Let’s CHAT! Check, assess, takeover : Exploring the necessity and effectiveness of behavioural interventions to guide drivers during planned control transitions in vehicles with conditional driving automation.
Publication Date
December 9, 2025
Creator
Abstract
The introduction of automation technology to vehicles within the public domain promises to revolutionise everyday travel with anticipated benefits of improved road safety, comfort and mobility. Vehicles equipped with conditional driving automation, (or Level 3 (L3) automated driving systems (ADS), will be able to take control of the full driving task, allowing drivers to turn their attention towards non-driving related tasks (NDRTs) for extended periods. This automation capability is bounded by specific conditions, meaning its operation will not cover all end-to-end journeys. Therefore, drivers will be required to seamlessly switch between automated and manual driving while the vehicle is in motion following a controlled request from the automation. The introduction of vehicle automation fundamentally changes the context of the driving task overall and these planned control transitions represent a new sub-task for the human driver. However, there is a tacit assumption that experienced drivers will be able to take on any new tasks and expectations required to safely drive vehicles fitted with this capability without any additional guidance or support. The balance between the technological benefits and consumer desire brought about by the use of automated driving systems; and the maintenance of driver performance and comfort in a safety critical system is important. Therefore, this thesis proposes to test the assumption of driver competence in the use of conditional automation and consider the design and evaluation of behavioural intervention solutions aimed to encourage desired driver behaviour during planned transitions of control.
The work within this thesis details three studies (one observation study on the dataset of previous work and two evaluation driving simulator experiments) and two sections detailing the design of driver behavioural interventions to support planned control transitions. To begin, an observation study was employed to analyse the behaviour of 49 drivers who had taken part in a 5-day, longitudinal simulator study investigating control transitions in a vehicle, equipped with conditional automated driving systems. Analysis focused on driver behaviour during four planned and one emergency control transitions. Results revealed a lack of understanding from drivers on the shared control requirements during planned control transitions, and the checks and procedures necessary to safely and adequately prepare to resume manual driving. Drivers focused on improving performance in resuming operational control of the vehicle, neglected tasks associated with monitoring and planning future movements and prioritised extending shared attention with NDRTs prior to resuming manual control of the vehicle.
Findings from this foundational study were used to form an evidence-based, theoretical justification for the necessity of behavioural guidance for drivers using future vehicles equipped with conditional automated driving systems. In particular, highlighting a need to conduct more research on planned control transitions. Literature, including research from the aviation domain, alongside the analysis of driver behaviour during the observation study inspired the concept and development of a standard operating procedure (SOP): Check, Assess, Takeover (CHAT), as a way to standardise and specify the manner (and order) that drivers should undertake tasks during a planned control transitions. This CHAT procedural framework was used to inform the design and development of two behavioural intervention solutions: the Behavioural ‘CHAT’ training and the CHAT Human Machine Interface. These interventions were subsequently evaluated.
A driving simulator study was designed to evaluate the effectiveness and necessity of the behavioural “CHAT” training intervention, in a between-subjects design, with 24 participants. The CHAT behavioural training intervention was evaluated against an operational training intervention based on a user manual for a commercially available vehicle fitted with automated technology. Driving performance, glance behaviour and subjective ratings of trust, workload and situational awareness were assessed. Evidence suggested that behavioural “CHAT” training had a positive influence on tactical level task performance during automated driving and following a transition to manual driving. In particular findings suggested a positive impact on visual behaviour and complying with protocols to discharge secondary devices sooner following a takeover request when compared with drivers who were given a more traditional training method (reading a user manual). Thematic analysis carried out on interviews conducted with the experienced drivers who participated in the study revealed further evidence that many drivers do not accurately understand the level of engagement with the driving task required during a planned control transition from automated to manual driving. Findings supported the view that, without adequate training or guidance, drivers find it hard to remember or, in fact, recognise the importance of attention allocation to the external driving environment during transition of control. However, responses from drivers who undertook the behavioural CHAT training suggested that it was effective in raising the need for glance behaviour that provided ongoing sampling of the external driving environment to support drivers regaining situational awareness ahead of taking over the driving task.
Finally, a driving simulator study with a between-subjects design was carried out (n=24) to investigate the effectiveness of the CHAT HMI over 3 consecutive drives when compared with a basic HMI design that gave drivers the same 60 second timeframe to prepare to drive, with audio-visual temporal cues ahead of manual driving mode engagement. Evidence demonstrated that drivers using the CHAT HMI followed the SOP, using the initial part of the transition period to discharge their devices and get themselves in position for driving, for example, checking feet position, during the first 10 seconds of the transition period. These drivers then engaged in significantly more monitoring and tracking activity engaging with the front, rear and peripheral road environment more often than the Basic HMI drivers ahead of taking physical control of the vehicle. Monitoring behaviour was maintained more consistently across the course of the transition period by the CHAT HMI and this behaviour was consistent across the three drives. The Basic HMI group were more reactive with their glance behaviour showing temporary spikes in monitoring the external environment if something caught their attention, such as vehicles joining the motorway, but not maintaining a focus on consistent situation monitoring behaviour ahead of manual takeover. The Basic HMI group were also significantly more likely to make glances towards the HMI during the first 10 seconds after manual driving was resumed and even drivers who engaged in more rear and peripheral checks during the transition period demonstrated poorer driving performance behaviours after manual takeover, such as extended glances away from the front roadway (>2 seconds) or insufficient checks prior to changing lanes. Glance behaviour and driver performance showed that the early engagement in situation monitoring demonstrated by drivers with the CHAT HMI led to improved performance post manual driving resumption. For example, drivers who did not receive the CHAT HMI intervention were statistically more likely to be involved in a near-crash incident during a high task demand drive than those in the CHAT HMI group.
This thesis demonstrates the necessity to provide experienced drivers with additional guidance to train desired driver behaviours for the safe resumption of control of tactical as well as operational driving tasks required for planned transitions in L3 automated vehicles. The findings provide evidence that the use of a standard operating procedure to guide necessary actions in a particular order before taking over physical control of the vehicle is an effective intervention that can positively influence driver behaviour and resulting performance and comfort, during and after planned automated to manual driving transitions. The evidence provided has direct implications for the design of driver training interventions and in-vehicle HMI devices for both experienced and novice drivers of vehicles with conditional driving automation capability.
Item Type
ethesis
Thesis Type
PhD
Supervisors
Subjects (LC)
Associated Schools / Departments
Faculty of Engineering
Department of Mechanical, Materials and Manufacturing Engineering (UK)
eprints ID
83146
UoN Repository URI
Except where otherwise noted, this item's license is described as
File(s)![Thumbnail Image]()
Name
Emily_Shaw_PhD_Thesis_post_viva_FINAL_format_change.pdf
Type
Full-text
Description
Examined
Size
5.49 MB
Format
Adobe PDF
Checksum (MD5)
eeca720e9d896f65bdd45c4a6113a818