This study investigates the factors influencing Take-Over Time (TOT) during transitions from automated to manual driving, emphasizing the novelty of applying XGBoost modeling combined with SHAP analysis to uncover non-linear and implicit dependencies between features. Using high-frequency data from a driving simulator, key variables such as automation level, driving measurements, different types of obstacles, and HumanMachine Interface (HMI) conditions were analyzed to understand their effects on TOT. The XGBoost model was optimized using a cross-validation approach, achieving strong predictive performance (R2 = 0.871 for testing set). Feature importance analysis revealed that Automated Driving (AD) level 2 or 3 was the most influential factor, underscoring how extended time budgets and reduced driver engagement interact in shaping TOT. Higher automation levels resulted in longer TOT, with SHAP values consistently positive for AD Level 3, demonstrating the added value of explainable machine learning in clarifying these patterns. Dynamic driving parameters, such as deceleration and speed variability, were also significant. Strong negative deceleration values were generally associated with shorter TOT, reflecting quicker responses under urgent braking. Speed showed a moderate positive effect on TOT at 80-110 km/h, with drivers taking additional time to assess the environment, but higher speeds (above 110 km/h) resulted in quicker responses. Beyond these established effects, SHAP analysis revealed how automation level, obstacle environment, and HMI design jointly condition driver responses. The HADRIAN HMI, slightly increasing TOT compared to the baseline, simultaneously seems to demonstrate potential safety benefits through tailored guidance and improved situational awareness. By combining methodological innovation with contextual insights, this study contributes to a deeper understanding of takeover behavior and provides actionable evidence for optimizing adaptive HMI design and takeover strategies in AD systems.

Automated Driving (AD) has the potential to significantly reshape the transportation industry by improving safety, efficiency, and user comfort and acceptance. This research investigates driver behaviors during Take Over Requests (TORs) in automated vehicles at SAE Levels 2 and 3 using the HADRIAN Human-Machine Interface (HMI), designed to enhance driver support through real-time feedback and countdown displays. Analysis included clustering to develop distinct driving profiles based on key measurements collected through a driving simulator experiment, such as acceleration, deceleration, and speed, offering a deep understanding of driver behavior in responses to TORs. Three primary driving profiles were identified: “Passive driving and slow TOR response”, “Nervous driving and moderate TOR response”, and “Normal driving and quick TOR response”. This study also provides specific insights for each automation level, identifying profiles of driver behaviors at both SAE Levels 2 and 3. Results reveal that the nervous driving profile, although less frequent, poses significant safety implications due to higher deceleration rates and variability in speed and deceleration. Additionally, the study highlights that Non-Driving Related Tasks (NDRTs) increase the need for longer Take Over Time (TOT), with greater variability observed at higher automation levels, making accurate estimation of required TOT more challenging. The HADRIAN HMI is shown to positively impact driver performance by increasing TOT, allowing drivers more time to transition from automated to manual control comfortably. These insights can inform the design of more adaptive HMI systems, enhance real-time feedback mechanisms, and improve driver training programs to ensure safer transitions during TORs.

In this chapter we describe the conduct and results of two field-demonstration studies where the HADRIAN innovations were integrated into real vehicles and their effectiveness in terms of the human driver role were evaluated. In the first field-demonstration, twelve participants evaluated HADRIAN innovations that were intended to improve the safety and comfort of using automated driving at SAE L2 and L3. The study was performed in a passenger vehicle in an open road environment where participants compared interactions with a baseline automated vehicle with the HADRIAN innovations. In the second field-demonstration, ten participants experienced HADRIAN innovations that were intended to facilitate older drivers in a small passenger vehicle while driving on a test track in Spain. The results of both studies confirmed the key assumptions of the HADRIAN approach and identified limits and opportunities. We discuss these main lessons learned and conclude with what these findings tell us about the benefits and problems of adopting a holistic, user-centered approach for automated driving solutions.

