How should we make the transition from driving to driverless vehicles in the future?

[PTR 444]

Make no mistake, driverless road vehicles are increasingly closer to becoming a reality in the future, and hence at some point we will need to find a way to gradually roll these out on our highways. The ideal scenario is for a vehicle to take you from door to door with no driving required, but as we all know, introducing AVs to environments where manual vehicles, pedestrians and cyclists already exist is going to be problematic for many reasons already discussed at length.

Ideally you are going to want AVs to run where there is no risk of external obstructions, but this rules out urban areas where most of the owners/users will be likely to live. I can therefore see driverless vehicles starting out in one of the following scenarios:

• No fully autonomous vehicles are allowed and any AVs must have a manual driving option. Once ownership is at a high level, the UK motorway network is adapted to allow autonomous driving only as this will increase capacity on such roads. On all other roads, only manual driving is possible. This option could make future motorways cheaper and easier to build since they would not need to be as wide due to computers taking over lane discipline and working out stopping distances.
• A phased trial/introduction of autonomous vehicles takes place but only on sparsely populated islands, such as Orkney and Shetland. As traffic levels are low there, manual and automatic vehicles could co-exist without causing too much issues. If this is successful, this could be rolled out to more densely populated islands, such as the Isle of Wight, Jersey and Guernsey where all manual vehicles would need to be replaced with AVs overnight. If successful, this would eventually be phased into the rest of the UK.
• Autonomous and manual vehicles co-exist but cannot share each other’s roads. Basically a separate road network has to be built to cater for AVs, which could be made up by converting controlled access roads and some unprofitable railway lines. The manual road network would remain for privately owned cars, while the AV network could become an on-demand transit service where people hail cars to take them places. It would cause a lot of inconvenience to people who do not live near an AV road, but for anyone living near one, it could become more useful than the railway as one who doesn’t drive would be able to make a wider range of journeys without having to change en-journey.

Please let me know what you think of these suggestions and feel free to share your own ideas.

 

[PeterC]

AVs will always have a manual driving option if only restricted to walking pace. This would be needed for a number of edge cases such as moving to the right bay in a workshop or for using temporary off road parking .

 

[87 027]

I can't see how fully autonomous vehicles could easily get along with the new Highway Code on 'mixed traffic' roads. Person standing on the corner of a junction of the road you are intending to turn into - are they waiting to cross or not? Driving down a country lane - cyclists can pootle along in the middle of the road in front of you with impunity.

I can see a case for co-existence on more restricted roads such as motorways. However look at all the controversy about HS2 for creating brand new dedicated high speed infrastructure.

I think the first of your three scenarios is the most likely in my lifetime subject to the caveat that drivers will always need to take back control if required.

 

[birchesgreen]

They'll perfect the technology the day after nuclear fusion goes mainstream.

 

[GrimsbyPacer]

The future I see has a standard lorry driver in a lorry with some assisted driving technology which communicates to other vehicles, and 40 computer controlled lorries following in the slipstream down a motorway between say Tilbury and a lorry depot in Birmingham.
A human will always be in control of any vehicles on the road, but with freight it's possible to link them together in trains, no couplers, but a microwave link sending other data on speed, direction, and breaking.

 

[Bletchleyite]

I think there are too many variables and complexities of fully automated cars so we won't see them in the mainstream for a very long time. Never say never, but I reckon 30 years off as a minimum.

In the meantime, the technology will be used to make manual driving safer, e.g. automatic speed limiting and automatic braking to avoid accidents. I would also expect adaptive cruise control and lane departure warning (or possibly even automatic lane holding) to become standard on all cars fairly soon, which would reduce motorway "rear endings".

 

[biko]

As others mentioned, I also believe we will see first sort of trains of HGVs on motorways and later on, autonomous driving on motorways and maybe some other high-standard roads, as these have no pedestrians or cyclists.

In urban areas, I don’t think we can give a good prediction when or if autonomous vehicles will be able to take over.

I can't see how fully autonomous vehicles could easily get along with the new Highway Code on 'mixed traffic' roads. Person standing on the corner of a junction of the road you are intending to turn into - are they waiting to cross or not?

Also without this new rule, it would have been a problem as autonomous vehicles will need to be (nearly) fail-safe. Also this rule is a long-standing one in many other countries in Europe, so the manufacturers will need to cope with it in any case.

 

[Jozhua]

I think there are too many variables and complexities of fully automated cars so we won't see them in the mainstream for a very long time. Never say never, but I reckon 30 years off as a minimum.

In the meantime, the technology will be used to make manual driving safer, e.g. automatic speed limiting and automatic braking to avoid accidents. I would also expect adaptive cruise control and lane departure warning (or possibly even automatic lane holding) to become standard on all cars fairly soon, which would reduce motorway "rear endings".

