Many automotive OEMs are currently developing automated driving functions that no longer require driver supervision in certain driving environments. If system boundaries are reached the driver has to retake manual control. The system needs to provide a sufficient period of time in which the driver can safely and comfortably regain control over the vehicle. To determine a period of time for such a transition, relevant concepts from the field of automation psychology were presented and applied to the automotive context. Additionally an analysis of existent studies concerning take-over transition times from highly automated driving to manual driving was carried out to gain first insights into the time required for this transition. Building on the analysis of the literature and based on interviews that were conducted with automotive OEMs and suppliers, complex take-over scenarios and secondary tasks were developed which represent realistic and relevant conditions for future automated cars. These scenarios and secondary tasks were implemented in a static driving simulator and tested with N = 60 test drivers aged 20 to 76 (M = 36.0, SD = 15.2) years. After an automated drive of approx. 5 minutes, during which the drivers were distracted by motivating game on a handheld tablet-pc, the drivers were generally able to regain manual control over the vehicle. This secondary task was compared to a manual drive, in which the take-over request was substituted by an auditory warning, to an automated drive, in which a different task was presented on a tablet-pc, as well as to an automated drive, in which the participants supervised the automation. It could be shown, that after a highly distracted drive 90% of the participants first had their eyes on the road after 3-4 seconds, had their hands on the steering wheel and their feet on the pedals after 6-7 seconds and disengaged the automation after 7-8 seconds. When analyzing the times taken to look at the side mirror and the speedometer as indicators of situation awareness during this driving situation, 12-15 seconds are needed. These reactions, which are necessary for the comprehension of the traffic situation, are therefore up to 5 seconds slower for the distracted automated driving compared to manual driving. For some automated drives collisions and critical situations were recorded, although they were comparably frequent in the manual drive condition. These critical situations might have been avoided through the use of suitable driver assistance systems that stay active and support the driver after the driver has disengaged the automation. Additional reaction times and driving data was analyzed to gain further insights into the process of the take-over transition process. Driving in a real world scenario may change the reaction times compared to driving in a simulator. Also, the effects of longer automated drives, sleepiness and more immersive secondary tasks on take-over process and reaction times are not yet clear. Some of these issues will be addressed in a planned simulator study. If there is to be a gain not only in comfort but in safety for automated driving compared to manual driving, drivers need to be not only sufficiently forewarned before takeover situations, but must additionally be supported by suitable assistance systems before, during and after the transition to manual driving. (Author/publisher)
Abstract