Amazulu

He sold one to the Socialist Republic of South Australia. They've replaced their coal-fired power stations with it - and as they contribute zero to the Australian economy (in fact, less than zero as the rest of the country carries them), it seems to be doing the job

GarryP

For those still saying the autonomous vehicles and robotaxis are a decades off.

https://www.news.com.au/technology/i...fa6bf870f3ecdb

Cat4 testing starting next year in israel. In addition to all the other test services ongoing in different parts of the world. 3 years hence for serious roll out seems if anything an overestimate.

Wol

For those still saying the autonomous vehicles and robotaxis are a decades off.


>>A car equipped with level four self-driving technology would be able to complete all the dynamic aspects of driving but only under specific road conditions such as a restricted area or road type.For instance a level four vehicle would be able to drive safely on highways and well trafficked roads but side streets and more urban areas would require human input.Outside of the controlled areas the car will request for driver input and will pull the car over if the driver fails to comply.<<

But the caveats are just the point!

No-one who has paid any attention to this over the past 40 years or so doubts that autonomous cars work in many environments - after all there are plenty of videos of them doing exactly that.

The problem is twofold: firstly, when will they be licenced to ply the roads without a "driver" or steering wheel - after all, if they are as safe and efficient as promised they should be able to, and secondly will they be able to run anywhere? If not, a whole different can of worms is opened. And it's a big can, too!

Shard

1. Waymo will be testing without a driver in Mountain View in March, I think. 2. Why keep raising this point, it's not a full and instantaneous deployment. The tech will continue evolving and in year's hence the AV will be able to navigate more remote or challenging terrain. And even with successful AV, there will likely remain plenty of human commercial and recreational drivers. It's not really an all or nothing proposition.

GarryP

I think they kind of missed the point. The local streets are if anything easier than the busy main ones, less traffic, less speed, less scope to go badly wrong. The limits of deployment initially will be those streets that have been surveyed and not having to be dealing with rural tracts. The urban and suburban streets will be covered - hell Elon is talking about deployment of software to enable their cars to navigate a car park, find and park, and do so - by next year.

Wol

>> ....Elon is talking about deployment of software to enable their cars to navigate a car park, find and park, and do so - by next year. <<

I am actually a big fan of Musk - he's not perfect by a long way but there aren't many who could set up what he has in less than two decades. But he does go down the hubris route a little. In any case I'm not sure that Tesla is actually at the leading edge of automation. When I have persuaded Mrs Wol to let me get a model S I will report further.

I came across this today:

https://www.researchgate.net/publica...on-road_study?

Seems like reinventing the wheel - in aviation there have been far too many hull losses caused in essence by human/automatics dissonance, lack of mode understanding etc - and this among highly trained and constantly checked crews, not your average Joe with a hangover.

BadgeIsBack

Sure the tech works, but commonsense tells you regulation does need to solve a lot of issues and make sense of inevitable problems first - that is what will take time

Sure, aeroplanes operate in 3d successfully but there are a lot less of them. You can't control airspace in the burbs.

BadgeIsBack

Spot on as always.

GarryP

Which is why they are aiming to jump to cat 4 - so there is no dissonance, etc.

BTW many of the problems in aircraft are down to the electronics (which aren't that smart) keeping the plane flying normally, right up to the point they can't manage any more and dump a barely flying mess onto the human pilot. There's quite a lot to be said for smarter autopilots and no dumping it back on human pilots at all. A lot of the time it's a passing of the buck "we are going to crash, so you can't make it any worse".

Wol

Actually, autopilots are a lot smarter than your post implies - and that's quite a lot of the problem! They have so many modes and combinations of modes that when things start to get complicated their activity can be misunderstood or not understood at all - which has been the cause of quite a few hull losses over the past thirty years or so especially with Airbus.

Even with relatively simple <g> operations like autoland, which on the surface involve little more input from the crew than setting a few buttons, training for the crew is extremely complex and takes up quite a proportion of the six monthly simulator checks. This is because the failure cases are many and varied, the responses also depend on the height or altitude the failure combinations occur, whether they happen before or after an engine failure etc etc.

I just can't see autonomous cars becoming common on public roads until they are fully driverless and are registered as such. The failure cases and maintenance requirements of the cars and nav systems are just not in the same league as in aviation. When you look at the host of sensors, radar, lidar etc in use it's not to me feasible.

Shard

When would that be ?

Robotaxis to roll out in 2021. Shall we say AV becomes 'common' by 2030?

GarryP

They are actually quite simple, and the fact they are modal is the big issue. They are at least an order of magnitude (probably two) simpler than an autonomous car, and very deterministic. Of course, the reason is, they only really have to deal with the airframe and a bit of environment - not other aircraft, signposts, cyclists, etc. And that airframe is stable.

Mainly to stop the pilots screwing things up by touching anything.

Cat 4 is driverless, and that is what they are looking to deliver pre-2021. Hell, Waymo vehicles are driving around NOW, on real roads, no driver and test public passengers.

Shard

Another trial. Garry's 2021 rollout is looking more and more realistic !

https://www.google.co.uk/amp/s/amp.u...amp/1997500002

Beoz

That article talks about a human driven car understanding how the market may want to use driverless.

Despite the hype from Elon and Garry, it sounds like driverless is more likely 2030. Even if they can get the merging thing sorted (which is a big item) there are so many little bugs that need resolving.

https://www.news.com.au/technology/innovation/motoring/hitech/apple-cofounder-steve-wozniak-says-dont-expect-a-selfdriving-car-anytime-soon/news-story/14f087b4710a3932139351923da1902f

Shard

2030 for the flying AV maybe. Robotaxis and specialised delivery tech seems pretty much ready. Trialling on live streets means rollout is within sight.