When I approach one of a dozen or so junctions, I start to get tense and look out for the car that will fly in and cut me off. Which inevitably they do. I have often wondered why the same traffic infringements often happen in the same places and are allowed to continue, to the detriment of general safety. Which technologies are going to have the biggest impact on traffic safety issues such as this?
While video monitoring and AI interpretation of computer vision are coming on a pace, I wondered if there is a deeper and cumulative effect that will ensue from a combination of technologies and their adoption over time.
For example, just as has been observed with vaccination against disease, could there be a growing herd effect on safety from the adoption of autonomous vehicles and other technologies? This theory would hold that the more self driving cars there are, that remove human irrationality, the safer the roads will be.
Why? Lets look at something beyond a simplistic statement of more safe cars equal more safety overall.
Let’s first examine if these cars are safer.
The promise of autonomous vehicles is that they are good at tirelessly monitoring traffic constantly and even predicting when accidents might happen, like the case witnessed by a Tesla unit in the Netherlands:
If this case is to be taken at face value, then the car detected the potential for an accident a full second before the accident actually happened. While this is behavior that most humans carry out without thinking, it is not uniformly so. Tiredness or poor visibility can interfere with our ability to predict the traffic situation. Add to that the percentage of drivers that have a reduced capability to predict what is happening in front of them or drive in a casual way not paying full attention to the journey. We can see therefore that there is ample scope for the machine to improve safety. Finally, there are all the drivers breaking the rules by either going too close to cars in front or being on their phones or not paying attention for some other reason such as impaired capabilities brought on by alcohol consumption.
Another pointer is the US National Highway Traffic Safety Administration (NHTSA) investigation into the first fatality recorded in a partially autonomous vehicles. This vehicle was also a Tesla as they are probably the widest distributed semi-autonomous cars.
https://www.wired.com/2017/01/probing-teslas-deadly-crash-feds-say-yay-self-driving
The NHTSA not only exculpated the autonomous tech but also went on to laud the positive effects on safety this technology can have. In their research, the NHTSA discovered that Teslas driven with the semi-autonomous function crash 40% less often than those driven by humans alone.
We could now apply the concepts of herd immunity and herd effect, as defined by
T. Jacob JohnReuben Samuel in the field of vaccination.
http://link.springer.com/article/10.1023/A:1007626510002
There, herd immunity is the proportion of subjects in a population who are immunized. A beneficial “herd effect” applies to those not immunized coming from the immunisation of a portion of their number.
http://link.springer.com/article/10.1023/A:1007626510002
If we take this principle to the world of traffic accidents, taking as parameter the 40% potential reduction in accidents observed by the NHTSA, we could see this herd effect. So the cars that use autonomous tech get 40% safer and the cars that don’t live in an environment where a number of vehicles around them are 40% less likely to crash into them. We could dispute the %ge, but the principle indicates that more self driving tech could mean fewer accidents for all the herd, not just those lucky enough to own a robotic car.
But going back to the original issue of accident and dangerous driving black spots and preventing repeat offending, can we join the observation capabilities of autonomous vehicles with something else to unfailingly record where accidents happen in real time?
The rationale for doing so would be to enhance self driving tech further and inform law enforcement on where to deploy resources. The technologies exist to do so and could be deployed in combination with policies on usage of the data generated by them, especially around privacy. Technology such as 4G or LTE telco networks can be used to relay appropriately anonymized data to manufacturers and to competent authorities to analyze infringements or accidents in real time. As in the herd concept, the autonomous vehicle could report on its own issues and those of others on the road, even when the car itself is not involved.
So now all we are left with is ensuring that there can be no dispute in the time and place of a given recorded incident or infringement. Do do so we can turn to Distributed Ledgers Technologies such as blockchain powered systems. These enable all relevant parties to have an identical, unarguable, copy of the incident details. Parties such as vehicle manufacturers, law enforcement, regulatory authorities, insurance companies and health authorities could share the same data and be sure they have exactly the same information. This is not only good for the security and reliability of the data but it makes it resilient too.
This concept is akin to, indeed the rationale behind, neural nets or crowd sourced distributed systems.
So in conclusion, our roads could get progressively safer with the deployment of autonomous tech. With direct and indirect effects, such as the usage of vehicle data with appropriate policies.
Andrew Rippon is a leading the implementation of Distributed Ledger Technologies and True Digital processes for disparate fields such as Smart City government services, real estate operation, mobility, law enforcement and healthcare.