Drivers can take 25 seconds to resume control from driverless cars, suggests study

Sprinters reacting to starting gun in race

A new study of drivers’ reaction times has found that the ‘takeover time’ – the time needed for a human driver to take back control from a driverless car – can last as long as 25 seconds.

A team at the University of Southampton demonstrated that the transition times taken for participants to switch from automated vehicle control to manual control fell into a very broad range – lasting from just 1.9 seconds to as long as 25.7 seconds.

We hope our findings can guide policymakers in setting guidelines for how much lead time a driver will need when changing in and out of automation
— Professor Neville Stanton, University of Southampton

During the experiment, conducted in the University’s driving simulator, participants were assessed on their ability to resume control of a car operating automatically, under non-critical conditions.

The study’s authors, Professor Neville Stanton and Alexander Eriksson, said that the wide range of reaction times reflected a variety of driver behaviour and environmental conditions.

During the study, participants engaged in simulated driving at 70mph in normal motorway driving conditions.

Drivers were asked to drive both with and without a potentially distracting non-driving task to perform, with their reaction times recorded as they took over or relinquished control of the automated driving system. 



The study found that drivers engaged in the secondary task took longer to respond – posing a safety hazard.

“We hope our findings can guide policymakers in setting guidelines for how much lead time a driver will need when changing in and out of automation,” said Professor Stanton. 

“The challenge for designers is accommodating the full range of response times rather than limiting parameters to mean or median transition times.”

In particular, say the researchers, there is a risk of unintended consequences if the lead time for normal handovers is based on data taken from testing in critical situations. 

“Too short a lead time, for example seven seconds prior to taking control, as found in some studies of critical response time, could prevent drivers from responding optimally,” said Eriksson. 

“This results in a stressed transition process, whereby drivers may accidentally swerve, make sudden lane changes, or brake harshly. 

“Such actions are acceptable in safety-critical scenarios when drivers may have to avoid a crash, but could pose a safety hazard for other road users in non-critical situations.”


DriverlessGuru commentary

What do the findings mean for the development of driverless cars?

The study’s findings are likely to be of particular interest to manufacturers developing ‘conditional automation’ models – those where a human driver is expected to respond to a request to intervene by an automated driving system.

Such models require the driver to remain alert to the driving environment and be ready to resume control at any time. So the idea that it could take a human driver almost half a minute to do so will pose a challenge as these technologies become more common.

Some manufacturers may simply advise drivers to stay alert – much like Tesla has with its Autopilot advanced driver assistance system. Even with Autopilot engaged, drivers are expected to keep their hands on the wheel and maintain control and responsibility for their vehicles.

However, as driverless technologies improve and the number of requests for manual intervention falls, carmakers will also have to take into account human nature – and the likelihood of distraction if drivers are lulled into a false sense of security at the wheel.

The promise of fully autonomous cars suggests motorists will one day be able to hand over all driving responsibilities and enjoy extra leisure time in their vehicles.

In the meantime, manufacturers must strike a balance between relieving drivers of some of that driving task, and ensuring they remain alert enough to take over when required.