Automated Vehicles Act 2024

[DerekC]

The Automated Vehicles Act 2024 came into force this week (20th May 2024) but didn't get much airtime. Here's the government announcement:

https://www.gov.uk/government/news/...by-2026-as-automated-vehicles-act-becomes-law

"Self-driving vehicles set to be on roads by 2026 as Automated Vehicles Act becomes law​

Road safety is at the heart of the legislation, with automated vehicles expected to improve road safety by reducing human error.
From:Department for Transport, Centre for Connected and Autonomous Vehicles and The Rt Hon Mark Harper MPPublished20 May 2024

• self-driving vehicles could be on British roads in just 2 years as new law receives Royal Assent
• move could create over 38,000 new jobs, cementing Britain’s position at the forefront of the self-driving tech industry
• travel set to be revolutionised by £42 billion sector, increasing road safety and unlocking opportunities for those who currently can’t drive

Self-driving vehicles could be on British roads by 2026, after the government’s world-leading Automated Vehicles (AV) Act became law today (20 May 2024).
Announced in the King’s Speech, the AV Act enables advanced technology to safely drive vehicles on British roads. The new law puts Great Britain firmly at the forefront of self-driving technology regulation, unlocking the potential of an industry estimated to be worth up to £42 billion and creating 38,000 more skilled jobs by 2035.
Road safety is at the heart of the legislation, with automated vehicles expected to improve road safety by reducing human error, which contributes to 88% of road collisions.
The law will require self-driving vehicles to achieve a level of safety at least as high as careful and competent human drivers, as well as meeting rigorous safety checks before being allowed onto roads. Therefore, in the future deaths and injuries from drink driving, speeding, tiredness and inattention could be drastically reduced."

The Act itself is published here:
https://bills.parliament.uk/publications/54189/documents/4416

What it actually does is to create a legal framework within which automated vehicles can be authorised for operation and legally operated, defining the responsibilities of the various parties involved. You might have thought that the Office for Rail and Road might have had a part to play, reflecting its role in Rail, but it doesn't. Also there is no equivalent to RAIB. The Secretary of State gets all the powers to define principles, issue licenses, investigate accidents and enforce the rules.

I found a summary of the Act here which is better than I could do:

https://blog.burges-salmon.com/post/102is9q/automated-vehicles-bill-overview

There are seven Parts of the Bill. Namely:

1. Regulatory Scheme – Chapter 1 establishes the fundamental principles, safety expectations, and regulatory framework for a vehicle to be licensed as "self-driving" in Great Britain. Central to the framework will be the newly introduced concept of an authorised self-driving entity (ASDE) who will be legally responsible for the AV. Chapter 2 provides for the implementation of ‘no-user-in-charge’ (NUiC) vehicles, for which there will have to be a licensed operator that has “oversight” of each vehicle (for example, ensuring it is properly maintained and insured). Chapters 3 to 7 then focus on enforcement; with Chapter 3, for example, seeking to ensure proper information provision by ASDEs and licensed operators, both in the initial licensing stage and subsequent inspections, and providing sanctions for failure to do so.

2. Criminal Liability for vehicle use – Chapter 1 provides a ‘user-in-charge’ (UIC) definition, immunity for the UIC in certain instances, exceptions to the immunity, and concludes with establishing when the UIC will be liable due to still being legally defined as a ‘driver’ (including via being a ‘deemed driver’ when they are not in a position to be in control of the AV, but should have been). Chapter 2 then provides the offences themselves, via adding AV-specific offences to existing road offences (such as ‘dangerous use’).
3. Policing and Investigation – Chapter 1 provides powers to order AVs to stop when suspected of an offence (namely, via ‘appropriate communication with equipment of the vehicle’) and also to seize and detain AVs (if there is reason to believe the AV has or is about to commit an offence). Chapter 2 then introduces the role of ‘automated vehicle incident inspectors’, whose investigations must seek to ‘determine what caused’ a relevant incident (not seek to lay blame).

4. Marketing Restrictions – creates an offence where a vehicle is incorrectly described in terms specified by the Secretary of State (with the Explanatory Notes citing “self-driving” as one likely such term) and a more general offence of producing (or causing to production of) a communication that would be likely to confuse the public as to a non-AV’s autonomous capabilities and licensing status.

5. Permits for Automated Passenger Services (APS) – provides the power to the ‘appropriate national authority' to grant such permits. It also confirms the disapplication of taxi etc legislation as recommended by the Law Commission (meaning that an APS could not be prosecuted for unlawfully plying for hire). Finally, it also requires a second level of consent where the APS ‘resembles’ a taxi or bus service (with this secondary consent need to come from each local body that the resembling service operates in).
6. Adaption of existing regimes – provides the Secretary of State with the power to amend ‘type approval legislation’ (as the EU did in September 2022) and extends the powers of ‘vehicle examiners’ to AVs (and allows them to prohibit one from driving if it fails an examination).

7. General Provisions – provides definitions and confirms the territorial scope (extends to England, Wales and Scotland, with a few exceptions).

