International Transport Forum: Automated Driving and Traffic Safety4/27/18
Indeed, the added value in improved productivity and efficiency that can be extracted in commercial transport operations appears to be substantial.
But, the question of the impact of automation on road safety tends to divide the public in two distinct camps – one lauding the life-saving potential, and the other envisioning a future of self-aware vehicles running wild, causing death and destruction.
So, where are we heading? Well, I believe the reality of the challenges involved is becoming apparent as more automation pilots are deployed and experience accumulates. Improving road safety on a system-wide level will prove challenging. Yet, under favorable circumstances automation concepts are showing great safety potential. Favorable is a key word here.
The vast majority of incidents and crashes involve human behavior. It is easy, then, to suppose that taking the human out of the loop will reduce crashes correspondingly. Yet, I think intuitively we understand there is something wrong with this logic. And the reason is most driving involves no crashes. An attentive person is fantastic at anticipating events in complex situations – much better indeed than any AI outside of Hollywood.
What strikes me in the debate on automation is the tendency to confuse automation and safety – especially, I fear, among those supposedly well-informed in the area. It is easy to be tricked by the fact that active safety and automation share technologies. Naturally, there will be no successful deployment of automation unless it is safe. But, the development of one does not follow naturally from the other. Both areas require dedicated efforts.
In conventional driving, the human driver performs both operational maneuvers and the highly intricate task of anticipating and mitigating critical situations. The objective of safety is largely to introduce barriers – conceptual and real – to minimize the consequences of mistakes and errors.
The challenge for a highly-automated driving system, then, is to accomplish both the basic task of driving and anticipating and avoiding emerging critical situations. In other words, it must be able to avoid or mitigate any situation it can reasonably encounter in its operational domain – not just mitigate the rare instances missed by a human driver.
Do I sound skeptic about automation? Well, in fact, I am quite the opposite. I believe vehicle automation will make great contributions to road safety. But, it needs to be done right – with safety as a primary design factor. This means vehicles need to behave safely – be sensibly cautious and use good margins.
It also means the set of conditions under which the vehicle can safely operate is actually an integral part of the solution itself. In other words, the when, where and how, are just as important as the what. With joint effort we have a chance to develop conditions on a system level that are favorable for safe automation.
It may mean starting in confined – or by other means controlled – areas, and then working systematically to increase the number of applications viable for automated driving.
Done right we will be able to reap all the benefits automation offers in terms of productivity and efficiency while enhancing safety.
That’s the Volvo way.