Future vehicles will combine different levels of driving automation characterized by varying responsibilities for users. This development will intensify system complexity which poses the risk of confusing the driver. We hypothesize that the users’ mode awareness suffers especially when changing from Level 3 “Conditional Automation” to Level 2 “Partial Automation”. Therefore, automated systems need to be designed in a way that minimizes confusion with regard to the automation mode. The article describes the influence of a gaze direction based Attention Request (ATR) to avoid mode confusion with the aim of contributing to the reliable operation of different levels of automation in one vehicle.

Two similar studies were conducted. One took place in a dynamic driving simulator with 40 participants. Every participant drove for 10 minutes with a partially automated driving (PAD) (SAE level 2) system and conditionally automated driving (CAD) (SAE level 3) system in the order PAD/CAD/PAD. The second study was conducted on a German highway in a Wizard-of-Oz car. All 40 test persons drove in each PAD and CAD phase 8 minutes in the order of PAD/CAD/PAD/CAD/PAD/CAD. The CAD-system was in both studies a high performing Hands-Off Level 2 system that required no input of the driver. To promote the same mental model for all participants as it is a requirement to measure the differences in mode awareness, all persons became a detailed description of the Level 2 and 3 systems presented by video and text.

Both studies used a between-subject-design to measure the influence of an ATR. The ATR was based on the gaze direction of the driver and initiated by the investigator when the drivers gaze was not in the street AOI for longer than 4 seconds. Mode awareness was operationalized by the visual attention towards driving-relevant areas, a qualitative analysis of a questionnaire and followed by an interview.

The ATR was proven to be an effective action to maintain the mode awareness by using a level 2 and 3 system within one car. Specifically, the visual attention did not decrease by an intermitted CAD drive during PAD. Moreover, the visual attention to the road scene increased for the group with an ATR during PAD. This was indicated by the measurement of a significant interaction effect for the development of the visual attention to the road scene for the groups with and without ATR. Thus, the gaze direction based ATR was proven to be an effective measure to maintain mode awareness, if different levels of automation are combined in one vehicle. This result helps to take the next step for realizing such combined multilevel systems with tailored HMIs for advanced driver assistance systems.

Moreover, it has to be considered, that the studies put the emphasis on the first glance of the drivers, during their first contact with partly and conditionally automated systems. Further studies should investigate the long term effect of an ATR.