General Insurance Article - Considering the Risk of Self-Driving Cars

 By Tim Marlow, Head of Autonomous & Connected Vehicle Research at Ageas

  

 Self-Driving Cars – Risk Considerations for Forthcoming ‘Level 3’ Conditional Automation Vehicles

  

 There are a number of definitions of vehicle automation or autonomy but probably the most relevant is the six level system defined by The Society of Automotive Engineers (SAE) – from Level 0 (a manual vehicle with no automation) through to Level 5 (Fully Automated with no manual input). We need to understand how these levels of autonomy impact risk and whether risk actually reduces as automation increases.

  

 We are already seeing a number of vehicles with Level 1 systems such as autonomous emergency braking (AEB) and lane-keeping assistant (LKA). Level 2 systems, such as Traffic Jam Assist, combine longitudinal and lateral automated control so that both braking and steering can be used to mitigate or avoid a crash at low speeds. These ‘driver support systems’ reduce claims frequency and costs and, as the driver retains ultimate authority over the vehicle, the liability for any accident stays with the driver.

  

 At SAE Level 4 (High Autonomy), vehicles will be capable of taking over the driving task entirely during pre-planned segment(s) of a journey. The vehicle will offer autonomous control when certain preconditions are met and will also make the driver aware of the point that control will need to be handed back. As the vehicle approaches the end of the pre-planned segment, it will remind the driver they need to take over and (probably) initiate a countdown. Should the driver not take back control, the vehicle will drive itself to a ‘safe harbour’, such as a lay-by or hard shoulder, where it can stop safely until the driver is ready to assume control.

  

 At the top end of the scale, Level 5 allows the entire journey to be undertaken without the involvement of a human driver. Vehicles with this level of automation are unlikely to be available commercially until some way into the 2020s.

  

 But it is cars at SAE Level 3 (Conditional Autonomy) that will start to appear in showrooms over the next two to three years and these may feature:

 The details are still being debated, but by 2018 a driver may be able to disengage from driving for up to fifteen minutes at a time, with the car taking over steering and speed control.

  

 All of this raises several concerns. Could the customer misunderstand the level of automation a Level 3 vehicle has? Is the ‘safe stop’ really safe? Active Cruise Control sensors currently have useable range of around 150m and can make decisions at perhaps 100m – a vehicle travelling at 81 mph covers this distance in just 3 seconds – realistically is this enough time for a driver to get back in control?

  

 Serious accidents could occur where the vehicle has dumped control back to the human driver and he or she has not been able to respond quickly and/or appropriately, not to mention disputes over liability because a manufacturer argues that the human driver should be in control while the driver/insurer argues that the vehicle systems should have been able to cope with the situation in the first place. A number of major accidents involving vehicles of this type could wipe out the safety reputation of ADAS and autonomous systems.

  

 To have safe roads we need safe systems but SAE Level 3 may potentially rely on car drivers reacting and taking control in two or three seconds – something we know stretches even highly trained pilots. Perhaps, from a risk point of view, it would be better for the manufacturers to jump straight to SAE Level 4 as some, such as Volvo and Ford, are now proposing.