Autonomous vehicle roadmap: 2015-2030

Anything goes in here.....
User avatar
Kelvin
Posts: 656
Joined: Thu Apr 20, 2006 12:28 am

Re: Autonomous vehicle roadmap: 2015-2030

Post by Kelvin » Tue Feb 16, 2016 4:02 pm

jason wrote:For road transport, I'm not certain we'll ever see truly autonomous vehicles - at least not for many generations. But this depends on the interpretation of autonomous.

If we consider transport operating on a network system (so we ignore lifts, etc!) what present, or near future, form of human transportation is truly autonomous, ie. does not, at minimum, at least involve live monitoring/management by remote operatives?

…so it's hard to imagine a time when each vehicle would not require supervision/monitoring, and surely the volume of vehicles and routing options mean this can only be carried out by a responsible (and sober!) occupant. I'm sure remote monitoring will come, one day, of course.

The same argument exists with lorry drivers, they need to be responsible and sober (and not lie about having blackouts!) and this is harder to control if they are at the wheel of a lorry than, say, sitting in a command room monitoring multiple lorries.

The issue is more the ethics and way our laws are written. While autonomy will reduce accidents it won't eliminate them and in certain circumstances autonomous lorries will need to take avoiding action that could result in people being hurt or killed. The onboard computer will need to evaluate the situation and decide the course of action with the smallest impact. e.g. I'll drive into this car rather than that bus. This means that the programmers will have had to write the algorithms to accommodate this. If the lorry makers get sued up the ying yang if this happens then fully autonomous vehicles will never be a reality. The most likely first step with freight lorries on the public road will be driverless lorries with a driver in the cab to take over in trickier situations. Autonomous vehicles are already being used in large quarries/mines (Rio Tinto) and by the military.

fd
Posts: 883
Joined: Mon Apr 23, 2007 4:56 pm

Re: Autonomous vehicle roadmap: 2015-2030

Post by fd » Tue Feb 16, 2016 8:48 pm

I do safety critical systems work, software and systems design. Stuff that will kill people if it fails. Our goal is less than 1 fatality per 10e8 to 10e9 hours of operation, pretty much at the sharp end. As I understand it this is about the same safety level as aviation ( it's difficult to compare as the standards don't measure the same things). Do the maths .. You can calculate how many people autonomous vehicles will kill per year if you know how many vehicle hours are used in the UK if they are as safe as aviation systems. I think autonomous vehicles are a long way off, on cost grounds, who will maintain them ? You can't have joe blogs freckling around with that kind of kit. Of course the bean counters may choose to make then less safe and kill more people, or they will be unreliable - reliable AND safe systems are difficult .. Lastly, most high safety criticality systems live in very controlled environments especially where they can kill people . . cars on a public road are not in a controlled environment . . that makes the safety case much harder .. As with most safety things it will come down to the value of life . .

fd
Posts: 883
Joined: Mon Apr 23, 2007 4:56 pm

Re: Autonomous vehicle roadmap: 2015-2030

Post by fd » Wed Feb 17, 2016 12:56 am

310 billion vehicle miles travelled in the UK in 2014.
Say at an average speed of 30mph => 630 billion vehicle hours.
At the least stringent safety we have one fatality per 100 million vehicle hours.
So that is 630 billion / 100 million = 6300 fatalities per year.
There were 1700 people killed on the roads in 2014.

So, aviation grade safety systems, if they were even possible in an uncontrolled road driving environment, would kill ~3x the number of people that currently die - a closer comparison than I expected.

I still wouldn't like to be on the marketing campaign for that.

This is just back of the envelope maths as it's been a long day, but there you have it . . . your cars will be safe from this threat for a long time . . . I recon you need 2 orders of magnitude reduction in fatalities before it's a goer . . .

User avatar
thinfourth
Posts: 3177
Joined: Tue Mar 15, 2005 12:06 pm
Location: Playing in the mud near aberdeen

Re: Autonomous vehicle roadmap: 2015-2030

Post by thinfourth » Wed Feb 17, 2016 7:43 am

fd wrote:310 billion vehicle miles travelled in the UK in 2014.
Say at an average speed of 30mph => 630 billion vehicle hours.
At the least stringent safety we have one fatality per 100 million vehicle hours.
So that is 630 billion / 100 million = 6300 fatalities per year.
There were 1700 people killed on the roads in 2014.

So, aviation grade safety systems, if they were even possible in an uncontrolled road driving environment, would kill ~3x the number of people that currently die - a closer comparison than I expected.

I still wouldn't like to be on the marketing campaign for that.

This is just back of the envelope maths as it's been a long day, but there you have it . . . your cars will be safe from this threat for a long time . . . I recon you need 2 orders of magnitude reduction in fatalities before it's a goer . . .
Ah but

I am not going to argue over your figures but i think you might be comparing apples with oranges

i don't know about anyone else but i have never had a phone call saying i am owned £3000 in compensation for the aircrash i was in. Air crashes are very rare.