AbstractThe development of Artificial Intelligence (AI) technologies experiences worldwide an ongoing challenge to become trustworthy and ethical for users and the general public. This challenge currently stands between the promise of AI to create immense societal and individual impact and its realization. Because of this, possible large marketplaces still remain hesitant or closed. We have investigated this problem and identified potential solutions in the InSecTT project, a large international EU research and development project that investigates ethical, smart technologies. Thereby, working with industrial and research partners, we assert that developing trustworthy AI technologies is not foremost a technical challenge, but increasingly an organizational and process challenge that results from applying traditional ways of conceiving, designing, and selling technologies to technologies with very new types of user and societal implications. Because AI technologies can shift the role that humans and society see as acceptable, traditional development processes that rely on the strict separation of specialties are overburdened. In our view, a research and development approach for trustworthy AI systems should put human concerns and needs at the start of the development process to effectively integrate humans and systems and thereby have a chance to meet EU guidelines for developing ethical AI. Our proposed approach to develop such ethical, trustworthy AI systems centers around the assessment of trustworthiness risks through intensive user involvement prior to the elicitation of system requirements that are then managed throughout the system’s life cycle. Also, the approach includes concrete recommendations to establish the organizational prerequisites for producing ethical, trustworthy AI systems. In this chapter, we describe the Human-Systems Integration (HSI) approach and motivate the underlying principles in some detail. The chapter intends to inform managers of technical organizations and product managers, as well as principal investigators, to set up the prerequisites for trustworthy AI, while also soliciting inputs for further refinement and discussion.

Supporting drivers in different levels of automation was one of the key goals in the HADRIAN project. Following the approach of a "fluid interface", i.e., an interface that considers the state of the driver, vehicle, and environment, and which uses different modalities to support the driver, a human–machine interface (HMI) was developed and compared to a baseline HMI in a simulator study (n = 39). The integrated fluid HMI aimed at supporting the driver in automated driving in SAE level 2 and 3 by providing mode relevant information, supporting the driver during take-over requests by the system, and supporting engagement in non-driving related tasks when allowed. The fluid HMI consisted of several components (head-up display, LEDs, haptic icons, sound, tablet) and featured driver monitoring and adaptive tutoring. Study results did not show significant differences between the HMIs regarding subjective measures such as user experience, usability, acceptance, or safety feeling. The various factors contributing to this conclusion are thoroughly discussed. However, objective measures in terms of eye movements and a safety analysis including driving data showed a significant benefit of the integrated fluid HMI over the baseline HMI. Participants had better mode awareness and a higher safety score with the integrated fluid HMI. Furthermore, valuable insights on how to further improve the HMI could be gained during the study.

Automated Driving (AD) technologies are transforming the mobility sector, promising enhancements in efficiency and safety as well as user experience tailored to diverse user needs. As the sector moves toward a more automated ecosystem, it's essential to recognize the varied expectations and requirements of its users. Elderly drivers, for instance, have distinct needs, from functional transportation to the aesthetic pleasure of driving. As age-associated impairments affect their driving capabilities, Advanced Driving Assistance Systems (ADAS) present an opportunity to augment their safety and confidence. Truck drivers or business people on the other hand have different sets of needs and expectations. The article emphasizes the role of User-Centered Design (UCD) in the HADRIAN project with the intent to fulfill the promise of AD. Through the iterative creation of personas, like Harold, representing the elderly, Sven, the seasoned truck driver, and Florence, the business woman, the project brings to light the requirements for AD from a user-centric focus. Such personas, backed by comprehensive research and expert insights, steer in the HADRIAN project not only the design of interfaces but influence the development of AD systems ensuring broader acceptance and inclusivity. As the landscape of urbanization and digitalization expands, coupled with the emergence of smart cities and shared mobility solutions, the integration of AD with these broader trends becomes imperative. This will require a larger ecosystem that allows to steer vehicle development processes increasingly toward the specific needs of its user groups, moving away from one-for-all vehicles. In the HADRIAN project we envisioned such an ecosystem to be able to step beyond currently prevailing vehicle development approaches and show how with a user-focused approach, the future of transportation can become a harmonious blend of safety, efficiency, and inclusivity, resonating with the real-world needs of diverse commuters.

