Popular scientific abstract
Vehicle automation has emerged as a significant innovation in the fields of automotive and transportation, offering considerable potential for enhanced mobility and safety. While substantial advancements have been made in perception, planning, and control technologies, the interaction between automation and users, as well as users’ reactions to automation, remain under-examined. For example, the extent of automation’s reliability and the perception of safety within automated driving systems are not yet fully understood. This can lead to unwanted consequences, such as undertrust and disuse of some automated driving systems or overtrust and misuse of driving assistance systems. Consequently, accidents involving driving automation still occur.
Under the SHAPE-IT framework, which aims to facilitate the safe and acceptable integration of user-centered, transparent automated vehicles into tomorrow’s mixed urban traffic environments, I, as ESR12, will address the challenges of fostering user trust in driving automation and ensuring users feel safe in automated vehicles. To achieve this, I will explore how perceived safety and trust function in automated vehicles, and then work to investigate the user interface design inside automated vehicles to enhance drivers’ perceived safety and trust.
My research will involve conducting driving simulator experiments to identify factors that influence perceived safety and trust, and to obtain model parameters. Furthermore, I will investigate how perceived safety and trust can be enhanced by providing user interfaces in automated vehicles. The expected results of my research will encompass three key aspects: first, methods to develop natural, trusted, and accepted interactions between automated vehicles and their users; second, a model of perceived safety; and finally, recommendations for human-automated vehicle interaction design to ensure these vehicles are perceived as safe and reliable.
Contact details to my supervisors:
Trust is a vital factor that determines user’s willingness to use and rely on the automated systems . Therefore, many researchers tried to explain and model trust, which can give insights into the interaction design in automated vehicles. However, many trust models are conceptual models , . For the measurement of trust, although questionnaire is the most popular measurement of trust, it is not realistic to use questionnaire to obtain on-road driver’s trust level in real time. Therefore, surrogate measures need to be created to obtain driver’s subjective feelings in real time.
Perceived safety is a psychological item which has a close relationship with trust, being pivotal for the user’s acceptance of automated vehicles. It captures the subjective evaluation of safety by users of automated driving systems but also by other road users interacting with automated vehicles. The models from the existing researches are similarly conceptual ones and are always combined with trust models in some driver acceptance model of automated driving systems –. Consequently, building a mathematical model of perceived safety to predict driver’s perceived safety in real time is also necessary.
These existing researches can only provide a basis concept of perceived safety and trust. Some others built effective metrics for perceived safety but need further research to confirm real-time capability . In addition, how to explain the relationship between perceived safety and trust and how to enhance perceived safety and keep trust in an appropriate level still need further research and experiments.
I am Xiaolin He, holding a Master of Engineering degree from China, with a focus on human adaptive design for Advanced Driver Assistance Systems (ADAS). Prior to joining the SHAPE-IT project, I worked as an engineer at SAIC Volkswagen, where I was involved in the development of driving automation technologies. My expertise in human adaptive design and driving automation will be invaluable in contributing to the success of this project.
Aims and objectives
There are currently 4 overall objectives of my ESR-project
- To find out which factors determine perceived safety and trust and to what extent they affect perceived safety and trust.
- To model perceived safety and trust. The model can be used to design automated driving styles.
- To explain the relationship between perceived safety and trust.
- To design suitable internal user interfaces to enhance perceived safety and trust
The general goal of my research is to study perceived safety and trust in automated vehicles considering their modelling, their relationship between each other and the enhancement. Some new models and methods are going to be created and control theories are likely to be used to adapt the driving style of the automation in order to enhance the perceived safety and trust. The following approaches would be used to fulfill the aims.
- Literature review
In this phase, survey of literature will be conducted to have a better understanding of perceived safety and trust on the determinants and the modelling method. After the literature review, the factors which determine perceived safety and trust can be basically found. Meanwhile, I can lean some effective ways to obtain drivers perceived safety and trust.
- Experiment on the simulator for the data collection
After some determinants are basically found, experiment on the simulator would be designed and conducted to find out and validate the factors which have an influence on perceived safety and trust. I need to design the scenarios, collect some data from the vehicle, other road users and the users inside the automated vehicle. Then, based on the experiment control and analysis of the data, the factors can be found.
- Modelling of perceived safety and trust including indicator selection, mathematical modelling and calibration of the model on driving simulator
At this stage, based on some models of functional safety, models of perceived safety and trust will be built with the subjective feelings of drivers. Then, simulator experiment will be conducted again to calibrate the model.
- Designing User Interfaces for Partially Automated Vehicles to enhance perceived safety and trust based on the model.
Based on the mathematical model of perceived safety and trust, the mechanism of perceived safety and trust will be explained well. Then, we can design user interfaces for partially automated vehicles with the models to get a higher perceived safety and trust level will be built.
- Online perceived risk survey to collect large-scale perceived risk data and validate the computational perceived risk model.
An online perceived risk survey will be conducted to collect large-scale perceived risk data and validate the computational perceived risk model.
The expected results of my research are as following:
- Publications (journal papers, conference papers, dissertation thesis)
- Model of perceived safety
- Methodology for the adaptive design of automated vehicle in order to get a higher level of perceived safety and trust.
- Open datasets and codes
Perceived risk and trust modelling
- Regression models predict event-based changes in perceived risk and trust in automation.
