Where have you got that crap from ? When we have a significant number of driverless cars running then there'll be a proper comparision.
But with the example shown I don't believe even you would have had the accident the driverless car had, and the other tesla with the lorry killed it's pasenger. But that wasn't a driverless car either as it had manual controls and was NOT meant to be always driven driverless anyway.
I hate it when I can't see what's in front of the vehicle in front. In older days it was much easier to see through the vehicle when no headrests and additional bodywork existed. Also, and I guess as most decent drivers would, I'd drive slightly offset so I could see what is going on.
The Tesla does exactly what human idiots do in fog.
Which AI can never have.
This is where cognative inteligence is needed, most things can be predicted and this is where fuzzy logic came in.
Some are, others prefer to point out the problems rather than ignore them and hope they go away.
That has only just come into effect as previously learners weren't allowed to drive on the motorway making it pretty difficult to learn on them.
what about night driving I was suprised that it isn't tested.
There's more cars than simulators and cars are cheaper too.
Some assume the word inteligence means a good driving practice.
Well I don't .
"Musk reiterated his dislike of LIDAR and defended Tesla?s strategy of achieving ?full autonomy? using only cameras, radar, and ultrasonic sensors."
Assuming the main system at longish range is radar, I imagine a cardboard car would not reflect anything much at all, so it would be invisible to the sensors. Only when the range shortened would the tesla 'see' the cardboard car with its other systems.
Alternatively, the Tesla system has to be tuned to ignore some rather whispy objects, such as fog or drops of rain. So, they may have some simple and robust algorithm that says "if it shows up on the camera system but not the radar, then it's just fog, so ignore it".
I think it did eventually. Perhaps once it's in sonar range. It did, after all, stop before demolishing the second cardboard model.
I think this 'accident' was rather carefully contrived.
Personally, I don't drive that closely behind the car in front, precisely because my concentration on long motorway journeys is not
100%. I'm forever getting people pulling into what I regard as my braking distance and they regard as a large gap. :/
^ That!!!
Avpx
But they already demonstrated that the Tesla would stop for a real car! They needed something it would crash into. ;)
Well why not use a student ;-)
They knew it would fail and wanted a more dramatic vid than just having a human do an emergency stop.
This automation is something the airline business has been working on for years. Flying an aeroplane is one thing. Monitoring an autopilot flying an aeroplane is a completely different task (and in a lot of cases more difficult because when things start to go wrong it can take longer to get "in the loop").
Bearing in mind how easy it is to fall asleep on motorway journeys now I think it entirely unreasonable for a car manufacturer to say "the car will drive but you still have to monitor it and be ready to jump on the brakes/steering when it doesn't do its job."
If the car will do the lot and the human can read a book then that's fine but expecting the human to get the car out of trouble will never be a satisfactory solution.
lot-what-happens-next
ultrasonic
He said it's too bulky and expensive, but he's tesla isn't exactly cheap.
I find it strange it didn't see it via camera deteaction as that's what Et on Musk seems to prefer, well if it doesn't work then..... he won't have hi s full autonomy car. Can this bug that has been shown be fixed in software is the key, otherwis e he's going to have to come up with a new sensor array.
I wonder if there's a bouncy castle in a carpark a tesla would drive into t he castle because it couldn't see it using radar ?
No you can;t igmore these, you have to account for them NOT ignore then as it will rain there will be fog and things are made out of card, paper and o ther things that radar might not see as solid or worth breaking for.
Again some cognative awareness is needed, the awarness of what is around u s and what might appear around us at any particualar time.
r away to be seen, but the humans saw it, why didn't the tesla ?
why did it wait ? why didnlt it swereve to avoid it. Surely it worked out the speed of approach that is pretty basic stuff for s ensors, even our basic ones the SFR05s less than a quid on ebay, but they c an't do more than about 4 metres, but surely a visual method could be used.
I was it cardboard ?
who only observes the car infront when driving ?
Yees but it proved a point, which was the difernce between a driver assiste d car and a driverless autonomous car, and how humans are still better than computers at driving cars.
Was it that close ? Most drivers I see have less than a car lenght bewteen them and the next car. and with a driverless car this problem shouldn't occur anyway.
I'm waiting to see driverless car racing formual E type. When that happens we'll have arrived, although still won't be sure where we ;re going.
The point is it's less likely for an automated car to crash than a human controlled one. By all means improve the automated one further, but it's already safer than a human, so saying they're "dangerous" and shouldn't be allowed is stupid. They may have a few weak points, but humans have way way more.
A study.
Now go look at every single human caused accident and realise that a robot would not have made any of those mistakes.
Exactly so.
The aeroplane cases we hear about are the ones where the human pilots failed to pick up the pieces. There may be plenty where they didn't fail.
Nevertheless, we saw with the recent uber car accident that the human fall-back driver was literally half asleep. It's impossible to maintain alertness when the machine does 99% of the driving.
Handing the driving over to the human just in the most difficult circumstances means that the human gets out of practice and can't handle those circumstances.
They can read signs.
It wasn't a normal object. Stuff falling off lorries would be full boxes. I remember as a teenager we used to put empty boxes and carrier bags in the road and laugh at drivers who avoided them. Every so often a sensible one would just drive over it.
Because that wouldn't have fooled the car and made their invalid point.
Are they annoying you again?
I like the Volvo test where it ran the developer over.
If it had been a real car, it would have worked. The whole thing is bullshit.
yep
because there's no way of knowing that the box is empty
tim
because a cardboard box *is* a real object
you've no way of knowing what's inside it
tim
No it is NOT, that is a fact stat. automated cars havn;t be allowed on busy roads long enough to properly test them, they can drive very well on opwn uncongested roads but they havent; been tested in teh real world for very long thtre's only a few location in the world where they allow automated cars without a driver in them.
No they aren't. It's a distorted statistic.
I;ve not said that, they need proper testing on proper roads. That's why there are restrictions on where they are allowed to drive.
Not true, automated cars haven't been tested enough as yet, there's too sma ll a sample and NO car is allowed on a pubilc road without a driver.
HomeOwnersHub website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.