Sur le site du CCFA, reprenant un article des Echos ce jour, on peut lire que "Le Groupe PSA ne développera pas de fonctionnalités d’autonomie au-delà du niveau 3 pour les voitures particulières"

Le président du Groupe PSA Carlos Tavares a annoncé au Salon de l’Automobile de Genève que le groupe abandonnait les développements relatifs au véhicule autonome au-delà du niveau 3 pour les voitures particulières. Cela signifie qu’il n’est plus question de travailler sur une voiture individuelle capable de se passer de conducteur à certains moments et de se mettre en sécurité toute seule (niveau 4), et encore moins sur une voiture totalement autonome (niveau 5). Au niveau 3, en revanche, le conducteur peut déléguer la conduite dans certaines circonstances (autoroutes, embouteillages), mais il doit être capable de reprendre le contrôle à tout moment. « Entre le niveau 3 et le niveau 4, il y a une vraie divergence », a expliqué le dirigeant. « Compte tenu du coût additionnel de la technologie, le coût de la voiture devient tel que celui qui peut se la payer n’est de toutes les façons pas derrière le volant, mais plutôt sur la banquette arrière », a-t-il ajouté.

The Irish Times (non, je ne suis pas abonné à leur feed ...) parlait également il y a 2 semaines de l'annonce de Tavares : Self-driving cars: Is the autonomous dream slipping away from us?

The coming of the robot car has been screamed loud from every headline and tech blog for the past decade, but increasingly car makers are being more cautious about handing over control to the microchips.
Carlos Tavares, the head of PSA Group and its multitude of car-making brands – Peugeot, Citroen, DS, Opel, and Vauxhall – has said that fully autonomous cars may never make a mass-market breakthrough, because the technology will be too expensive for retail buyers. Fully autonomous tech, which would allow cars to entirely drive themselves, will remain the preserve of “those who could anyway afford to employ a driver” said Tavares. He said that PSA would continue to work on advanced driver aids, up to Level 3 (where the car can steer and brake itself on some roads but requires the driver to still pay attention and take control of the vehicle at regular intervals).

... ainsi que de la position de Ford :

Tavares’ sentiments seem to echo those of Ford, which apparently took a step back from committing to full vehicle autonomy.

According to Automotive News, Ford – having recently been a firm advocate of vehicle autonomy, and having gone so far as to build a Hollywood-style fake town on a Michigan test track to aid in the development of robot cars – seems to be rolling back on its original commitments.

Ford had championed the idea of leaping over Level 3 autonomy – where the car is mostly in control but which still requires human control and intervention – and going straight to Level 4, where the computer systems are broadly in control all of the time.

Now though, Ford has changed its corporate and technological mind, and it reckons that systems which ensure that the driver is actually paying attention and keeping an eye on things will be just as critical as those that actually control the car.

“Taking steps to get there as opposed to a big bang is more practical,” said Marcy Klevorn, Ford’s president of mobility.

“I think it allows us to provide autonomy in step functions to get people used to it. Acceptance is going to be a big deal; this is totally a different way to move. If you go right to full autonomy, the acceptance curve will be very steep, and you won’t have as much to learn about how humans interact with autonomy. If you just take that leap, you miss out on all those learnings.”

Ford still says that it plans to have Level 4 vehicles in use by 2021, but that in the meantime it will focus on “semi-autonomous systems”.

Un petit mot sympa également pour le projet d'Apple :

The American tech giant had been working on a much-publicised-and-or-rumoured autonomous car project codenamed, Titan.

It had been touted as a major disrupter into the automotive world, one that could seriously upset the apple-carts of existing major car makers.

Titan, though, seems to have proved rather more Titanic, and is sinking fast. Apple has confirmed that, far from pressing ahead, it’s planning to lay-off 190 people associated with the project, including product design engineers, and a machine shop supervisor.

While Apple hasn’t entirely abandoned the idea, it looks now more likely that it will “do an IBM” and instead of designing a car, will create plug-and-play autonomous and automated driving tech systems to be used by existing car companies.

On découvre avec un intérêt certain, à défaut d'un certain intérêt, qu'il existe une "Society of the Irish Motor Industry" (!) dans un passage sur l'éthique :

BMW’s special representative to the UK government, Ian Robertson, recently gave a speech to the Society of Motor Manufacturers and Traders (SMMT, the UK’s equivalent of Ireland’s Society of the Irish Motor Industry) in which he has questioned the very notion of fully autonomous driving. And it’s not the tech, that’s in question, it’s the morals.

“Imagine a scenario where the car has to decide between hitting one person or the other – to choose whether to cause this death or that death,” Robertson said in his address.

“What’s it going to do? Access the diary of one and ascertain they are terminally ill and so should be hit? I don’t think that situation will ever be allowed.”

Encore plus fort, on apprend que BMW envisage un niveau d'autonomie fractionnaire ! Déjà que la différence entre niveau 3 et niveau 4 n'était pas claire pour tout le monde ... (moi compris)

BMW has previously stated that it intends to have cars capable of hands-off driving available by 2021, and has even shown off a Rolls-Royce concept car that dispensed with a steering wheel and other driver controls entirely.

The Munich firm has spoken of its 2021 model being a Level 3.5 autonomous capable car.

L'explication n'est pas bien convaincante :

That would theoretically mean that it’s capable of driving without any driver input, but still needs a human operator to cover some more unusual scenarios. That is the very gap, the one between Level 3 and Level 4 autonomy, that many commentators have described as the most dangerous stage of autonomous vehicle development – the one where the machines are almost, but not quite, fully capable of taking over.

Ben non, c'est le niveau 3 qui pose problème à (presque) tout le monde [1]. La preuve avec ce que disait Ford il y a 2 ans : Ford to Skip Level 3 and Go Straight to Fully Autonomous (Self-driving) Cars

Je crois que je suis bien parti pour gagner mon pari.

Note

[1] sauf Musk qui s'en cogne, puisqu'il vend du L2 pour du L3 ... ou L4 ... enfin on ne sait plus