I agree - ultimately being able to have a computer checking a human's work is better than relying on a human checking a computer's work. People get distracted too easily when they stick autopilot on!

Make no mistake, driverless road vehicles are increasingly closer to becoming a reality in the future, and hence at some point we will need to find a way to gradually roll these out on our highways. The ideal scenario is for a vehicle to take you from door to door with no driving required, but as we all know, introducing AVs to environments where manual vehicles, pedestrians and cyclists already exist is going to be problematic for many reasons already discussed at length.

Ideally you are going to want AVs to run where there is no risk of external obstructions, but this rules out urban areas where most of the owners/users will be likely to live. I can therefore see driverless vehicles starting out in one of the following scenarios:

• No fully autonomous vehicles are allowed and any AVs must have a manual driving option. Once ownership is at a high level, the UK motorway network is adapted to allow autonomous driving only as this will increase capacity on such roads. On all other roads, only manual driving is possible. This option could make future motorways cheaper and easier to build since they would not need to be as wide due to computers taking over lane discipline and working out stopping distances.
• A phased trial/introduction of autonomous vehicles takes place but only on sparsely populated islands, such as Orkney and Shetland. As traffic levels are low there, manual and automatic vehicles could co-exist without causing too much issues. If this is successful, this could be rolled out to more densely populated islands, such as the Isle of Wight, Jersey and Guernsey where all manual vehicles would need to be replaced with AVs overnight. If successful, this would eventually be phased into the rest of the UK.
• Autonomous and manual vehicles co-exist but cannot share each other’s roads. Basically a separate road network has to be built to cater for AVs, which could be made up by converting controlled access roads and some unprofitable railway lines. The manual road network would remain for privately owned cars, while the AV network could become an on-demand transit service where people hail cars to take them places. It would cause a lot of inconvenience to people who do not live near an AV road, but for anyone living near one, it could become more useful than the railway as one who doesn’t drive would be able to make a wider range of journeys without having to change en-journey.

Please let me know what you think of these suggestions and feel free to share your own ideas.

So what you're saying that is if you remove all other traffic besides driverless cars, they will be feasible?

In that case, you could achieve automation with 70's technology - see the Morgantown PRT.

If you need to build a dedicated highway network, at that point, it would be cheaper to build automated light metro, like the DLR or SkyTrain.

Perhaps there may be dedicated AV lanes on motorways and the like. But ultimately any such system will have the floor wiped clean by high speed rail in terms of capacity and cost any day.

 

[AM9]

As others mentioned, I also believe we will see first sort of trains of HGVs on motorways and later on, autonomous driving on motorways and maybe some other high-standard roads, as these have no pedestrians or cyclists.

So with a continuous train of HGVs all travelling in the slipstream of the vehicle in front, how will other traffic get onto and off the motorway?

 

[biko]

So with a continuous train of HGVs all travelling in the slipstream of the vehicle in front, how will other traffic get onto and off the motorway?

It won’t be continuous trains, but groups that coordinate their driving, but also coordinate sufficient gaps for other road users.

 

[AM9]

It won’t be continuous trains, but groups that coordinate their driving, but also coordinate sufficient gaps for other road users.

So how many are likely to be in a typical convoy?

 

[biko]

So how many are likely to be in a typical convoy?

Not sure about that, but testing of these kinds of convoys is already happening in some places. I think it will be around 3 to 5 HGVs

 

[Annetts key]

Driverless road vehicles on normal roads are at least ten years away from being introduced.

Make no mistake, driverless road vehicles are increasingly closer to becoming a reality in the future, and hence at some point we will need to find a way to gradually roll these out on our highways. The ideal scenario is for a vehicle to take you from door to door with no driving required, but as we all know, introducing AVs to environments where manual vehicles, pedestrians and cyclists already exist is going to be problematic for many reasons already discussed at length.

Ideally you are going to want AVs to run where there is no risk of external obstructions, but this rules out urban areas where most of the owners/users will be likely to live. I can therefore see driverless vehicles starting out in one of the following scenarios:

• No fully autonomous vehicles are allowed and any AVs must have a manual driving option. Once ownership is at a high level, the UK motorway network is adapted to allow autonomous driving only as this will increase capacity on such roads. On all other roads, only manual driving is possible. This option could make future motorways cheaper and easier to build since they would not need to be as wide due to computers taking over lane discipline and working out stopping distances.
• A phased trial/introduction of autonomous vehicles takes place but only on sparsely populated islands, such as Orkney and Shetland. As traffic levels are low there, manual and automatic vehicles could co-exist without causing too much issues. If this is successful, this could be rolled out to more densely populated islands, such as the Isle of Wight, Jersey and Guernsey where all manual vehicles would need to be replaced with AVs overnight. If successful, this would eventually be phased into the rest of the UK.
• Autonomous and manual vehicles co-exist but cannot share each other’s roads. Basically a separate road network has to be built to cater for AVs, which could be made up by converting controlled access roads and some unprofitable railway lines. The manual road network would remain for privately owned cars, while the AV network could become an on-demand transit service where people hail cars to take them places. It would cause a lot of inconvenience to people who do not live near an AV road, but for anyone living near one, it could become more useful than the railway as one who doesn’t drive would be able to make a wider range of journeys without having to change en-journey.