Research and development of high levels of automated driving (AD) vehicles has received considerable attention in recent years. Thereby, for the time being, humans will remain actively involved to assure overall safety, whether as drivers, safety drivers, or tele-operators. This effectively shifts human tasks and responsibilities compared to manual driving. To make these shifts as safe and comfortable as possible and also reliable and predictable to use, the Horizon Europe research and innovation project HADRIAN (Holistic Approach for DRiver role IntegrAtioN) investigated and evaluated holistic, user-centered solutions. Thereby, the HADRIAN consortium envisioned a larger eco-structure from which it would be possible to reconceptualize what is part of driving automation and how it works. This meant to include parts of the roadside information infrastructure as well as to directly include the human driver in two specific ways: first by designing AD solutions that directly support their mobility needs and constraints. And secondly, by shaping the AD solutions that allow drivers to perform their new tasks and responsibilities more safely and comfortably. In this volume we describe how such holistic, user centered approach allows to derive better and more powerful solutions than those that are merely focused around the vehicle. For this we report the results of a series of innovations and their evaluations and demonstrations in the field. We conclude with how such approach also requires more tightly connected and inter-disciplinary team collaborations than are often found in current research and development organizations. In this first chapter we introduce the underlying human factors problems of currently available levels of AD and thereby motivate the starting point for the holistic user-centered approach and solutions that are then described to greater extent in the following chapters.

AbstractThe current paper was performed within the HADRIAN project and focuses on exploring the effects of innovative Human–Machine Interface (HMI) prototypes on safety, driving performance, and driver perceptions. Employing driving simulator experiments and questionnaires, this study investigates whether HADRIAN innovative HMI enhances safety and receives positive evaluations from drivers. Specifically, the research centers on a driving simulator experiment that evaluates novel HMI prototypes designed to improve automated driving at SAE Levels 2 or 3. To facilitate HMI assessment, a tailored safety and impact assessment methodology was developed using unique Key Performance Indicators (KPIs). To benchmark and generate a total score for the HADRIAN HMI, data envelopment analysis was deployed based on the aforementioned KPIs. The findings shed light on the influence of HADRIAN HMI innovations on safety and perceived impact when compared to a baseline “state-of-the-art” HMI. Subsequently, a comprehensive discussion unfolds, highlighting the key KPIs that contributed significantly to the safety and perceived impact scores. This method and its outcomes can serve as a valuable resource for other HMI stakeholders, enabling them to employ similar human-centered assessment methodologies to assess the safety and perceived impact of potential HMI configurations.

AbstractMany people in Europe still have limited access to transportation modes overall. Socio-economic constrains as well as cognitive, sensory and physical impairments affect everyday life of these citizens, posing challenges to access mobility services.Technologies for vehicle automation have advanced greatly in recent decades and it is expected to become part of vehicle fleets in the foreseeable future. Yet, the implementation and use of automated and autonomous vehicles (here jointly referred to as AVs) entails chances but also hurdles regarding accessibility and inclusivity of vulnerable groups. This concerns both the use of the vehicle by humans as well as the interaction between humans and vehicles as participants in road traffic.In this chapter, these aspects shall be presented by identifying opportunities and risks as part of our mobility system, starting from a narrowing down of the vulnerable social groups we are looking at. Subsequently, we present the benefits that co-creation and universal design can have in overcoming or, in the best case, avoiding these obstacles. Even though the authors are aware that no detailed recommendations for action can be given within this framework, at least suggestions for solutions are outlined.