It is notable (speaking as an engineer) for not including any technical requirements other than one clause on transitions from non-user-in-charge to user-in-charge operation. The Secretary of State has an awful lot to do, including defining the principles for safe operation. I can't see anything other that maybe an automated taxi service in one or two places getting going by 2026 as promised in the hype.

From a technical perspective, the big problem I haven't heard an answer to is how the automated vehicle copes when it detects an incipient failure when cruising in Lane 3 of a motorway. The system must alert the user-in-charge (aka the driver), wait for a response and if one isn't forthcoming, "deal safely with the situation", to use the words in the Act. That doesn't (it seems to me) include stopping in a live lane. So you need highly redundant systems that can (at least) drive the car safely to the next exit.

 

[61653 HTAFC]

Quite possibly the biggest example of legislation putting the cart before the horse in history. As for it creating jobs, how exactly will that work? Maybe for lawyers and the people who repair crash barriers I suppose...

The argument that self-driving cars would be inherently safer also doesn't stand up to scrutiny. Whilst human error is a factor in many crashes, the errors that humans tend to make are way more predictable than the errors an autopilot system throws up. In particular any "transitional period" where roads are populated by a mixture of human drivers and autonomous vehicles, will produce chaos of the sort not seen since Sweden switched to driving on the right.

 

[RailWonderer]

I can't wait for the lawsuits and litigation in years to come...
Why does the UK always have to address everything before getting it right?

 

[DerekC]

Quite possibly the biggest example of legislation putting the cart before the horse in history. As for it creating jobs, how exactly will that work? Maybe for lawyers and the people who repair crash barriers I suppose...

The idea is that we continue to be a developer of some of the technology and an early adopter - hence creating jobs. I am not at all clear that the framework they are trying to create will work, in that I think it underestimates the systems engineering and regulatory effort needed to ensure that the whole shebang will work safely, but the alternative is to try to pretend that this technology will go away.

The argument that self-driving cars would be inherently safer also doesn't stand up to scrutiny. Whilst human error is a factor in many crashes, the errors that humans tend to make are way more predictable than the errors an autopilot system throws up.

I don't think that's true. Well-designed automated systems have predictable failure modes. Humans suffer not only from error, but from anything ranging from bad temper through fits of road rage to deliberately dangerous driving. You can reduce the incidence of human error to a great extent in a controlled environment such as train driving or piloting an aircraft through candidate selection and training in technical and non-technical skills, but it's much harder to do for the general public.

In particular any "transitional period" where roads are populated by a mixture of human drivers and autonomous vehicles, will produce chaos of the sort not seen since Sweden switched to driving on the right.

It's very true that the mixture of autonomous, semi-autonomous and human driving will be hard to handle, but short of banning any further development and use of autonomous vehicles in the UK, how can that situation be avoided? (BTW, the accident rate immediately after "Dagen H" when Sweden switched to driving on the right went down rather than up).

I can't wait for the lawsuits and litigation in years to come...
Why does the UK always have to address everything before getting it right?

So what would you do?

 

[61653 HTAFC]

The idea is that we continue to be a developer of some of the technology and an early adopter - hence creating jobs. I am not at all clear that the framework they are trying to create will work, in that I think it underestimates the systems engineering and regulatory effort needed to ensure that the whole shebang will work safely, but the alternative is to try to pretend that this technology will go away.

The sensible thing to do from a legislation point of view is to ban the technology until it can be proven to work safely... something that is far more complex than you seem to realise. There are so many variables at play on the road, and I trust a human to deal with those better by using their brain, than I would trust an algorithm-based operating system to prioritise those variables.

I don't think that's true. Well-designed automated systems have predictable failure modes. Humans suffer not only from error, but from anything ranging from bad temper through fits of road rage to deliberately dangerous driving. You can reduce the incidence of human error to a great extent in a controlled environment such as train driving or piloting an aircraft through candidate selection and training in technical and non-technical skills, but it's much harder to do for the general public.

"Well-designed automated systems" are all well and good in a closed system, but our public roads are not a closed system. So-called autonomous vehicles may well be able to "react" to something quicker than a human can, but it will never be able to interpret human behaviour better than a human can. By doing observations and paying attention, a human can detect a hazard coming from another human (for example, a pedestrian who isn't paying attention because they're looking at their smartphone) way better than any automated system could. This idea that humans are fundamentally flawed and that automated systems can read our behaviour better than we can, is misanthropic and dystopian.

It's very true that the mixture of autonomous, semi-autonomous and human driving will be hard to handle, but short of banning any further development and use of autonomous vehicles in the UK, how can that situation be avoided? (BTW, the accident rate immediately after "Dagen H" when Sweden switched to driving on the right went down rather than up).

I was referring to the chaos caused by the change, not the accident rate. Its no surprise the accident rate actually went down as everyone was being extra cautious due to the change... however contemporary news reports show gridlocked interchanges as people adjusted to the change.

 

[DerekC]

The sensible thing to do from a legislation point of view is to ban the technology until it can be proven to work safely... something that is far more complex than you seem to realise.

