Could driverless cars make us worse drivers?

With approximately 90% of road traffic accidents attributed to human error, it is widely assumed that driverless cars will lead to improved road safety.  However, a new study at the University of Nottingham suggests that Level 3 automated vehicles may have an adverse effect on driver performance.

Conditional automation

While industry is pursuing the goal of fully autonomous vehicles, the first 'driverless cars' that are likely to be seen on public roads will not be fully autonomous but rather offer conditional automation.  These vehicles, also referred to as Level 3 vehicles (on the Society of Automotive Engineers classification), are capable in certain scenarios of controlling all dynamic, non-strategic, driving activities but expect the human driver to intervene if required.  Examples of this technology include traffic-jam assist.

The issue of transferring control between the vehicle and the human driver – known as 'handover' –has proved controversial.  In particular, there is a significant debate as to whether this can be done safely as a matter of design.  A split has emerged in industry as a result, with some manufacturers seeking to jump straight to Level 4 automation ('high automation) on the basis that it cannot.

The study

The study – How will drivers interact with vehicles of the future? – involved a group of 49 experienced drivers undertaking a 30-minute commute-style journey in a longitudinal driving simulator at the same time on each of five consecutive days.  The journey involved periods of manual and automated driving.  Each participant also brought with them their own devices and artifacts to use.

The aims of the study were to explore:

• the type of activities that drivers chose to undertake in a vehicle offering Level 3 automation;
• what impact these had on the manual resumption of the driving task under both routine and emergency conditions; and
• how participants’ behaviour changed during the course of the week.

The results

 The results of the study showed the following:

• Drivers were quick to develop high levels of trust in and acceptance of automation and undertook a range of activities while the vehicle was in control, with smartphones being the most popular device;
• The proportion of time spent on non-driving activities increased over the course of the week, with less than 20% of time spent directed at the road scene by the end of the trial;
• Initial driving performance after handover was poor with drivers swerving and varying their speed for up to 10 seconds;
• Although driving performance after handover improved over the week so too did driver complacency with an increased tendency among drivers to use the 60 second prepare-to-drive notification period to casually dispense with their secondary activities rather than actively preparing to drive;
• Human-machine interface (HMI) appeared to play an important part in facilitating efficient handover, with a feedback HMI – showing the current status of the automated control system – reducing the time to 'driver readiness'.

The study, a full copy of which is available here, provides interesting insights for how people will interact with Level 3 vehicles in practice nd, critically, how that may impact on their ability to take safe control of such vehicles on handover.  This will be relevant to both policy makers and industry, particularly in relation to designing vehicles in a manner that satisfies a manufacturer's duty of care to consumers.

Interestingly, the authors of the report also conclude that the results indicate a strong need for new driver skills when it comes to vehicles with automated functionality, to ensure that drivers efficiently monitor system status displays, and are able to resume control smoothly and safely.