Scents and self-driving carsBy Dmitrijs Dmitrenko, Sussex Computer-Human Interaction (SCHI) Lab - 9th September 2019
Dmitrijs Dmitrenko from the Sussex Computer-Human Interaction (SCHI) Lab explains how scents can be used to convey driving-relevant information, and asks the question – can they make driving safer?
In-car smell is a big topic in the automotive industry. Scent diffusers are becoming increasingly popular in modern vehicles, but their current function does not go beyond the enjoyment of the driver and/or passengers. Academic research is investigating scents in the context of Advanced Driving Assistance Systems (ADAS) and autonomous vehicles (AV). Recent findings have demonstrated that scents can be used to convey driving-relevant information, but can they also make the process of driving safer?
Driving is a highly visual task. Nevertheless, it is a process that involves other senses too. When we drive, we touch the steering wheel, we listen to what is happening around us, and, even if we are not paying attention to that, we smell what is happening – a scent of gasoline, burning rubber, plastic heated by the sunlight, for example. In fact, smell is able to provide a much more vivid experience than any other human sense.
Moreover, the sense of smell could deliver benefits in terms of safety. If we see a "service due" light on the dashboard, we might ignore it, but if we smell something burning, we would probably stop immediately. Self-driving cars could potentially reduce visual demand, allowing the driver to focus on a secondary task. Nevertheless, even in such scenarios, the driver might need to monitor the automation performance. Could this requirement be minimized with the help of scents and, most importantly, would this contribute to the safety of driving?
Such a benefit of using olfactory stimulation in the car as a strong link to memories and emotions has been the motivating factor for the recent advancements in the automotive industry, where multiple manufacturers, including Mercedes-Benz and BMW, have created their own in-car scents to enhance hedonic driving experiences. Moreover, research has shown that scents have a positive impact on the alertness and mood of the driver,1 drivers’ braking performance,2 and on keeping drowsy drivers awake.3 However, the interfaces used in these examples are delivering only ambient scents, without making use of the ability of scents to provide hints to the driver. This is happening despite the potential of scents to convey information, which is evidenced by findings in psychology and neuroscience.
Scents have been demonstrated to mediate cross-modal correspondences.4 They are able to support object localization and identification tasks, as well as help capture people’s attention. If olfactory notifications are used to notify the driver about an upcoming danger (e.g. a pedestrian suddenly appearing in front of the car), then a carefully chosen scent could make the driving safer. For example, one of our studies has shown that the scents of rose and peppermint result in a significantly lower number of crashes than when dangerous events are signalled using the scent of civet (Figure 1). This was true for drivers in an induced angry state.
Anger is a dangerous emotion to experience when driving a car, as it may lead to road traffic accidents. The study mentioned above has also shown that the scents of rose and peppermint improve the mood of drivers, suggesting that they could become less angry if they are exposed to scents like rose and peppermint throughout the driving. Consequently, this might help maintain safer driving behaviour.
In a context of autonomous driving, olfactory notifications could help the driver monitor the self-driving car better, employing an olfaction-enhanced reliability display. A recent study5 demonstrated that there is an improvement in the secondary task performance (i.e. interacting with a mobile device) when scent-based reliability displays are in use. The scent of lemon was used to notify the driver about a switch from high to low reliability, and lavender signalled a switch from low to high reliability. Moreover, such displays resulted in a better trust in the AV, while contributing to its ease of use and perceived usefulness.
In conclusion, research provides evidence for the ability of scents to make ADAS and AVs safer to use. Scents could help us become calmer on the road and maintain safer driving behaviour. In a context of self-driving cars, scents could support the task of monitoring the reliability level of the automation (i.e. prepare the driver for a potential emergency), while also making us more likely to keep using such systems. This suggests that we should continue exploring olfactory interaction and expanding the range of its application scenarios in the car.
1. Raudenbush B et al. N Am J Psychol 2009; 11:245–56.
2. Martin G, Cooper J. In The British Psychological Society Annual Conference 2007.
3. Oshima C et al. In Workshop on DSP for in-Vehicle and Mobile Systems 2007.
4. Castiello U et al. Chem Senses 2006;31:665–71.
5. Wintersberger P et al. In IUI ’19 (2019), ACM, New York, NY, USA.
Dmitrijs Dmitrenko, Sussex Computer-Human Interaction (SCHI) Lab, Creative Technology Research Group, School of Engineering and Informatics, University of Sussex, Brighton, UK