Caporegime
- Joined
- 30 Jul 2013
- Posts
- 29,417
The DLR in London is driver-less
Autonomous Vehicles
- Thread starter adam cool dude
- Start date
It has and is done already. Much more accurately than a person can land as well. So much so, at first it was stopped being used because the planes wore the run way landing area out in very specific spots where the wheels touched down. Its very rarely used now mainly in low visibilty landings, actually easier for the pilot to land the plane at other times.
Soldato
Talking of cows, I am a bit worried we will get an infestation of kangaroos in the UK, which the volvo system seems to be having an issue with. Apparently the unique movement fools the system into what distance the kanga is at and means it doesnt operate correctly.
So in Jiggers world the result will be total shutdown.
But you ask, how would humans deal with situation better?
Well we would just plough into them, 16000 collisions a year apparently.
http://www.bbc.co.uk/news/technology-40416606
The system monitors for deer, elk and caribou. I hope it could handle cows.
Interesting as I just remembered a week or so ago I came across a deer standing in the middle of the road, I only saw it when it was in range for dipped beam or so, as i had just passed a car coming the other way and flicked on main beam, so deer must have moved other car came past. In theory the monitoring system works a decent distance ahead so should have detected it earlier.
Certainly jumped me travelling around 50mph down a pitch dark country road
So what happened, I saw it, braked, slowed to about 10mph, beeped and it jumped off across the road into a field
What would driverless have done, I believe, detected object earlier, slowed earlier, stopped whilst hazard was there (it was literally in the middle of the road) Would the deer have moved, I hope so but in jiggers world I guess I would sill be there, the deer refusing to move and me dehydrated and dying from thirst and hunger.
What I really need is a pheasant detector and built in shotgun, bloody things shoot out of the verges and try to run (dont they know they are birds) across the road. One part of my commute journey is nicknamed bird strike alley now, its a wooded section that they love to meet up in.
So in Jiggers world the result will be total shutdown.
But you ask, how would humans deal with situation better?
Well we would just plough into them, 16000 collisions a year apparently.
http://www.bbc.co.uk/news/technology-40416606
The system monitors for deer, elk and caribou. I hope it could handle cows.
Interesting as I just remembered a week or so ago I came across a deer standing in the middle of the road, I only saw it when it was in range for dipped beam or so, as i had just passed a car coming the other way and flicked on main beam, so deer must have moved other car came past. In theory the monitoring system works a decent distance ahead so should have detected it earlier.
Certainly jumped me travelling around 50mph down a pitch dark country road
So what happened, I saw it, braked, slowed to about 10mph, beeped and it jumped off across the road into a field
What would driverless have done, I believe, detected object earlier, slowed earlier, stopped whilst hazard was there (it was literally in the middle of the road) Would the deer have moved, I hope so but in jiggers world I guess I would sill be there, the deer refusing to move and me dehydrated and dying from thirst and hunger.
What I really need is a pheasant detector and built in shotgun, bloody things shoot out of the verges and try to run (dont they know they are birds) across the road. One part of my commute journey is nicknamed bird strike alley now, its a wooded section that they love to meet up in.
So the technology is there to identify different textures in the road while traveling at 60mph?
Soldato
I don't believe it is yet, but I am sure its going to come.
However most incidents are not due to road conditions directly, but drivers, doing one of many things:
- Not driving to the conditions
- Speeding
- Not paying full attention
- Paying full attention, but needing more brain power than is available
Lets face it, there are plenty of times when its nigh on impossible to see the road well enough to detect a surface change, yes in daylight on a decent road fine
But at night, or in heavy rain nope
There will be situations when a human may exceed the ability of automated systems, but there will be more situations when automated systems will exceed the ability of a human (even if not distracted)
ABS is example 1
So we could take all the self driving systems and put them in a car, without the car actually doing anything other than flagging them as warnings. So going back to ABS, how helpful would it be when braking if it kept informing you of locked wheels? Not very, so its automated and it does a better job. So maybe that could be extended to normal braking, it is now, emergency city braking etc (names differ by manufacturer)
What about keeping a safe distance, most drivers fail this.
Auto parking
Parking sensors, we all know people who said they were for women who couldn't park. However I have no shame in admitting I would park in spots in my 5 series I wouldn't have attempted without, why? because its a longish car and pretty good but not awesome visibility, the safety the sensors added mean't I could park in spots that I probably couldn't have before.
Lane assist
So many other techs that "drivers" are relying on, we are getting closer to being operators than drivers.
Probably the most contentious one would be speeding. Some cars now have the ability to detect road signs, so could by default limit the speed to the correct limit for that area. Would this reduce incidents? Of course, and it would certainly reduce the impact of incidents if they were unavoidable such as someone pulling out of a sideroad just as you were going to pass
That's nowhere near as safe as me and about the same record as my wife. My Dad would probably be 10 times safer again.
