Remote Operation in Teleoperated Railways: Impact of a Supplementary Task, Attention-Guiding Take-Over Requests and Work Experience on Situational Awareness and Workload

Abstract

Background: Remote railway operations can improve efficiency and reduce costs, but operators must still intervene when onboard systems reach their limits, requiring high levels of situational awareness (SA). Passive monitoring can affect SA, especially under low workload conditions. This study on SA and workload in remote train operation explores the effects of a supplementary task (ST) intended to maintain engagement, potentially at the cost of SA. It also examines whether attention-guiding takeover requests (AGTORs) and train-driving experience can mitigate these effects. Methods: A mixed design online study was conducted with 160 participants, including 28 active train drivers. Participants completed two blocks, one with an ST and one without, and were randomly assigned to receive either AGTORs or simple warning tone takeover requests. Each block featured four eight-second video scenarios in which participants had to identify the reaons for the takeover and assess additional information from a dynamic remote environment. SA was measured using takeover reason recognition and SAGAT questionnaires. Additionally, workload after each block and a number of additional variables were assessed. Data were analyzed using mixed linear and logistic regression models. Results: Engagement in the ST reduced recognition of takeover reasons but increased SAGAT scores. The ST condition also led to higher workload scores, though levels were above optimal across both conditions. AGTORs significantly improved recognition performance regardless of the ST condition. No significant performance differences were found between train drivers and non-train drivers. Additionally, age was negatively associated with SA, and participants evaluated the remote environment favourably. Discussion: The study highlights the benefits and challenges of STs in remote train operations. Working on an ST led to higher SAGAT scores, suggesting a benefit for maintaining perception-level SA. However, it also resulted in fewer recognized takeover reasons, potentially reflecting task-switching costs and reduced comprehension-level SA. A promising design direction to tackle this might be AGTORs, that, in this study, improved takeover reason detection without affecting overall SA. Previous experience did not show beneficial effects, but this might be due to the small train driver sample size. Future research might investigate the effects of experience, using a larger sample and more complex scenarios. Overall, this study provides a valuable insight into the challenges associated with remote train operation and offers solutions in the form of maintaining engagement through STs and compensating switching costs with AGTORs to support operators in takeover situations.