A Road Map for Driverless Cars
Engineers have estimated that fully driverless cars could be on our roads from as early as 2030. A look at the literature has provided us with a road map of what to expect when the roll out does happen. This shows how during the early stages of the transition, driverless cars will be expected to operate normally, alongside their traditional driven counterparts in a ‘mixed fleet’ car scenario.
During this time the integration is likely to occur in two phases. The first of which will involve a shift from traditional, to ‘overridable’ self-driving cars. These will have a handover feature which gives control back to the human operator during critical situations which the onboard AI will not be expected to handle. The second phase will begin when overridable self-driving cars eventually begin to significantly outnumber the norm. At this stage we will expect to see fully driverless non-overridable cars introduced to the roads.
But before the idyllic end goal of completely safe driverless roads can be realized, it should be cautioned that the pathway to fully driverless roads will not be completely without risk. A note of caution which forms the focal point of this blog will consider the risks inherent to the transition to new driverless paradigms, especially when considering the contributory roles of human and machine error in our daily experiences.
Human Error: At Home on the Roads and Behind the Scenes
Today, we can already begin to observe how human error plays a significant role in our every day interactions. As humans, we are notoriously prone to making mistakes,not just when driving but during normal everyday tasks too.
Think of how many times you might have forgotten to turn your cooker off, after coming home tired following a long evening. Think of how many times you might have reached for your phone in the morning after a mad dash to work only to realize that it’s been forgotten back on your living room couch -although for some, leaving your mobile phone behind might be too unthinkable a mistake. An even better example; think of how many times you might have tried accessing a G-mail only to find yourself having entered your password once, twice and even the dreaded ‘thrice’ times in frustration, not knowing why for the life of you of you, you can’t seem to get it right?
Such is the case today with our natural tendencies towards making mistakes that when looking at national statistics you shouldn’t be surprised at all to learn how in the UK alone, 95% of road traffic accidents could be traced back to human error as being the single causal implicating factor. In 2016 over 180,000 accidents were reported 25,000 of which were known to result in serious injury or death. In the US the figures are known to be much higher and steadily growing with each passing year.
Many of these accidents have largely been down to behavioral factors such as drink driving, being distracted at the wheel, drowsiness or driving without seat belts. One of the main arguments in favor of the rapid adoption of self-driving cars and other forms of technology utilizing AI, lies in the fact that with their advent, the role of human error will be relegated to the backseat.
Unlike the case with humans, Self-driving cars will not, or more importantly cannot, get distracted, tired, clumsy or intoxicated whilst on the road. Nonetheless, they will still be prone to other errors and defects, only in this instance, these will tend to concern their manufacturing, testing, programming or design.
The Things Most Likely to Make Your Driverless Car Car Go Bump – Manufacturing and Design Defects
Manufacturing defects are issues that typically result from the improper construction of a product later leading to its malfunction. A manufacturing defect could be caused by things like the raw materials, components sourced or the manufacturing process of a self-driving car being defective.
On the other hand, Design defects are problems that can still occur despite the self-driving car being manufactured strictly to specification. Nonetheless a defective design, possibly at the pre-production or design stages makes the design of the entire product line unsafe. Normally, where the defect could have been discovered by careful testing, a liability situation will be created for those involved in its production and design. But whilst manufacturing defects would have once previously been prominent; given the present state of manufacturing standards, with high quality checks and controls augmented by machine processes at almost every stage, these are expected to be rare.
Whilst the risk of fatal road accidents occurring is expected to be low when fully driverless vehicles dominate the roads. During mixed fleet car scenario noted upon earlier, self-driving cars will still be vulnerable to accidents. Nonetheless, these will most likely be due to the error and the dangerous driving of other road users rather than the technology itself. So far many of the accidents involving driverless cars today have been down to the fault of their human users or other drivers, as opposed to technical or design defects.
Full autonomy has the potential to save thousands of lives within the UK and millions globally. But given the very real risk of human error, such as by vehicle operator inattentiveness during handover scenarios, or manufacturer oversight leading to unforeseen problems surpassing the testing stage, or more typical design faults; human error will still have a strong role to play in future road traffic accidents.
The Roads Ahead
Given future and present-day scenarios involving high vehicle usages and the ever present risk of human and machine error, sheer probability dictates that even fully driverless vehicles will too inevitably crash. More so than ever during the mixed fleet car scenario where these incidents might begin high but taper as more self-driving cars are eventually introduced to the roads. At first this process might involve traditional cars frequently colliding with self-driving cars, followed by fully driverless cars with less automated cars. And eventually in rare limited circumstances fully autonomous vehicles colliding with other fully autonomous vehicles.