Is text-based user manual enough? A driving simulator study of three training paradigms for conditionally automated driving

• Published in: Transportation research. Part F, Traffic psychology and behaviour Vol. 95; pp. 355 - 368
• Main Authors: Chen, Huei-Yen Winnie, Guo, Zhi, Ebnali, Mahdi
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.05.2023
• Subjects: Conditionally automated driving; Driver takeover; Driver training; Driver-automation interaction; User manual; Conditionally automated driving; Driver training; Driver takeover; Driver-automation interaction; User manual
• ISSN: 1369-8478
• DOI: 10.1016/j.trf.2023.05.006

•Simulator study compared training by written guide (WG), video guide (VG), and interactive guide (IG) for automated driving.•In written tests, WG scored lower than VG and IG groups in strategic knowledge, and lower than VG in procedural knowledge.•IG led to improved takeover performance (steering reveral rates) compared to WG.•IG group showed more visual monitoring behavior during low automation reliability periods compared to other groups. Training drivers about the automated driving system is one way for establishing safe and efficient driver interactions with a conditionally automated vehicle. This study investigated the effectiveness of three training paradigms—written, video (contains use cases), and interactive guides (support practices of use cases)—in a driving simulator study. Thirty participants with little or no prior experiences in driving automation were recruited and randomly assigned to participate in one of the training paradigms followed by a written test and two driving trips in a simulator. Explicit knowledge acquisition following training, driving performance during the simulated trips, and changes in acceptance of automation after training were analyzed. Results showed that: (1) there were no significant differences among the training groups in the declarative knowledge section of the test; however, the video guide group outperformed the written guide group in the procedural knowledge and strategic knowledge sections, and the interactive guide group outperformed the written guide group in the strategic knowledge; (2) Training type produced a significant effect on manual response times, but not visual response times, to non-scheduled takeover requests: the interactive guide group produced the shortest manual response times; (3) the interactive guide group performed better in takeover control during non-scheduled takeover scenario and (4) spent more time viewing the road during periods of low-reliability automated driving; (5) the written guide group reported decreased perceived safety after training. These findings suggest that the inclusion of use cases (in video and interactive guides) and providing hands-on practices (in interactive guide) have unique potentials in developing drivers’ knowledge in conditionally automated driving. This work has important implications for the effective design of conditionally automated driving training, as well as for the design of human-machine interface in automated vehicles to support users in developing and maintaining situation awareness.