safe AV/cyclist interactions is to establish how the interaction between cyclists and AVs can be enhanced by cyclist-oriented HMIs. Our findings suggest that cyclists are enthusiastic about adding more technology to their bicycles, but there are concerns about the responsibility of safety being imposed on vulnerable road users if such HMIs are required for safety. External on-vehicle HMIs (eHMIs) that are visible from all around the vehicle are likely the best solutions to accommodate cyclists’ movement patterns and to make sure the message can be observed at higher speeds.
Who am I?
My name is Siri Hegna Berge, and I have a background in psychology. I have a Master of Philosophy in Psychology degree from the University of Oslo, specializing in human factors and human-computer interaction by studying trust in artificial intelligence during emergency management in my master thesis. After finishing my degree, I have worked as a Research Psychologist at the Institute of Transport Economics, the Norwegian Centre for Transport Research (TØI) at the Department of Safety, Security and Behaviour.
Working on various transport research projects in an inter-disciplinary environment such as TØI has sparked an increased interest in the interaction between humans and technology and how human factors affect the safety, trust and transparency of this interaction. What I find particularly fascinating is how the interaction between humans and AVs may be enhanced by designing HMIs from the perspective of vulnerable road users.
The SHAPE-IT project and ESR 10: HMIs promoting safe AV/cyclist interactions, is the perfect opportunity to pursue this research interest further and at the same time, contribute to the safer traffic environment of tomorrow.
My host university is Delft University of Technology (TU Delft), and I work in the TU Delft Faculty of Civil Engineering and Geosciences, at the Department of Transport and Planning.
My supervisors are Prof.dr. Marjan Hagenzieker and Dr.ir Joost de Winter, both at TU Delft.
Increased urbanisation and economic globalization, paired with dramatic population growth and consequently increased environmental impact, call for new and better transport systems. Removing the human driver by integrating AVs into the traffic system may contribute to substantial reductions in road-traffic fatalities and injuries while improving mobility. However, introducing fully automated vehicles into the traffic system may also increase traffic congestion and cause breakdowns of the transport system [1, 2].
AV technology is already being implemented on public roads and highways, and continuously tested in today’s traffic. This generates an increasing amount of knowledge on automation and the development of AVs. However, the perspective of this research is inherently technological- or driver-focused, with insufficient consideration to how humans interact with and influence AVs in traffic.
In line with the unsustainability of today’s transport systems, road users are encouraged to use more active forms of transportation, like walking and cycling [3, 4]. Even though traffic in the future may become fully automated, there will still be vulnerable road users (VRUs) like pedestrians and cyclists interacting with AVs. We currently lack information on how this mixed urban traffic will be affected by the interaction between the different road users regarding safety, trust and transparency.
Today, the interaction between road users is determined by factors such as traffic rules and regulations, individual differences, expectations, behavioural adaptation, informal rules and non-verbal communication . For instance, road users may signal their intentions using behavioural cues like facial expressions, hand gestures or eye contact . Other actions, like a reduction in vehicle speed or flashing the headlights may indicate behavioural intentions as well. To further complicate road user interaction, the implicit cues may differ from country to country and culture to culture. External human-machine interfaces (eHMIs) show promise as potential substitutes for human non-verbal communication [6-9]. Still, we are at the very beginning of gathering information about how the AV-VRU interaction will be affected once vehicles become automated and the driver is no longer in the loop.
To ensure the safe and acceptable integration of AVs into tomorrow’s mixed urban traffic environments, we need insight on the interaction between road users such as AVs, pedestrians and cyclists. The knowledge generated by SHAPE-IT is our contribution to help solve the upcoming challenges and prepare us for the traffic environment of tomorrow.
Aims and objectives
In ESR 10, we will investigate cyclist-oriented HMIs to aid cyclists in future automated traffic and to develop realistic test scenarios for testing cyclist HMIs in the context of AVs. The aim of the project is to facilitate safe and acceptable interactions between cyclists and AVs.
The objectives of the research are to:
- Establish and evaluate realistic, complex test scenarios for AV/cyclist interaction.
- Establish how communication from AVs to cyclists can be enhanced by cyclist-oriented devices.
- Establish what HMI modalities are useful for cyclists, and how the HMIs interact with the cyclist.