While cars have a surprisingly low fatality rate i don't think this is due to humans being awesome drivers. It is due to modern cars being very good at crashing without harming the squishy bit in the drivers seat. Humans drivers are remarkably bad at not driving into things.

I think autonomous driving could seriously reduce the accident rate while leaving the fatality rate not hugely effected. To kill yourself in a modern car you have to of done something quite dumb. As long as you have that manual option quite dumb will still be possible.


I cannot see the manual option ever disappearing as how many cars do you know where you can't turn off the traction control for the driving gods. the driving gods must be able to turn off traction control for trackdays

I had a leaf on a test drive even in that you could turn traction control off

Why?

It was a rolly polly electric hatchback that will never see a trackday
Landrover 90 = Muddy shed spec
Fiat panda = Couldn't care less spec
Landrover ?? = Muddy shrek spec
Unimog 404S = Very slow silly offroader spec
Kubota F1900 = Snowplough spec

fd
Posts: 883
Joined: Mon Apr 23, 2007 4:56 pm

Re: Autonomous vehicle roadmap: 2015-2030

Post by fd » Wed Feb 17, 2016 1:24 pm

You haven't been in a place crash have you ?

So you're not due a phone call or any compensation . . . or am I missing something ?

AV's may or may not reduce the accident rate, but statistically are likley to (per the figures) KILL 3x the number of people.

The statistics and figures remove how awesome (or not) drivers are, don't care about track days or traction control, egos or bullsh1t . . . that is why they are so much better at giving guidance on what is likely to happen.

Anyway - good luck with your marketing campaign . . .

Fd

User avatar
Kelvin
Posts: 656
Joined: Thu Apr 20, 2006 12:28 am

Re: Autonomous vehicle roadmap: 2015-2030

Post by Kelvin » Wed Feb 17, 2016 1:43 pm

How many of us would be prepared to get on an aircraft that didn't have people in the pointy end telling the stewardess to put the bomb bay doors to manual? I just don't think I could bring myself to get on a pilot less aircraft even if the most common cause of crash is pilot error.

User avatar
jason
Posts: 2183
Joined: Mon Oct 09, 2006 2:15 pm
Location: East Lothian

Re: Autonomous vehicle roadmap: 2015-2030

Post by jason » Sat Mar 05, 2016 11:44 pm

Announcement expected about driverless Lorre convoys to be trailed on the M6 in Cumbria. One driver in the leading lorry. I think this is more a remotely operated slave setup rather than autonomy.

http://www.bbc.co.uk/news/uk-politics-35737104

User avatar
Dominic
Posts: 14447
Joined: Sat Mar 19, 2005 10:14 am
Location: Milton Of Campsie
Contact:

Re: Autonomous vehicle roadmap: 2015-2030

Post by Dominic » Mon Mar 07, 2016 3:55 pm

http://www.dsaccountancy.com

1999 Lotus Elise Sport 135'99

User avatar
j2 lot
Posts: 7660
Joined: Tue Dec 19, 2006 8:47 pm
Location: Strathaven / Glasgow

Re: Autonomous vehicle roadmap: 2015-2030

Post by j2 lot » Mon Mar 07, 2016 8:37 pm

' buses and large vehicles are less likely to yield' - ain't that the truth :blackeye
2015 Lotus Evora
2022 Polestar 2 LRSM Plus
2023 Skoda Kodiaq Sportline

User avatar
Dominic
Posts: 14447
Joined: Sat Mar 19, 2005 10:14 am
Location: Milton Of Campsie
Contact:

Re: Autonomous vehicle roadmap: 2015-2030

Post by Dominic » Thu Mar 10, 2016 4:22 pm

Pistonheads on the new Merc E class. The future is coming soon!.... ish
Pistonheads wrote:'Just as significant is the new Drive Pilot, an evolution of the Distronic Plus seen on the previous model. What's new this time around? It can now follow the car in front autonomously at up to 210km/h, or 130mph... At speeds of up to 130km/h (81mph), it can "continue to actively help with the steering by taking account of surrounding vehicles or parallel structures, even if the lane markings are unclear". And it can change vehicle speed to the limits detected by the camera. How does it work in the real world? Activation couldn't be simpler - just pull the cruise control stalk towards you twice - and a display in the dash tell the driver their Drive Pilot is on. It will then maintain a speed, brake when required, steer through bends and even change lanes with the indicator on for two seconds, though that latter tech won't be in the UK. It's initially pretty spooky to have steering and throttle inputs done for you, but as it continues to work so confidence builds.

Trouble is, it's not flawless. In certain instances the E-Class just didn't feel like it would respond to the situation, or it would overreact unnecessarily when it was safe and set the seatbelts to strangle mode. For the vast majority of the time it worked well, but technology like Drive Pilot must be flawless for us to trust it. On this impression it really isn't far off though.
http://www.pistonheads.com/road-tests/m ... view/33865
http://www.dsaccountancy.com

1999 Lotus Elise Sport 135'99

Post Reply