Human-Machine Interface

The information and warnings judged to be relevant for the driver are given via the Human- Machine Interface (HMI). The HMI can rely on visual, acoustic and haptic channels for conveying information to the users, used either separately or combined. In the latest years, the in-vehicle’s HMI has evolved by incorporating in the car some of the functionalities available in smartphones, i.e. allowing mobile devices to be operated through the car’s dashboard screen. Android Auto is the standard developed by Google, while CarPlay is the one developed by Apple. Mirrorlink is an intermediate solution between both platforms which is compatible with either of the two operating systems, Android and iOS. Among the functionalities offered, there is navigation, messaging, music playback, internet search. This increases in-car connectivity and allows a plethora of information and entertainment services available to the driver and passengers. The future automated driving will allow for further digital interactions, profiting from driver’s transition from driving to non-driving tasks and making use of innovative forms of interaction (e.g. augmented reality, gestures recognition) (Boyadjis, 2015; see Figure 18).

Figure 18. The interconnection of HMI and automated driving (according to NHTSA levels)

Source: Boyadjis, 2015.

It is in the intermediate levels in which the DDT is performed by both the driver and the vehicle that important Human-Machine Interaction (HMI) concerns emerge. As defined by the International Harmonized Research Activities (IHRA) Working Group on ITS (IHRA WG ITS, 2010), the notion of driver-in-the-loop means that a driver is involved in the driving task and is aware of the vehicle status and road traffic situation as an active player of the driver-vehicle system. On the contrary, out-of-the-loop means that the driver is not immediately aware of the vehicle and the road traffic situation because he/she is not actively monitoring, making decisions or providing input to the driving task (Kienle et al.,

2009). The HMI has a key role in keeping the driver in the loop, by informing the driver of required status information transparently and unambiguously. This means that systems need to be designed to detect the limits of their own range of effectiveness during highly automated driving and clearly inform the driver (visually, haptically and acoustically) with sufficient advance time to be able to resume the driving task (15_{). A recent study found} that resuming control in automated driving requires up to 5 seconds more than conventional driving (Kühn, 2016). Specifically, the authors found that after a drive with a high level of distraction, 90% of the drivers looked at the road again for the first time after 3-4 seconds, had their hands back on the steering wheel and their feet on the pedals after 6-7 seconds and had turned off the automated system after 7-8 seconds. Even drivers who were not distracted at the moment where the request of intervention came, had delayed reactions compared to users in normal manual driving.

When it comes to proposing best practices for AVs HMI, the following recommendations can be highlighted (Kühn, 2016):

— The driver needs to be notified as early and clearly as possible of the need to resume vehicle control (preceded by an early identification of the need to transfer the vehicle control). The takeover period must last longer than 8 seconds.

— The automated system must remain active during the takeover process, until the driver has shown readiness to take over vehicle control.

— A minimum risk manoeuvre has to be put in place if the driver cannot handle the control takeover request.

— Comprehensive but succinct information on the current driving situation needs to be provided in order to facilitate the driver’s situational awareness after an automated drive.

— The vehicle readiness to assist after the driver has resumed control needs to be increased to avoid inappropriate reactions of the driver.

— To show the urgency of a given situation, a cascade of different types of warnings could be issued to the driver.

— Instructions about capabilities and limitations of automated systems could be specifically given to drivers for better user reactions in the event of a control takeover request and to avoid wrong or reduced system use.

Standardising HMI could be beneficial to minimise the risk for users’ misunderstanding/confusion when using different AV models, especially in consideration of the car sharing trend.

About interaction with other road users outside the vehicle, a 2015 study (Lagström and Lundgren, 2015) on the pedestrian - driver communication provided the following interesting insights. Pedestrians need to know when a vehicle is in automated mode, given that the decoupled driver’s inattentive behaviour can otherwise be interpreted as uncertain and dangerous and may as a consequence impede the pedestrian to cross. This investigation proposed an external HMI prototype with LED strip lights showing different sequences depending on the AV intended manoeuvre: about to yield, about to start, resting or in automated driving mode. The prototype evaluation delivered promising conclusions as regards the suitability of having an external HMI. The users who participated in this study were able to understand the interface messages after a short training. They reported that the interface could replace the role of the driver and even excel today’s interaction as the communication was clearer and available earlier.

NHTSA also provided a set of HMI considerations (NHTSA, 2016a). Constantly showing the system status is highlighted as a minimum requirement for AV systems (e.g. whether it is properly functioning or requesting a driver’s intervention). In fully automated systems, the

(15_{) See VDA’s Networked and automated driving site, https://www.vda.de/en/topics/innovation-and-} technology/automated-driving/automated-driving.html.

need to design an HMI that accommodates for people with disabilities is explicitly indicated. For fully automated vehicles that operate without any humans in it (e.g. automated delivery vehicles), the central control authority or remote dispatcher should know the status of the operated AVs at all times. Driver engagement monitoring for those systems that may require drivers to regain vehicle control is suggested. Equally relevant, the consideration of signalling vehicle intentions to other road users such as pedestrians, cyclists and other vehicles. Since this is an area which is rapidly evolving, a special mention is given to considering the guidance, best practices, and design principles published by SAE International, ISO, NHTSA, American National Standards Institute (ANSI), among others.

Further research in this area may be required to better understand the AV interface needs.