Action-Meaning Networks - A Novel Methodology to Identify Unsafe Use of Driving Automation

Paper i proceeding, 2023

The transition from manual to automated driving holds the promise of safer traffic and numerous societal benefits. However, ensuring the safe usage of driving automation is crucial for realizing these promises and facilitating a successful transition. Introducing driving automation into vehicles presents challenges for drivers, often resulting in behavioural adaptations that undermine the anticipated advantages. Therefore, it is imperative to understand how drivers engage with driving automation systems (DAS) and the significance they attribute to them, as the meaning assigned to these systems significantly influences their usage. This paper presents a novel methodology aimed at investigating the utilization of driving automation by examining both users' actions and the meaning they associate with the system during operation and organizing them into action-meaning networks. To illustrate this methodology, it was applied to takeover and hand-over requests, on a previously collected dataset from an on-road study. By employing this methodology, usage patterns were revealed, enabling the distinction of how drivers' actions and comprehension evolved over time and with repeated interaction. Additionally, unsafe actions and meanings associated with the interaction with the DAS were identified, highlighting potential hazards in long-term or alternative usage scenarios. Based on the evaluation of the results, it is reasonable to conclude that this methodology has the potential to complement existing approaches. By uncovering usage patterns and associated meanings of users’ interaction with DAS, it has the potential to facilitate the development of strategies to ensure safe usage and promote a positive user experience with such systems.