Problem is they have only really been tested in either controlled environments or they have worked up most of their average statistics in the US on long, straight, big roads in a grid layout. The figures are a bit fudged.
We have yet to see them deal with rural Britain, bad weather or how they would react when e.g. meeting another car on a single lane road, or even a large puddle/small flood.
And yea, their stats are still worse than my personal driving history. Which means a self driving car would be more dangerous than me...
We have yet to see them deal with rural Britain, bad weather or how they would react when e.g. meeting another car on a single lane road, or even a large puddle/small flood.
And yea, their stats are still worse than my personal driving history. Which means a self driving car would be more dangerous than me...
Last edited:
I'm sure someone is racing to bring the AI cow to the market. AI cow 10 times safer than regular cow and zero emissions. The problem is you need a guy in a control room to take over when they fall down.
Defiantly. A lot of our roads are based on tarmac'd ancient footpaths from long before we invented North America.
Soldato
I guess if there ever was an admission of defeat thats it
Yeah driverless cars will be a no go until the AI cow is released.
TBH Buddy post #227 is a pretty big admission of defeat. If you need to relay on those systems then AI could be the best option for you. I wouldn't sound your horn at animals on the road either. You could cause an accident or breach of the peace. Maybe even spook a driverless car into doing something silly.
Last edited:
Soldato
I think your swimming against the tide on this one
They are coming, it just depends on how fast and how fully they integrate
I can see for example some main road integration happening very quickly, small country lanes less quickly
But the biggest benefits would be on main roads as they are the bulk of the miles travelled
Will we require infrastructure changes to support them? Who knows, we may need cow sensors strung along the roads.
What i am certain is we will see a lot of the tech come to manual cars, it all functions as driver aids at the end of the day so once developed why not
Like I said for example, sign recognition, automatically be able to set "apply speed limits" switch to on, no worries about begin caught speeding when driving in unfamiliar areas and missing a 40 to 30 change or something.
I am sure some people thought satnav would never beat a good map
They are coming, it just depends on how fast and how fully they integrate
I can see for example some main road integration happening very quickly, small country lanes less quickly
But the biggest benefits would be on main roads as they are the bulk of the miles travelled
Will we require infrastructure changes to support them? Who knows, we may need cow sensors strung along the roads.
What i am certain is we will see a lot of the tech come to manual cars, it all functions as driver aids at the end of the day so once developed why not
Like I said for example, sign recognition, automatically be able to set "apply speed limits" switch to on, no worries about begin caught speeding when driving in unfamiliar areas and missing a 40 to 30 change or something.
I am sure some people thought satnav would never beat a good map
Cars will increasingly follow the Tesla model; install all of the tech necessary to enable autonomous driving. Sell the software as an optional extra. Use the hardware to gather data on how people drive.
Tesla will already have a somewhat detailed map of UK roads and driving behaviours. They have ~5000 cars driving around this country, gathering data. Yes, autopilot has its flaws. But it's coming along nicely. In the early days corners on unmarked roads without kerbs could be somewhat hairy. Now the discussion is about how to tackle obstacles in the road, rather than how to navigate the road itself.
Level 5 automation is still some way off. But more autonomous tech in production cars means more data is gathered. That, in turn, leads to faster development.
Tesla will already have a somewhat detailed map of UK roads and driving behaviours. They have ~5000 cars driving around this country, gathering data. Yes, autopilot has its flaws. But it's coming along nicely. In the early days corners on unmarked roads without kerbs could be somewhat hairy. Now the discussion is about how to tackle obstacles in the road, rather than how to navigate the road itself.
Level 5 automation is still some way off. But more autonomous tech in production cars means more data is gathered. That, in turn, leads to faster development.
It's unlikely that there's an established solution yet, so it would hand back control to the driver. Perfectly acceptable for Level 3 and Level 4 automation.
I guess it could also record data on the car blocking the road and send that off to the local authority, as the car would be causing an unnecessary obstruction. That could, in turn, lead to better enforcement of traffic laws, mitigating the issue.
Last edited:
I think the example you've given is one where a) people shouldn't be parking on pavements anyway so an AV would simply park somewhere else or b) falls into the category of 'things we'll need to change in our cities to accommodate autonomous vehicles'.
Soldato
But parking on the pavement is illegal, and more and more places are starting to clamp down on it. Yes its a problem as lots of housing predates everyone having 2 cars.
Imagine, car "cannot park no space available"
Driver "override parking restrictions"
car "parking restrictions overridden, input parameters"
Driver "parallel park"
Car slots in in line with previously parked cars bumped up onto the pavement
Soldato
Its an option if they choose to enforce the law
They would have to follow the law them selves then as even the police and ambulances park on the pavement.
Edit: Just read that its only illegal in london else where only if signs say so.
Whats all this level 4 and level 5 stuff, what are the levels?