Higher levels of Automated driving (AD) vehicles require new allocations of functions among drivers, vehicles, and road infrastructure. The European Horizon 2020 project HADRIAN investigates how such reallocations could be practically achieved as part of Collaborative Connected and Automated Mobility (CCAM) to meet the benefit expectations of drivers while increasing safety. In a field demonstration it is shown how road infrastructure can be used to expand the prediction horizon of AD vehicles and how multimodal, driver-state dependent human machine interactions (HMI) could help address European mobility needs with AD vehicles and increase operational acceptance and safety. Whereas performance results of the various innovations are reported elsewhere, in this paper the evaluation of the feasibility of the HADRIAN innovation in an open road field-demonstration is described.

Vehicles offering conditional automated driving (SAE Level 3 [1]) are now becoming available worldwide. At this automated driving level, it is possible for drivers to engage in non-driving related activities while the vehicle takes over all lateral and longitudinal maneuvers. However, drivers must remain fallback ready to reengage control when requested to do so by the vehicle. Human factors research in automation has for a long time established that it is difficult for drivers, who are out-of-the-loop, to quickly, effectively, and safely re-engage in the driving task. This video submission presents two methods to increase the acceptability and safety of automated driving at SAE level 3. These methods were developed in the EU Horizon 2020 project HADRIAN (Holistic Approach for Driver Role Integration and Automation Allocation for European Mobility Needs; https://hadrianproject.eu/). The first method makes the duration of SAE L3 drives more predictable and expectable to the driver by displaying the duration. Duration information is received by road infrastructure messages that are sent to the vehicle. Also, the amount of time that a driver has to reengage manual control after driving at SAE L3 is predicted and displayed to the driver. The second method provides the driver with various types of learning information and feedback to help improve the driver's skills in handling the automated vehicle. Detailed analyses describing the impact of these methods are currently under investigation and will be reported soon.

With increasing levels of automation on modern vehicles, drivers need more knowledge about their vehicle than ever before. Thereby, the display of vehicle status and alerts are important means for safe and effective use. Nevertheless, there are currently no common and consistent ways of providing such status in‐formation to drivers. This is different from other domains where standardized alert management structure are common such as in modern aviation. In current vehicles, initiation, use, and termination of automated driving assistance differs considerably between vehicle models and brands and require special learning ex‐periences by the driver and alerts are not prioritized and harmonized across the various onboard systems. With higher levels of automation such complexity is expected to increase enormously. To address this problem, we propose a common taxonomy of vehicle display status and alerting terminology that is based on flight deck regulations in aviation and propose simple principles for their use in a man‐aged display status and alerting system.

A critical challenge of higher levels of automated driving (SAE level 3) is the reengagement of the driver to take back manual control. In this relatively novel mode of automated driving that is slowly becoming commercially available, the driver can perform non-driving related tasks but has to take over when the vehicle reaches the boundaries for its design domain for automated driving. The challenge here is for the driver to build sufficient situation awareness of the vehicle and the environment before taking back control in a timely and safe manner. In this study we investigated to what extent the driver could be helped by receiving predictive information about the duration of the automated driving as well as the available time for the reengagement.To address this research question, we conducted a simulator study where 41 participants drove alternating in manual mode and automated mode. Multiple times the participants had to take back control of driving prior to a stationary vehicle that blocked the lane. Audiovisual cues informed the participants about the necessary take-over 15 seconds in advance. The cockpit display showed the current driving mode (automated versus manual), as well as one of four types of prediction information: a) The baseline display type showed no prediction of time at all, b) the transition prediction (TP) display type showed the available time for the take-over, c) the automated driving prediction (AP) display type showed a remaining time during automated driving, and d) the combined display type showed both types of information (TP and AP). We compared the perceived usefulness of the prediction types in a questionnaire, the gaze behavior during control transition/automated driving segments as well as driving performance.The results indicate that the combined display type was perceived to be highly useful by the participants when transitioning control from automated to manual driving. This perceived usefulness is positively associated with their intention to use such a system in their daily lives. An analysis of the driver’s gaze indicates that drivers used the combined display type during takeovers and automated driving more than the other display types. Furthermore, drivers apparently acquired over time a safer gaze behavior with the combined display type as they monitored the road environment more during control transitions than in the other conditions. This effect was only present after a couple of takeovers, showing a gradual increase, which indicates that it requires some learning to fully utilize the time predictions.The results imply that a combined display type of predictive information about the duration of automated driving and reengagement time was perceived to be highly useful by the participants. In this paper we describe the results along with performance results and a comprehensive assessment with implications for further research.