- Increased understanding of how HMI systems for cyclists can incorporate communication from infrastructure, such as traffic lights and other infrastructure-to-everything functionalities.
- Evaluate cyclists’ trust and acceptance of these methods across different biker characteristics (e.g., age, gender, and personality).
During this PhD project, three main studies have been conducted in a progressive manner to investigate the potential of equipping cyclists or bicycles with HMIs in future automated traffic. The first study was an exploratory interview study with a sample of 30 cyclists, 15 in Norway and 15 in the Netherlands, aimed at understanding cyclists’ experience sharing the road with vehicles, identifying their needs in increasingly automated traffic, and assessing the feasibility of equipping cyclists or bicycles with an HMI such as a communication device or warning system. Following the interview study, we performed a literature review study to examine the available devices and communicative technologies for cyclists, and to assess their potential for supporting cyclists in future traffic. The focus of the third study was on developing a selection of scenarios of cyclist interaction with AVs, and providing a realistic description of AV behaviour for future studies if an actual AV cannot be used.
The findings from the interview study revealed that cyclists are enthusiastic about integrating more technology on their person (e.g., cyclist wearables) or their bicycles, but not necessarily for the sake of AVs. The literature study concluded that there is a lot of potential for cyclist-oriented HMIs, with the option to wear AR-glasses, equip helmets and bicycles with signalling or warning systems, or connecting cyclists and bicycles to a larger network of AVs and infrastructure. However, the potential problem with most of these solutions is that they shift the responsibility of safety onto the vulnerable road user, which is not ideal from the cyclist’s perspective.
Similarly, the idea of connecting vulnerable road users for increased safety is not entirely feasible in the near future, as it would require everyone to be connected for the system to have an impact on safety. AVs dependence on data from all road users to venture safely in traffic could lead to an increased risk of accidents if the device or system malfunctions, or the user forgets their device at home. Therefore, we propose that on-vehicle eHMIs, are more viable solutions. On-vehicle eHMIs targeting cyclists should be designed with visibility all around the vehicle to accommodate for cyclists’ movement patterns and higher speeds than pedestrians. Utilising eHMIs instead of cyclist wearables or on-bike devices also would place the responsibility of safety where it belongs. Future research could focus on further development of eHMI designs for cyclists and testing the feasibility of these designs.
The scenario development study resulted in a scenario collection comprising of 20 prototypical scenarios of cyclist/AV interaction, grouped in four categories based on the road users’ direction of movement: crossing, passing, overtaking, and merging scenarios. Our assessments indicated that right-turning vehicles, dooring scenarios, and more complex situations have the highest likelihood of accidents. Passing and merging scenarios are particularly relevant for studying AV communication solutions, as these types of scenarios involve more negotiation. Future research on cyclist/AV interaction should consider phantom braking and the driving styles of vehicles.
- Cyclists are enthusiastic about the prospects of technology improving their cycling experience, but they do not desire mandatory devices or systems for safe interaction with AVs.
- A device requirement might become a barrier to cycling, as bicycles are traditionally cheap and simple, and additional costs might deter people from choosing cycling as a transport mode.
- In our literature study, we identified 92 concepts with the potential for supporting cyclists in future automated traffic. The concepts were analysed using a taxonomy of 13 variables based on the physical, communicational, and functional attributes of the systems. The systems were divided into four categories according to placement: cyclist wearables (39%), on-bike devices (38%), vehicle systems (33%), and infrastructure (23%). Most systems communicated visually (77%).
- We suggest that interfaces on motorised vehicles accommodate cyclists with visibility all around the vehicle and incorporate two-way communication.
- For operationalising AV behaviour in simulator or Wizard-of-Oz studies, accounting for phantom braking and differences in AVs’ driving styles are important.
