The 737 MAX, a twin-jet airliner series made by Boeing, is
currently grounded by regulators in over 80 countries for an investigation into
whether the automated flight control (autopilot) system is safe. Should this concern the automotive industry?
Aircraft autopilot systems have less "unknowns" to manage than
even the current autonomous vehicle systems, pilots are selected and better trained than
drivers, the airline industry is more regulated, and yet it is now thought that, in multiple scenarios, an autopilot system that the pilots did not know was
present had triggered incorrectly.
Could we see similar problems with autonomous
cars? Will cars and their systems be able to clearly communicate with their drivers?
Cars will need less driver support as
autonomous systems become more capable, but for the foreseeable future, they will need a driver as a backup. We’re
not just talking about the Robotaxis of the future, these are cars that can be
purchased now, cars that are an over-the-air update away from being able to, in
certain circumstances, cover significant parts of a journey without driver
supervision. Are drivers ready for this?
For most journeys, the ‘driver’ will have no need to monitor
traffic situations ahead, vehicle locations around them, even the route they
are taking. But, in an emergency, the driver will need to take control of the vehicle,
understand which functions the vehicle is still handling and which it isn’t, which systems are active,
and take in why the vehicle needs assistance in the first place. Unfortunately, benchmarking has shown that
the way vehicle and driver interact with each other, the HMI, is currently woefully
inadequate for autonomous-capable vehicles.
When the autonomous system realises
that it cannot handle a situation, the vehicle will have to support the driver
in the handover task. It needs to be a controlled handover, a supported
takeover with gradual reduction in the support the system offers the driver. If
the driver does not assist, the vehicle needs to initiate a minimum risk manoeuvre, such as to slow down or come to a stop. It must
not simply plunge an unsuspecting driver into a situation that the vehicle itself
is unable to handle.
We cannot just hand over autonomous-capable vehicles to the
public and hope for the best. Driver education will need to be managed.
Vehicles, drivers and legislation will need to be optimised for these
Driving licences will need to be revisited. For example, in
the UK, student-drivers can choose between taking their test with a manual
gearbox or an automatic. The catch is that if you take the test in an
automatic, you are only licensed to drive an automatic vehicle, whereas the
manual licence allows both.
While it would sound like learning to drive an
autonomous-capable vehicle should only allow you to drive other autonomous-capable
vehicles, that should not be the case. Driving an autonomous-capable vehicle
is an additional skill on top of driving a normal vehicle, not a lack of skill.
Drivers will need to be trained for it. This is another challenge, as how do
you train someone to safely, suddenly take control of a vehicle?
Cabin-monitoring technology will be vital too. A driver
needs to know exactly what their vehicle is capable of in a situation and,
conversely, the vehicle needs to be able to tell what that driver is capable of
too. Ultimately, a good HMI is essential.
“Being on the cutting-edge of technology is not easy” says
Dr. Alain Dunoyer, Head of Autonomous at SBD Automotive. “There is no road-map for
a well-designed, well-implemented, autonomous-capable vehicle. There are risks,
opinions and limitations, but, for the OEMs that get it right, the opportunities
“By benchmarking the industry’s most innovative autonomous-capable
vehicles, finding what works and what doesn’t, and by applying the HMI Principles, we can help
OEMs to create an ADAS package that is competitive in the market and intuitive for
Alain Dunoyer, PhD
Head of Autonomous