Community Research and Development Information Service - CORDIS

Periodic Report Summary 1 - HFAUTO (Human Factors of Automated Driving)

Progress summary of HFauto: Human Factors of Automated Driving


Project objectives
Road transport is an essential part of society but the burden of crashes, congestion, and pollution is enormous. Highly automated driving (HAD) has the potential to resolve these problems. However, before HAD can be deployed we have to address imminent human factors questions regarding safety, traffic flow efficiency, human machine interfaces (HMI), and drive state assessment. The mission of HFauto is to generate knowledge on human factors of automated driving towards a better road safety. HFauto bridges the gap between engineers and psychologists through a multidisciplinary research and training programme. Furthermore, by means of secondments in automotive industry, road safety institutes, and academia, the researchers gain transferable knowledge. HFauto involves 13 Early Stage Researchers (ESRs) and 1 Experienced Researcher (ER), clustered in five synergistic work packages (WPs):

WP1. Human behaviour during highly automated driving (2 ESRs).
WP1 determines the effects of automation on the state of the driver. We investigate how platooning (i.e., being transported in a tightly spaced group of car) influences psychological constructs such as workload, situation awareness, trust, fatigue, vigilance, arousal, and stress. Moreover, this WP investigates the human factors associated with control transitions between manual and automated driving.

WP2. Human machine interface of the future highly automated vehicle (4 ESRs).
WP2 develops a HMI that supports the driver of a highly automated vehicle. The HMI should facilitate human-to-vehicle instruction (e.g., to change automation mode) and offer automation-status feedback. WP2 develops a visual ambient/head-up display, dynamic vibrotactile feedback, and spatial sound concepts for enhancing situation awareness. A demonstrator will be created as final deliverable.

WP3. Driver state monitor for highly automated driving (4 ESRs).
WP3 develops a system that monitors the driver’s vigilance level and automation-mode awareness in real time. The ESRs adopt a cognitive modelling approach to understand the relationship between eye gaze and driver perception. Based on eye-scanning behaviour, the system should be able to estimate whether the driver is at risk of collision under high automation. A demonstrator will be developed together with WP2.

WP4. Predicting real-world effects of highly automated driving (3 ESRs).
In WP4, microscopic traffic flow models are developed that predict the safety, stability, and efficiency of traffic flows as a function of the penetration rate of highly automated vehicles. The focus is on determining within- and between-driver variability when interacting with automation, and on integrating this knowledge into realistic traffic models.

WP5. Legal and market perspective of highly automated driving (1 ER).
In close cooperation with industrial partners, working groups, and European and national authorities, the ER reviews the legal perspectives and liability dilemmas in HAD, and proposes a framework for market introduction. Surveys and interviews are the prime research methods in WP5.

Work performed
The researchers were recruited between January and September 2014, and all ESRs are pursuing a PhD degree. After about 1.5 years of research, all researchers have now completed literature reviews on their topic. Most researchers have performed an experiment using volunteer participants, and have completed secondments at one or more project partners. Moreover, the ESRs have presented their results at conferences and workshops. The HFauto consortium was particularly well represented at the 6th International Conference on Applied Human Factors and Ergonomic held in July 2015 in Las Vegas. Moreover, the ESRs have been involved in various outreach and training activities, including biannual project consortium meetings. A full list of scientific journal publications, conference presentations, and outreach activities is available at Moreover, hardware and software integration has been performed towards the demonstrator (WP2 & WP3).

Main results achieved so far
In WP1, a literature review has been completed in which the authors proposed a psychological model of automated driving. Additionally, a literature review on human factors of transitions in automated driving has been completed. This review showed that experiments thus far have been mostly concerned with ‘automation-initiated driver-in-control’ (AIDC) transitions in ‘take-over’ scenarios. The review made clear that not only AIDC transitions, but also driver-initiated transitions (DIAC & DIDC) should be considered. Accordingly, the review proposed a classification of transitions based on who initiates and who takes control (human vs. automation), and whether the transition is mandatory or optional. Furthermore, a driving simulator experiment on the effects of platooning on workload, stress, and fatigue has been completed, as well as a study into how much time drivers need to perceive complex traffic situations. Both experiments made use of eye tracking to determine how participants distribute visual attention. The results of the experiments question prevailing dogmas in the human factors literature regarding how long drivers can sustain attention in automated driving, and about required take-over times.
In WP2, two literature reviews have been completed on vibrotactile interfaces, as well as a review and Internet-based questionnaire on auditory interfaces in HAD. These articles show that vibrotactile and auditory feedback hold promise in HAD, because they can attract the attention of a visually distracted driver. The reviews concur that feedback should be multimodal instead of unimodal, especially in urgent situations (Fig. 1). A cross-institutional driving simulator experiment examined the effects of vibrotactile, auditory, and multimodal take-over warnings on driver reaction time in an urgent AIDC transition scenario. Consistent with the literature reviews, steer-touch reaction times were slightly faster for the multimodal feedback (Fig. 2). In addition, several HMI concepts are under development, where the driver is continuously informed via haptics and a computerized “chatty co-driver”.
In WP3, literature reviews have been completed on driver state monitoring and on psychological vigilance in driving. The reviews made clear that results of laboratory-based vigilance studies (the ‘Mackworth Clock’) do not generalize to the open road because driving is a complex task. This signifies that the design of the driver state monitor should be ‘ecological’ and not driver-centred. The ESRs cooperate to create a real-time driver state monitor using high-end eye tracking, body posture measurements using Kinect, in conjunction with a cognitive modelling approach called Cosmodrive.
In WP4, a driving simulator experiment has been performed to determine the effects of AIDC, DIAC, and DIDC transitions. Results showed that driving speeds decrease and headways increase after a transition, because drivers need time to react and press the gas pedal again. A second study was performed on a real highway using a longitudinal automated driving system. Moreover, an Internet-based questionnaire and another driving simulator experiment using event-related brain potentials have been completed. These studies have yielded knowledge regarding individual differences in driver workload, and information needs of automated driving as a function of traffic complexity and automation status. The results obtained from the simulator-based and on-road tests allow the researchers to determine the parameters to be used in predictive-valid traffic flow models.
In WP5, an international survey among 5,000 respondents has established that there are large individual differences in the willingness to buy an automated car. An interview study among 12 scientists specialized in human factors and automated driving was conducted to identify critical research challenges. The interviews showed that HMI design and the interaction between automated and manually driving vehicles are among the most critical human factors challenges. Moreover, an analysis of two datasets showed that uptake of existing automation, such as adaptive cruise control, has been slow in Europe, especially in countries with a low income. We expect our findings to be instrumental for stakeholders involved in the development of automated vehicles.

Expected final results and their potential impact and use
The expected results are (1) a comprehensive understanding of human capabilities and side effects of automated driving, both in monotonous and transient situations, (2) a HMI that optimally interacts with the driver of a highly automated car, for situations of different criticality, (3) an ‘ecological’ driver monitor that estimates the operator’s vigilance level and hazard awareness, (4) realistic traffic flow models that predict the effects of HAD on safety and efficiency, (5) a roadmap for market introduction of highly automated driving, and (6) trained researchers having the multidisciplinary and generalizable knowledge, skills, and vision required to address human factors challenges of automated driving.

Reported by



Life Sciences
Follow us on: RSS Facebook Twitter YouTube Managed by the EU Publications Office Top