Please let me know what you think of these suggestions and feel free to share your own ideas.

I don’t see any of these ‘options’ as being likely. Are you really thinking it likely that the government will prohibit manual road vehicles from using the public road network anywhere in the country?

I can just imagine the outcry from classic car groups and from consumer groups speaking up for the people who can’t afford a new car.

I also don’t see there being a very extensive private road network for driverless road vehicles. It rather defeats the purpose if you have to manually drive to/from your starting point to a private road, and then from the private road to your destination…

You also have to consider that other technologies will not stand still. So there may be even less need to travel in the future. It’s entirely possible that by the time driverless road vehicles are available to use on our roads, the amount of commuting may have fallen drastically.

An ever increasing amount of goods is being delivered directly to people’s homes. Conventional shops may have changed radically by then.

Better virtual/3D video services may have significantly increased home working.

They'll perfect the technology the day after nuclear fusion goes mainstream.

Ahh, you mean that once the vehicles can be powered by a nuclear fusion ‘engine’

 

[Bletchleyite]

I don’t see any of these ‘options’ as being likely. Are you really thinking it likely that the government will prohibit manual road vehicles from using the public road network anywhere in the country?

There is one very big elephant in the room with this - cycling. Cycling is to be encouraged, and will by definition never be automated. Thus, unless we were to build cycle lanes on the entire road network, we can never, other than motorways and some special roads, ban manual road vehicles.

 

[Meerkat]

Not sure about that, but testing of these kinds of convoys is already happening in some places. I think it will be around 3 to 5 HGVs

Merging onto a motorway where there are convoys of 3-5 HGVs is going to be very dangerous, and cause much congestion

 

[Bletchleyite]

Merging onto a motorway where there are convoys of 3-5 HGVs is going to be very dangerous, and cause much congestion

If this system forced the "HGV trains" to comply properly to the Two Second Rule, it would actually be easier. So often lane 1 is just a continuous parade of HGVs far too close together.

 

[87 027]

Isn't the 2 second rule mainly to allow for driver reaction time which wouldn't be as relevant for a convoy of automated vehicles?

 

[Bletchleyite]

Isn't the 2 second rule mainly to allow for driver reaction time which wouldn't be as relevant for a convoy of automated vehicles?

You would need it between the convoys unless manually driven vehicles were prohibited from the road. You'd not need it within the convoy.

 

[bspahh]

Fully automated cars could be used for valet parking, particularly for charging electric cars when there is a limited number of charging points. The car park can be set up with beacons so that the cars can get their exact position without having to rely on GPS or cameras. The cars can be parked closer than would be possible with a driver, so the price could be cheaper than a normal parking space. This would be particularly good for car clubs.

 

[Meerkat]

If this system forced the "HGV trains" to comply properly to the Two Second Rule, it would actually be easier. So often lane 1 is just a continuous parade of HGVs far too close together.

Not easier to merge if you have to be in front or behind a five truck convoy!
Credit to truck drivers they are generally pretty good at creating spaces for merging cars, slowing down or moving out a lane - indeed clockwise junction 9 M25 I almost always needed a truck to be generous to be able to get out the dreadful entry slip!

Autonomous vehicles are going to get the mick taken out of them once people get used to them. Pedestrians will know they can make them stop, and manual drivers will know they can cut them up - I was on the M1 once and a whole line of cars steadily cut into the outside lane in front of a Range Rover that must have had adaptive cruise on as it would just drop back and recreate the space for the next cutter in.

 

[Bletchleyite]

Autonomous vehicles are going to get the mick taken out of them once people get used to them. Pedestrians will know they can make them stop, and manual drivers will know they can cut them up - I was on the M1 once and a whole line of cars steadily cut into the outside lane in front of a Range Rover that must have had adaptive cruise on as it would just drop back and recreate the space for the next cutter in.

That sounds good. Can't see a problem with that. An efficient merge, and good safety.

The drop back will have cost him about 20 seconds on his journey if that, and has resulted in safer use of the road.

 

[Meerkat]

That sounds good. Can't see a problem with that. An efficient merge, and good safety.