A difficult challenge for today's driver monitoring systems is the detection of cognitive distraction. The present research presents the development of a theory-driven approach for cognitive distraction detection during manual driving based on temporal control theories. It is based solely on changes in the temporal variance of driving-relevant gaze behavior, such as gazes onto the dashboard (TDGV). Validation of the detection method happened in a field and in a simulator study by letting participants drive, alternating with and without a secondary task inducing external cognitive distraction (auditory continuous performance task). The general accuracy of the distraction detection method varies between 68% and 81% based on the quality of an individual prerecorded baseline measurement. As a theory-driven system, it represents not only a step towards a sophisticated cognitive distraction detection method, but also explains that changes in temporal dashboard gaze variance (TDGV) are a useful behavioral indicator for detecting cognitive distraction.

While the market of smart technologies is steadily increasing, there is much research to be done regarding the interaction between human users and Artificial Intelligence (AI) technologies. Specifically, the field of Explainable Artificial Intelligence (XAI) focuses on making AI explainable to users. To provide a user-centered approach to this growing field, this paper describes a study to investigate possible processes and methods. For this purpose, 20 participants were asked to use an AI system that provided them with the results of a personalized COVID-19 risk calculation. The study results indicate that while participants generally seemed to think that the presented results of the system were accurate, only a few said that they would change their behavior after receiving the results, and many asked for additional information to better understand the results. This paper discusses the findings along with possible approaches to increase behavior change in users of smart systems.

Truck platooning is a form of convoy cooperative driving of connected trucks assisted by a lead truck. The aim is to reduce the fuel and driving costs, improve road safety, and reduce CO 2 emission. Being semi-autonomous, platoons must be trustworthy in many perspectives. This paper presents a high-level trustworthy requirements analysis on three key perspectives: driver, communication, and security. In addition, we observed that any trustworthy requirement analysis is incomplete if perspectives are addressed independently. Therefore, we propose a simple holistic methodology that addresses the different perspectives as well as their dependencies, and we exemplify the use of the methodology with two use cases presented in the paper. However, we draw attention to the importance of more research to drive a more exhaustive and validated methodology 1 .

Train drivers in Sweden had the opportunity to test the new European Rail Transport Management System (ERTMS) during the early implementation of this signaling system in 2011 and 2012. Unfortunately, they have reported many implementation problems. These ERTMS lines have been evaluated as more challenging to drive compared to the previous signaling system. A systematic literature review was conducted to understand what network planning designers typically focus on when analyzing railway driveability and how driveability can be assessed. In the overall picture the impact on driveability originates from both technological, organizational, and train driver aspects. One technological issue found during the investigation is related to the speed profile design. Frequent speed changes were identified as the primary source of negative experiences during driveability evaluations, especially on retrofitted lines. Results also indicate the need for a methodology that includes the driver earlier in the network planning test process, to avoid the costly and time-consuming analysis done after project implementation. The results highlight the importance of the speed profile design to provide good driveability. Cognitive Task Analysis, simulations, and statistical calculations to predict driver and train performance were identified as applicable methods to involve train drivers earlier in the driveability analysis process.

Assistance systems designed to help workers in their jobs are increasingly used in industry. Technological progress makes these systems more powerful and extensive, but often nobody questions the extent to which they actually support the users and do not patronize them. For the development of such systems, we found the requirement analysis to be rather complex because human factors and social constraints are more difficult to determine than technical requirements. To counteract these difficulties, we pursue in our approach the involvement of people as knowledge carriers in the development of new technologies. In this paper we outline our framework how human factors aspects of acceptance and job satisfaction can be taken into account in the conception and design of assistance systems.