Autonomous Vehicles

Caporegime
Joined
28 Feb 2004
Posts
74,822
Very interesting personal contribution to AV thread. I am curious if you work for a major company or a small company involved in AVs---car manufacturer or tech company?

I am particularly interested in learning why you have not tested your AVs on road in those States that permit it---eg Nevada.

This link in PC Mag caught my eye. Do you share this view about a "God View" eye in the sky? Particularly for the edge cases where the vehicle is unsure what to do. How do you see teleoperation implemented?

http://uk.pcmag.com/news/92963/why-self-driving-cars-will-require-a-god-view-eye-in-the-sky



I work for a manufacturer independent, test, research and development company, we work on all and any type of vehicle, not just AV.

Manufacturers come to us because of our skills in testing any and all types of vehicles and because we have the tracks and facilities to put any and all types of vehicles through their full paces.

We do not test the cars in Nevada or other US states as we are based in the UK.

We have done some work with drones coupled to individual vehicles, and with high altitude balloons with cameras and other sensors linked to a main hub that any AV can access, amongst other options.
 
Permabanned
Joined
17 Aug 2016
Posts
1,517
I work for a manufacturer independent, test, research and development company, we work on all and any type of vehicle, not just AV.

Manufacturers come to us because of our skills in testing any and all types of vehicles and because we have the tracks and facilities to put any and all types of vehicles through their full paces.

We do not test the cars in Nevada or other US states as we are based in the UK.

We have done some work with drones coupled to individual vehicles, and with high altitude balloons with cameras and other sensors linked to a main hub that any AV can access, amongst other options.

Nice to hear that you are making good progress with your track testing. On your AV testing, do you test them on public roads in the UK too? If so how do you think your testing compares and contrasts with Waymo's.

I see that they are bringing their Pacifica SUVs to Atlanta, Georgia, an interesting city. This link and comments mentions that Atlanta is an "incredibly car-centric" city, with only 16% of households without vehicles. One of the comments in the link says: "if Waymo can get around the crazy Atlanta drivers, pot holes, and exploding bridges and water main breaks I’ll be impressed. Driving in ATL is always an adventure."

https://www.theverge.com/2018/1/22/16919716/waymo-self-driving-minivan-atlanta
 
Caporegime
Joined
25 Jul 2005
Posts
28,851
Location
Canada
There are surely other means if your goal is safer roads.

For one - legally mandated vehicle performance tracking. Your vehicle records its location, speed, acceleration and various other parameters like an aeroplane's black box. This data is automatically periodically sent for analysis - like when you recharge.

Yes you would still face the prospect of criminals et al removing/hacking/disabling the devices. But you could also reduce the need for police manning mobile speed cameras. Maybe couple it with monitoring technology embedded in the roads themselves, street lights, etc. Every street light could be turned into an average speed monitoring device, communicating with something embedded in each vehicle. Etc, etc.

For me I see this as a more tolerable form of technology impacting my driving experience (mostly because it won't). I can still hop in the car and go where I please, how I please. So long as I drive legally and safely I have nothing to worry about from the systems monitoring my performance. And perhaps idiots will get caught out/grassed up by their own cars...

For me, personally, the interesting thing about AVs isn't their safety. It's the idea that we might not need to own a car but could summon one. I'd just personally like to be able to summon one and then drive it manually. That would be pretty cool.

The goal isn't just safer roads. That's almost more of a byproduct than anything else. For me, the interest in AV is so I can do something else rather than have to drive.

That 5am two hour journey I need to regularly take? I can snooze until my destination... Driving long distances, I can sit reading a book or browsing the internet or play a game rather than stare in front for hours on end... Being able to stick the "kids" into the car and get it to drop them off at school or their club... The aim being having what is essentially a personal chauffeur without the cost.

That doesn't mean I don't want to drive at all, but there are a lot of journeys and situations I would prefer not to have to. Perhaps I'm feeling tired after a hard days skiing, I can snooze while it drives me back as another example.

It also benefits those that can't drive at the moment, the disabled, the elderly, those that never got their driving licence/nervous, children etc etc.

Realistically it's unlikely that "manual" driving will be banned in our lifetimes. Sure, it may be restricted. Perhaps to only one lane on the motorway, outside of towns etc, but that's not exactly unique. We already have plenty of roads where certain types of transportation are banned. Whether that be scooters, bikes, tractors etc on some dual carriageways and motorways, whether that be cars in town centrest, or other examples.

OK. 30-40 years is a lot more realistic a timescale than some expectations I've heard lately :) There are a lot of people who think this technology is just around the corner, but I think 30+ years is much more realistic.