The drop back will have cost him about 20 seconds on his journey if that, and has resulted in safer use of the road.

The drop backs will be after the fact - they dont necessarily make the merge safer (technically it can't be as the separation doesn't double instantly).
In this instance it was possibly iffy on being undertaking - can't remember if it was really busy enough to justify but some of those taking part were undertaking to hit this gap that kept appearing.

 

[Bletchleyite]

The drop backs will be after the fact - they dont necessarily make the merge safer (technically it can't be as the separation doesn't double instantly).

They do compared to the more usual approach of people not dropping back at all.

In this instance it was possibly iffy on being undertaking - can't remember if it was really busy enough to justify but some of those taking part were undertaking to hit this gap that kept appearing.

You are allowed to go to the end of the sliplane to merge if that suits your purposes, passing traffic in the left lane if it suits. Undertaking (actually "overtaking on the left") requires one to actually move one or more lanes to the left *specifically* to pass a vehicle, and isn't actually an offence, though would likely be seen as driving without due care and attention.

 

[Meerkat]

They do compared to the more usual approach of people not dropping back at all.



You are allowed to go to the end of the sliplane to merge if that suits your purposes, passing traffic in the left lane if it suits. Undertaking (actually "overtaking on the left") requires one to actually move one or more lanes to the left *specifically* to pass a vehicle, and isn't actually an offence, though would likely be seen as driving without due care and attention.

Just to clarify this wasnt from a slip lane - it was people undertaking in lane 3 and then cutting into lane 4 in busy traffic. Thinking about it it would be a feedback loop as it would make lane 4 go slower and lane 3 faster.....

 

[Jozhua]

Fully automated cars could be used for valet parking, particularly for charging electric cars when there is a limited number of charging points. The car park can be set up with beacons so that the cars can get their exact position without having to rely on GPS or cameras. The cars can be parked closer than would be possible with a driver, so the price could be cheaper than a normal parking space. This would be particularly good for car clubs.

I can't imagine this being a cheaper option than a typical car park. You need cars to come with a few grand's worth of equipment, alongside adding significantly more technical complexity to a car park. You now also make the car park only usable by people in possession of said technology.

The easier option would be those car lifts, seen in the VW factory, where the car is driven on to the platform, then lifted to a specific slot.

 

[bspahh]

I can't imagine this being a cheaper option than a typical car park. You need cars to come with a few grand's worth of equipment, alongside adding significantly more technical complexity to a car park. You now also make the car park only usable by people in possession of said technology.

The easier option would be those car lifts, seen in the VW factory, where the car is driven on to the platform, then lifted to a specific slot.

Car lifts are heavy engineering, which needs to be maintained and locks you in to a maximum size and weight of car.

The hardware for a self-driving car will be commonplace soon. An modern electric car already has enough sensors and cameras, and control of the acceleration and braking. If it doesn't already have control of the steering, then it won't be expensive to add in. Its not something that will be retrofitted to existing cars.

The software for a self-driving car is hard for use in an open environment, and the legal and insurance side of things would be difficult. In a closed environment, its a lot simpler.

This sort of thing will appear first where parking space is limited and expensive, for the convenience of being able to drop off a car, and pick it up fully charged, without having to pay for staff to drive them around.

 

[biko]

Merging onto a motorway where there are convoys of 3-5 HGVs is going to be very dangerous, and cause much congestion

No, the idea is that either the trains are not too long, or the following HGVs make some space if they detect a car wanting to join. See for example this explainer video (around 3:09 a car merges):

Congestion will actually reduce because headways are reduced.

 

[stuu]

My employer (big 4 accountant) released a report 7 or 8 years ago proclaiming 2020 to be the year autonomous cars would be mainstream and widely accepted. Can't find it on the website any more, oddly

 

[Meerkat]

No, the idea is that either the trains are not too long, or the following HGVs make some space if they detect a car wanting to join. See for example this explainer video (around 3:09 a car merges):

Congestion will actually reduce because headways are reduced.

It doesn't explain how its done. Does the car have to shove its way into the subsafe gap and be unsafe until the truck has slowly adjusted?
From that video the front driver is on the clock, suggesting that rotation will be desired, and on the fly platooning will require trucks to slow down for others to catch up. Both of these things will lead to platoons overtaking or being overtaken by other trucks -the thing that really really increases congestion.

Also allowing cars into the middle of the platoon might have legal impacts on whether the following drivers are driving or not.

 

[Ediswan]

Also allowing cars into the middle of the platoon might have legal impacts on whether the following drivers are driving or not.

I wonder if trucks in a platoon will show some kind of light to indicate that situation. In other words, how would a car driver know that they have joined in the middle of a platoon ? Would there be any legal obligation on a car driver to change lane and allow the platoon to re-form ?