Is it necessary to wait for the perfection of autonomous cars before authorising them?

One technological advance which may most affect our daily life in the near or not-so-near future is that of autonomous vehicles, those which circulate without anyone steering or controlling them. At this moment, apart from technical issues, public debate about these vehicles involves safety, a key factor related to the authorisation by public administration to be able to circulate.

Nidhi Kalra and David G. Groves, Rand Corporation researchers have published research that analyses 500 different scenarios assessing the introduction, adoption and improvement of autonomous vehicles, in order to respond to the following questions:

  • How safe do these vehicles have to be before they can be authorised for public use?
  • Under what conditions can lives be saved by each of the policies in the short and long term and how many are saved?
  • What do the tests suggest about conditions that generate minor costs while waiting for significant improvements prior to their implementation?
  • What does this imply for policies that direct the introduction of autonomous vehicles to their use by consumers?

The model that they have designed to carry out the analysis compares the accident rate and deaths that would be caused if autonomous vehicles were authorised to circulate when it has been confirmed that these improve by an average of 10% the accident rate connected to human-driven vehicles or if it is expected that this improvement is between 75 and 90%.

The conclusion they come to is that it would be necessary to authorise the circulation of autonomous vehicles once the average of safe driving by humans has been improved by 10%. The main reason would be to save human lives (hundreds or thousands in the short term and thousands in the long term) which would transpire for two reasons. The first reason would be that once autonomous vehicles, on average, acquire a safety level higher than that of human driving its authorisation would reduce accident rates. Even if this reduction were small, the time that has passed between the level of acceptable safety and the ideal level of safety, would accumulate a number of human lives saved. The second reason is that the authorisation of these vehicles would encourage their use, and considerably increase the data available for analysis. This way, the evolution of autonomous driving would speed up and levels of safety of over 75 or 90% better than those involving human driving would be reached before.

Source: Kalra, Nidhi; Groves, David G. The enemy of Good. Estimating the Cost of Waiting for Nearly perfect Automated Vehicles. Rand Corporation

https://www.rand.org/pubs/research_reports/RR2150.html

_____

Aquest apunt en català / Esta entrada en español / Post en français

Deixa un comentari