As others have discussed, realistically 30-40 years will be the timeframe where most vehicles sold are likely to be level 4/5. Previous milestones will probably include (some already happened, some from company announcements and some hypothesised):

2017: Level 3 autonomous lorries are on the road (there's a person in the cab, but they don't have to be at the wheel), level 4 cars are driving around as "taxis" (a "driver" sits in the back seat for now, expected to change this year with rule changes), picking up real customers in select cities to prove proof of concept.
2019: GM are hoping to have level 4 ride hailing service in a limited number of cities, with no "operators" - legislation depending.
2021: Ford aim to have it's own ride hailing service up and running. https://media.ford.com/content/ford...onomous-vehicle-for-ride-sharing-in-2021.html
2022: ~5 years - from now the first level 4 lorries will be doing test runs on real roads without drivers/operators in the cab. There may well be the first level 4 consumer vehicle available from someone like Tesla, allowing fully autonomous running in select locations.
2027: ~10 years - the first commercial runs of level 4 lorries will start appearing. Probably about the time when several vehicles will be available to the public that are level 4/5 and become more mainstream.
2050 ~ most vehicles will have an AV options package for people to choose and vehicles become truly level 5, able to travel virtually anywhere (excluding some random mountain track covered in deep snow, for example). Most lorries and almost all taxis will be autonomous vehicles with no operator in the vehicle (but perhaps a connection to a central depot to take over if required in exceptional circumstances).
 
Caporegime
Joined
28 Feb 2004
Posts
74,822
Nice to hear that you are making good progress with your track testing. On your AV testing, do you test them on public roads in the UK too? If so how do you think your testing compares and contrasts with Waymo's.

I see that they are bringing their Pacifica SUVs to Atlanta, Georgia, an interesting city. This link and comments mentions that Atlanta is an "incredibly car-centric" city, with only 16% of households without vehicles. One of the comments in the link says: "if Waymo can get around the crazy Atlanta drivers, pot holes, and exploding bridges and water main breaks I’ll be impressed. Driving in ATL is always an adventure."

https://www.theverge.com/2018/1/22/16919716/waymo-self-driving-minivan-atlanta


Currently in the UK you are not allowed to have autonomous vehicles on the road without someone in the drivers seat, at all times, to take over if necessary, so we cannot test on public roads without someone in the drivers seat at all times.

However the Government recently announced that by 2021 the law will be changed to allow autonomous vehicles with no one anywhere in the vehicle at all, to drive on any road anywhere in the UK.

https://www.autocar.co.uk/car-news/industry/new-legislation-allow-driverless-cars-uk-roads-2021

That will allow us to dramatically increase our testing programme, and will mean the UK will be way ahead of many other Countries in the world in this field.
 
Soldato
Joined
12 Jul 2007
Posts
7,911
Location
Stoke/Norfolk
That will allow us to dramatically increase our testing programme, and will mean the UK will be way ahead of many other Countries in the world in this field.

If that field is "accidents involving AV's" then yeap, I agree.

When you mix AV's who have very strict rules and humans, who to be fair are flexible when it comes to some rules of the road and are therefore used to driving with other "rule flexible" humans, there will always be issues. A common one from the US is AV cars coming to a complete stop at junctions, while the human in the car behind can see there is no traffic and, expecting the "human" in front to have seen the same thing and therefore NOT come to a complete stop, leading to a rear-ending of the AV (the most common form of AV accident and also a big human one too).

So in the UK imagine a scenario where the AV car is coming up to a roundabout which has no traffic on it. Based on the US experience the UK car would still be programmed to come to a complete stop before moving off, where as the human behind can also see the empty roundabout and won't be expecting the car in front to stop leading to a bump.

None of that would be avoided by having a human driver in the AV but having that limitation currently reduces the amount of AV cars on the road. Without that limitation the number of AV's can sky-rocket and so will the number of accidents caused by this Strict rule vs Flexible rule interaction.
 
Soldato
Joined
9 Mar 2003
Posts
14,232

So you think it's acceptable not to leave sufficient space between yourself and the car in front in case it stops? The car in front does not need to justify the reason as to why it stopped. Its the person doing the rear ending 100% of the time.

I'm fairly sure that most of the test vehicles all have 'autonomous vehicle, may stop suddenly' all over the back of them just like learners driver cars do.

No excuse really.
 
Caporegime
Joined
28 Feb 2004
Posts
74,822
So we can expect to see AVs with no one in them just stuck on small country roads at some point then :p

Same chance of any other vehicle being stuck on a small country road.
But you will never need the RAC or anyone to come out to any of our test vehicles as we have full remote control functionality ;)
 
Caporegime
Joined
28 Feb 2004
Posts
74,822
If that field is "accidents involving AV's" then yeap, I agree.