Berge, S. H., de Winter, J. & Hagenzieker, M. (2023). Support systems for cyclists in automated traffic: A review and future outlook. Applied Ergonomics [in press]. https://doi.org/10.13140/RG.2.2.29549.05606/1
Nordhoff, S., Lee, J. D., Calvert, S. C., Berge, S., Hagenzieker, M., & Happee, R. (2023). (Mis-)use of standard Autopilot and Full Self-Driving (FSD) Beta: Results from interviews with users of Tesla’s FSD Beta. Frontiers in Psychology, 14, . https://doi.org/10.3389/fpsyg.2023.1101520
Berge, S. H., Hagenzieker, M., Farah, H., & de Winter, J. (2022). Do cyclists need HMIs in future automated traffic? An interview study. Transportation Research Part F: Traffic Psychology and Behaviour, 84, 33-52. https://doi.org/10.1016/j.trf.2021.11.013
Berge, S. H., De Winter, J., & Hagenzieker, M. (2023). User Interfaces for Cyclists in Future Automated Traffic. In IUI 2023 – Companion Proceedings of the 28th International Conference on Intelligent User Interfaces (pp. 91-94). Association for Computing Machinery (ACM). https://doi.org/10.1145/3581754.3584140
Berge, S. H., de Winter, J. C. F., & Hagenzieker, M. P. (2022). Cyclist support systems for future automated traffic. In Proceedings of the 10th International Cycling Safety Conference https://doi.org/10.25368/2022.478
Berge, S. H., de Winter, J., & Hagenzieker, M. (2022). Cycling in automated traffic: Scenarios and test criteria. Abstract from ICTTP 2022, Gothenburg, Sweden.
Berge, S. H., de Winter, J., & Hagenzieker, M. (2022). Cyclist support systems for future traffic: A review. Abstract from ICTTP 2022, Gothenburg, Sweden.
Berge, S. H., Hagenzieker, M., Farah, H., & de Winter, J. (2021). Do cyclists need HMIs in future automated traffic?. Abstract from ISCS 2021 International Cycling Safety Conference, Lund, Sweden. https://www.icsc-2021.net/conferences/09-lund-2021/
Berge, S. (Creator), de Winter, J. (Creator), Hagenzieker, M. P. (Creator) (5 Oct 2022). Research data for the paper: Support systems for cyclists in automated traffic: A review and future outlook. TU Delft – 4TU.ResearchData. 10.4121/21130373
Berge, S. (Creator), de Winter, J. (Creator), Farah, H. (Creator), Hagenzieker, M. P. (Creator) (22 Sep 2021). Research data for the paper “Do cyclists need HMIs in future automated traffic? An interview study”. TU Delft – 4TU.ResearchData. 10.4121/C.5559372
References and links
1. Nenseth, V., A. Ciccone, and N.B. Kristensen. (2019). Societal Consequences of Automated Vehicles–Norwegian Scenarios (1700/2019). Institute of Transport Economics. Retrieved from https://www.toi.no/getfile.php?mmfileid=50576
2. Kellett, J., et al. (2019, 2019/10/02). How Might Autonomous Vehicles Impact the City? The Case of Commuting to Central Adelaide. Urban Policy and Research, 37(4), 442-457. https://doi.org/10.1080/08111146.2019.1674646
3. POLIS. (2014). Securing the benefits of active travel in Europe: Position paper of the POLIS environment & health in transport working group. Retrieved 16th of June 2020 from https://www.polisnetwork.eu/wp-content/uploads/2019/06/polis-paper-health-and-transport_final-3-dec.pdf
4. Centers for Disease Control and Prevention. (2011). Strategies: Promote Active Transportation. Retried 16th of June 2020 from https://www.cdc.gov/healthyplaces/transportation/promote_strategy.htm
5. Vissers, L.K., et al. (2017). Safe interaction between cyclists, pedestrians and automated vehicles: what do we know and what do we need to know? SWOV. Retrieved from https://library.swov.nl/action/front/cardweb?id=131121
6. Rouchitsas, A. and H. Alm. (2019). External Human–Machine Interfaces for Autonomous Vehicle-to-Pedestrian Communication: A Review of Empirical Work. Frontiers in psychology, 10. https://doi.org/10.3389/fpsyg.2019.02757
7. De Clercq, K., et al. (2019). External human-machine interfaces on automated vehicles: effects on pedestrian crossing decisions. Human factors, 61(8), 1353-1370. https://doi.org/10.1177/0018720819836343
8. Mahadevan, K., S. Somanath, and E. Sharlin. Communicating awareness and intent in autonomous vehicle-pedestrian interaction. in Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems. 2018.
9. Habibovic, A., et al. (2018). Communicating intent of automated vehicles to pedestrians. Frontiers in psychology, 9, 1336. https://doi.org/10.3389/fpsyg.2018.01336