Consumer advocate Ralph Nader, whose book, Unsafe at any speed, forced carmakers into fitting seat belts as standard, has his eyes on Tesla.
In a post regarding the car company’s autonomous driving technology, Nader said: “Tesla’s major deployment of so-called Full Self-Driving (FSD) technology is one of the most dangerous and irresponsible actions by a car company in decades. Tesla should never have put this technology in its vehicles. I am calling on federal regulators to act immediately to prevent the growing deaths and injuries from Tesla manslaughtering crashes with this technology.”
The argument that autonomous cars are dangerous is only half the story. Just like seat belts and airbags – if every car is fitted with the same level of autonomous technology – the risk of something going wrong isdrastically reduced. In other words, autonomous cars are safest when humans are taken out of the equation.
There is definitely a need for cross industry collaboration, to ensure that all the use cases tested are shared among manufacturers so that the software systems and AI models used for autonomous driving have the widest possible sample of learning experience combined with real world data of how all autonomous cars behave in real world conditions on public roads.
Let’s go a step further and pull in all the CCTV and speed camera data that exists to understand the key part of the AI testing that has no data inputs – other road users. How does the AI know how a pedestrian – or for that matter a cat, a dog or, if you are unlucky enough to encounter one on the road, a moose – will respond to an autonomous car? No amount of data about drivers can possibly build an accurate picture of the behaviour of other road users. So with regards to road safety, Nader does have a point.
Looking in from the outside, it does seem the US has a tendency to prioritise motorised vehicle users over other modes of transport. One can’t help but worry if one of the leading car companies developing autonomous technology is a US firm that is testing its algorithms on the US road network.
The more data collected, the better the data model. Michael Taylor, IT director of the Mercedes-AMG Petronas Formula One Team, says that data has always been part of Formula One. And while the only real piece of data that matters is the lap time – which in the early days was measured on a stopwatch, the data models used by F1 teams means they are getting very close to predicting events on track by analysing the build up in real time. Formula One is a pretty controlled environment, public roads are not. Carmakers will need a heck of a lot more data before autonomous cars can be truly trusted.