When you mix AV's who have very strict rules and humans, who to be fair are flexible when it comes to some rules of the road and are therefore used to driving with other "rule flexible" humans, there will always be issues. A common one from the US is AV cars coming to a complete stop at junctions, while the human in the car behind can see there is no traffic and, expecting the "human" in front to have seen the same thing and therefore NOT come to a complete stop, leading to a rear-ending of the AV (the most common form of AV accident and also a big human one too).

So in the UK imagine a scenario where the AV car is coming up to a roundabout which has no traffic on it. Based on the US experience the UK car would still be programmed to come to a complete stop before moving off, where as the human behind can also see the empty roundabout and won't be expecting the car in front to stop leading to a bump.

None of that would be avoided by having a human driver in the AV but having that limitation currently reduces the amount of AV cars on the road. Without that limitation the number of AV's can sky-rocket and so will the number of accidents caused by this Strict rule vs Flexible rule interaction.

Where is this "strict rule" idea coming from?

All our test AV's currently going round our tracks alongside human driven cars, have systems that react almost exactly like you say humans do.
They will all quite happily carry on through a junction, or around a roundabout if the way is clear on approach, exactly like a proficient human would.
We have had far more occurrences of humans expecting the AV to stop, so coming to a stop themselves, and being very surprised when the AV just carries on and leaves the human parked at the junction of a clear road.
 
Soldato
Joined
12 Jul 2007
Posts
7,911
Location
Stoke/Norfolk
What you really mean by this is incompetent human drivers.

Yeap. Humans like patterns and if a pattern changes we're generally slow to adapt and as we've become used to "looking ahead" to see if we need to slow etc we generally expect other road users (human or AI) to do similar and when they don't it causes problems. If lots of humans overnight started to drive exactly as the highway code says there would be the same problems so it's not an AV problem, it's a human problem. However when you mix humans and AV's on the same road more and more, if the programming is too rigid it'll will naturally lead to more accidents in the short term with the vast majority caused by humans not expecting the AV to do something that human wouldn't, like the US complete stop example etc. After a while though we'll start to see the "pattern" of AV driving and the accident rate would drop.

So you think it's acceptable.......................

Your accusation is called a Straw man Fallacy. What you posted is not "what I think" and it is not "what I posted" either, it is a product of your own bias leading to a misinterpretation of my post so while I'm happy to discuss my opinion, please don't make things up. Anyway back to the discussion, if you read my post again without adding bias (i.e. don't think "he's already anti-AV") you'll see I only gave reasons "why" accidents happen but did not say "who was to blame" for them. However if you read my reply to Kenai above I think that should answer the "who is to blame" question and the reason why I believe that a greater mix of strict rule AV and flexible rule Human drivers would lead to more accidents in the short term, as has been shown in the US.

Where is this "strict rule" idea coming from?

From the US testing being carried out as they are the biggest tester of AV's in the world currently and in the US they are programmed to obey every rule exactly, no "flexibility" at all. However if UK AV's are allowed some flexibility (such as the junction approach you mentioned) that's a great step forward over the US testing and will make a much bigger difference to safety.
 
Soldato
Joined
5 Apr 2009
Posts
24,856
Yeap. Humans like patterns and if a pattern changes we're generally slow to adapt and as we've become used to "looking ahead" to see if we need to slow etc we generally expect other road users (human or AI) to do similar and when they don't it causes problems. If lots of humans overnight started to drive exactly as the highway code says there would be the same problems so it's not an AV problem, it's a human problem. However when you mix humans and AV's on the same road more and more, if the programming is too rigid it'll will naturally lead to more accidents in the short term with the vast majority caused by humans not expecting the AV to do something that human wouldn't, like the US complete stop example etc. After a while though we'll start to see the "pattern" of AV driving and the accident rate would drop.
The AVs being tested according to your strict rule theory don't have a high accident rate though do they? They have a high level of attention and publicity but I'm not aware of any statistics showing the actual rate of accidents is higher than expected of a human equivalent.
 
Caporegime
Joined
25 Jul 2005
Posts
28,851
Location
Canada
What you really mean by this is incompetent human drivers.

Or law breaking drivers.

The US doesn’t do roundabouts or many Give Way/Yield signs so I’d be inclined to suggest what actually happens is the AV car draws to a stop at a stop sign and the driver behind then rear ends them because they were intending to blow the stop sign...

I’m sure there data somewhere for Ian to provide to show this is not the case though...
 
Soldato
Joined
12 Jul 2007
Posts
7,911
Location
Stoke/Norfolk
The AVs being tested according to your strict rule theory don't have a high accident rate though do they? They have a high level of attention and publicity but I'm not aware of any statistics showing the actual rate of accidents is higher than expected of a human equivalent.