- Relative motion of other vehicles (gap, TTC, braking intensity), personal characteristics (age, gender, driving experience) effectively predict perceived risk and trust.
- Perceived risk and trust were confirmed to be highly correlated. People trust the automation more after events where they perceive a lower risk, and people who perceive less risk trust the automation more.
- Brake behaviour could effectively indicate perceived risk and trust. Pupil dilation could reflect perceived risk if events were sufficiently risky. The merging and braking events increased heart rate, but no specific relation was found between heart rate increase (variability) and perceived risk.
Internal user interfaces design
- The most advanced UI (SM-VA), delivering both surrounding and manoeuvre information via visual and auditory modalities, resulted in the highest trust and acceptance ranking and the lowest perceived risk among drivers.
- The study found that manoeuvre information conveyed through the auditory modality had a more significant impact on trust and acceptance compared to the visual modality, emphasizing the importance of auditory communication in UI design.
- Criticality of event types and individual differences of participants were found to have a more substantial influence on drivers’ behaviour, trust, and perceived risk compared to the UIs, highlighting the complexity of human-automation interaction.
- Eye-tracking results demonstrated that drivers checked the centre-console UI when present, but no significant difference in gaze behaviour was observed between the four UIs.
Computational perceived risk modelling
- Driving task difficulty serves as an effective indicator of perceived risk.
- Perceived risk is 2-dimensional, originating from both longitudinal and lateral directions, and exhibits a non-linear increase as the distance to surrounding vehicles decreases.
- Incorporating uncertainties in the modelling process is crucial for a more accurate representation of perceived risk.
- Perceived risk fields are dynamic, reflecting human drivers’ risk perception under specific conditions rather than being static entities.
- There remains a lack of consensus regarding the most suitable fundamental unit for perceived risk.
He, X., Stapel, J., Wang, M., & Happee, R. (2022). Modelling perceived risk and trust in driving automation reacting to merging and braking vehicles. Transportation Research Part F, 86, 178–195. https://doi.org/10.1016/j.trf.2022.02.016
Lu, C., He, X., van Lint, H., Tu, H., Happee, R., & Wang, M. (2021). Performance evaluation of surrogate measures of safety with naturalistic driving data. Accident Analysis and Prevention, 162, 106403. https://doi.org/10.1016/j.aap.2021.106403
Nordhoff, S., Stapel, J., He, X., Gentner, A., & Happee, R. (2021). Perceived safety and trust in SAE Level 2 partially automated cars: Results from an online questionnaire. PLOS ONE, 16, e0260953. https://doi.org/10.1371/journal.pone.0260953
Nordhoff, S., Stapel, J., He, X., Gentner, A., & Happee, R. (2022). Exploring the Factors of Perceived Safety and Trust in Saelevel 2 Partially Automated Cars Using Principalcomponent Analysis. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.4156405
He, X. (2021). AV occupants’ perception of safety in relation to AV behaviour. Poster presented at the 13th International Conference on Automotive User Interfaces and Interactive Vehicular Applications. https://www.shape-it.eu/wp-content/uploads/2021/09/ESR12_Xiaolin-He-1.pdf
He, X. (2023). Computational perceived risk models in Level 2 automation: formulation, validation and comparison with experimental data. Poster presented at the 2023 TRB Annual Meeting. https://annualmeeting.mytrb.org/FileUpload/PDFUpload?ID=50026&SessionID=19459&ConferenceID=11
References and links
 J. K. Choi and Y. G. Ji, “Investigating the Importance of Trust on Adopting an Autonomous Vehicle,” Int. J. Hum. Comput. Interact., vol. 31, no. 10, pp. 692–702, 2015, doi: 10.1080/10447318.2015.1070549.
 K. A. Hoff and M. Bashir, “Trust in automation: Integrating empirical evidence on factors that influence trust,” Hum. Factors, vol. 57, no. 3, pp. 407–434, 2015, doi: 10.1177/0018720814547570.
 Frederick Steinke, Tobias Fritsch, and Lina Silbermann, “A Systematic Review of Trust in Automation and Assistance Systems for Older Persons’ Overall Requirements,” eTELEMED 2012, Fourth Int. Conf. eHealth, Telemedicine, Soc. Med., no. c, pp. 155–163, 2012.
 K. Garidis, L. Ulbricht, A. Rossmann, and M. Schmäh, “Toward a User Acceptance Model of Autonomous Driving,” vol. 3, pp. 1381–1390, 2020.
 Z. Xu, K. Zhang, H. Min, Z. Wang, X. Zhao, and P. Liu, “What drives people to accept automated vehicles? Findings from a field experiment,” Transp. Res. Part C Emerg. Technol., vol. 95, no. February, pp. 320–334, 2018, doi: 10.1016/j.trc.2018.07.024.
 J. C. Zoellick, A. Kuhlmey, L. Schenk, D. Schindel, and S. Blüher, “Amused, accepted, and used? Attitudes and emotions towards automated vehicles, their relationships, and predictive value for usage intention,” Transp. Res. Part F Traffic Psychol. Behav., vol. 65, pp. 68–78, 2019, doi: 10.1016/j.trf.2019.07.009.
 F. A. Mullakkal-Babu, “Modelling Safety Impacts of Automated Driving Systems in Multi-Lane Traffic,” Delft University of Technology, 2020. https://tudelft.on.worldcat.org/oclc/8541646509