Two-stage transition procedure reduces potential hazards on planned transitions in automated driving

Abstract

The feasibility of human-system transition in conditional automated driving systems remains an unresolved issue. These advanced systems necessitate support for planned transitions due to their frequent occurrence. Concerns in planned transitions include the lack of situation awareness and physical preparedness; specifically, drivers lack awareness of the traffic situation and their hands are occupied at the moment of transition. In this study, three types of two-stage transition procedures were investigated as potential solutions to these issues. In these procedures, the takeover request was issued after either a prenotification alert (notifying drivers of an impending takeover request) or a preparation request for takeover (requiring drivers to cease non-driving-related activities and monitor the traffic situation with both hands unoccupied for approximately 1 min before the transition). The effectiveness of combining road signs with the preparation request to facilitate takeover was also evaluated. The results indicate that a two-stage transition procedure incorporating a preparation request can mitigate potential hazards in planned transitions. This procedure resulted in shorter takeover reaction times and a lower rate of lane change failures post-takeover compared to the single-stage transition procedure. The addition of road signs to the preparation request further enhanced takeover support. However, the two-stage transition procedure was ineffective when only a prenotification alert was provided without a preparation request. Furthermore, the gaze behaviour of the drivers during the preparation period suggested the development of situation awareness for a safe and smooth takeover. The trend in gaze behaviour shifted from “busy and widespread” to “stably and centralised,” occurring within approximately 20 s. These findings may contribute to the design of automated driving systems, the planning of road infrastructures, and the development of driver monitoring systems.