It's a bit old (2015) but there have been reports about the accident rates per mile being higher that human - https://www.usatoday.com/story/money/cars/2015/10/31/study-self-driving-cars-accidents/74946614/ - however the very small number of AV's driving a very small amount of miles (compared to the total human cars/distances) does skew the results I'd say. Again, these have all been accidents where a human has hit an AV an I don't think there's been any yet with an AV being the "guilty" party.

Once we're as "fully AV" as possible (I think there will always be some humans drivers) the roads should be much safer for everyone.

I’m sure there data somewhere for Ian to provide to show this is not the case though...

You know it's funny but, if people would actually read my posts you'll see that I have NEVER give my opinion of AV's either Pro or Anti, so I'm amazed that people can, from thin air, decide that I am somehow "Anti" and use that as a reason, not to discuss or debate but to accuse and attack, but this is the internet after all so people will feel that they need to "win" to fill a sad void in their life I suppose, even if they have to make things up to do so.
 
Caporegime
Joined
25 Jul 2005
Posts
28,851
Location
Canada
You know it's funny but, if people would actually read my posts you'll see that I have NEVER give my opinion of AV's either Pro or Anti, so I'm amazed that people can, from thin air, decide that I am somehow "Anti" and use that as a reason, not to discuss or debate but to accuse and attack, but this is the internet after all so people will feel that they need to "win" to fill a sad void in their life I suppose, even if they have to make things up to do so.

I didn't accuse you of being anti...

You've made a statement about the "flexibility" of human drivers. I've just suggested that when you say flexible you mean illegal activities, and that the example you gave (stopping before entering a roundabout) may not be relevant to the situations in the US, where the road signage and systems are very different (basically no roundabouts and stop signs everywhere).

I'm asking you to provide evidence for what you've said, nothing else.
 
Soldato
Joined
5 Apr 2009
Posts
24,856
It's a bit old (2015) but there have been reports about the accident rates per mile being higher that human - https://www.usatoday.com/story/money/cars/2015/10/31/study-self-driving-cars-accidents/74946614/ - however the very small number of AV's driving a very small amount of miles (compared to the total human cars/distances) does skew the results I'd say. Again, these have all been accidents where a human has hit an AV an I don't think there's been any yet with an AV being the "guilty" party.

Once we're as "fully AV" as possible (I think there will always be some humans drivers) the roads should be much safer for everyone.

The last study I've found in a brief search was early 2016 (commissioned by Google but conducted independently I think) which showed AVs (US based) at 3.2 accidents per million miles vs 4.2 for humans. I would expect 2 years later that's probably only improved.

I think the suggestion that introduction of AVs will lead to a spike in accidents (or that they'd 'sky rocket' as you first hinted at) isn't really backed up by numbers even if thinking behind the ideas makes sense.
 
Joined
4 Aug 2007
Posts
21,427
Location
Wilds of suffolk
Personally I don't see an AV as any more likely to cause an accident than another random driver your following.
You never know how they are going to react.

Such as the person who without indicating at all did a 360 on a mini roundabout in front of me yesterday morning.
He flew into the roundabout, went round virtually on 2 wheels and then sped through a residential area so quickly that by the time I was halfway up the hill, he had disappeared over the top (and I was doing 30 and don't exactly hold back on accelerating)
At the top of the hill is a double hazard, road swerves left and people park on the edge of the road on the apex, and then its zigzagged yellows due to a primary school.

The sooner people like this lose the ability to drive the sooner we will all be safer ;)
 
Soldato
Joined
22 Nov 2006
Posts
23,372
Maybe it was stolen.

Which won't happen with AVs, because they will just hack it remotely and tell it to drive directly to the port :p
 
Caporegime
Joined
28 Feb 2004
Posts
74,822
Won't even get to court, there must be dozens of witnesses to say the motorcyclist illegally moved lanes, not to mention all the data from the autonomous vehicle itself most likely including camera views etc, showing the track and speed of the motorcyclist so he will be deemed to be at fault, as the article already says he was, therefore no case end of story.

Just some jumped up lawyer trying to make a name for himself I bet.

Will always be the case with any AV being involved in any incident, there will be so much data to show exactly what happened, including video footage, information on what signals were showing where, speeds, directions, etc of not only the AV but every other vehicle in the near vicinity, that it will be virtually impossible to fake an accident, or to put the blame on the AV when it so clearly wasn't.

Gone are the days of his word against yours. Now with tonnes of data to prove categorically exactly what happened almost every incident will be a cut and shut, case closed immediately.
 
Last edited:
Back
Top Bottom