I don't think banning is either practical or realistic. It isn't going to happen, and if it did, how would we ever get to prove that the technology works safely? There must be a case for further, wider and more comprehensive pilots, but the legislation doesn't preclude that. What the overall situation does need is an intelligent approach to managing the risks, building up confidence feature by feature in a increasingly complex environments and sharing and feeding back lessons learned into the development processes. That seems to be a gap in the thinking, at least judging by what I have read so far.

There are so many variables at play on the road, and I trust a human to deal with those better by using their brain, than I would trust an algorithm-based operating system to prioritise those variables.

"Well-designed automated systems" are all well and good in a closed system, but our public roads are not a closed system. So-called autonomous vehicles may well be able to "react" to something quicker than a human can, but it will never be able to interpret human behaviour better than a human can. By doing observations and paying attention, a human can detect a hazard coming from another human (for example, a pedestrian who isn't paying attention because they're looking at their smartphone) way better than any automated system could.

Even if that's true (and I don't think we are too far from automated systems that can interpret behaviour, at least in a traffic environment) humans very regularly don't observe sufficiently, get distracted, bored or tired or are just are having an off-day.

This idea that humans are fundamentally flawed and that automated systems can read our behaviour better than we can, is misanthropic and dystopian.

I don't see that - we are talking about controlling the movement of a motor vehicle and predicting the likely movement of other vehicles and pedestrians, not understanding human pyschology. Automatons are demonstrably better humans at repetitive tasks that require consistent application of skill. Personally I enjoy driving in the right circumstances, but if automation can free me from the stressful, intensive but essentially boring task of driving a car along a crowded motorway in the rush hour, particularly if I have to do it every day, that's great and I can spend the time talking to other occupants - hardly misanthropic and dystopian.

There are certainly some large buts, though - to me it's mainly about what happens when the automated vehicle suffers a failure of some kind.

 

[Gloster]

The reason that road casualties fell after Dagen H was that considerable restrictions were placed on traffic. All non-essential traffic was banned from 01.00-06.00 on the day and everything stopped at 04.50 before slowly moving from the left-hand side to the right and then starting moving again at 05.00; the ban was considerable longer in Stockholm, Malmö and one two other places. After that, for around a fortnight (I think), there were lower national speed limits. It could well be argued that the fall was due to accidents falling below the level of seriousness that made them recordable. It can be added that most cars were already left-hand drive, so overtaking became much less chancy.

 

[Bikeman78]

So what would you do?

The first question is what problem are we trying to solve? Do we need automated vehicles.

As already discussed, the change over period will be difficult and probably measured in years rather than months. Would there be an outright ban on manual cars? Would that mean the end of classic car shows?

I have tried to consider a few scenarios.

Motorways
These are relatively easy. Automated vehicles would always keep a safe distance and stick to the defined rules of the road. Regarding what happens if one has a serious fault, would that be any different to an existing car? If a car runs out of fuel or has a catastrophic engine failure, then it will likely be stuck where it is, unless it can limp on to the hard shoulder.

Narrow country lanes
Presumably automated vehicles would drive on sight? What happens when two meet? Will automatic vehicles have a detailed plan of every single road and passing place in the country? Otherwise, how will the two vehicles decide which one is going to reverse? Each vehicle will only know how far back the previous passing place is. They will have no knowledge of the road ahead. With people driving, at least one of them is likely to know the local area and if all else fails, they can have a conversation.

Urban
This is by far the most difficult, especially during the transition period. People park on double yellows and on pavements, which could potentially confuse an automated vehicle. In heavy traffic, humans often help each other out by letting someone pull out of a junction or proceed along a single file section when the other person should give way. Will automated vehicles do that? Road markings would have to be maintained properly. I can think of three lane roads where the markings are barely visible.

 

[HSTEd]

The first question is what problem are we trying to solve? Do we need automated vehicles.

I'd argue the economic case is extremely compelling.
Driving vehicles around is one of the most labour intensive tasks undertaken by modern society.

Hundreds of millions, or even billions, of hours of productive time a year would be otherwise available for economic, or leisure, purposes if it could be automated.

 

[Zamracene749]

I've operated automated systems all my life, controlling what should be relatively predictable reactions and PID feedback loops (that's a technical term used to describe how controllers predict and smooth their outputs) in the chemical process industry. If automation was easy or reliable I would have been out of work years ago.
These systems fail, and fail surprisingly regularly.
Driving is easy so AI will be safe, wail the ones that then go on to say driving is stressful, tiring, dangerous!
Well, barbering is probably easy and doesn't need a degree. Would any of the guys on here trust a robot with AI and a cut throat razor on their neck, every day? Some things, for now, are better with a human touch. I'll be the chap with the beard
I'm no Luddite, but I think the solution to traffic is not to automate it, but to reduce the need for traffic in the first place.
Yet eg work from home is being cut back, because it doesnt fit with the 'management presence' career bullies.
Too many lorries and vans? Let's cut back on needless consumption, inbuilt obsolescence, fast fashion etc.
Use public transport? Yes, once it's reliable and affordable. The short sighted may suggest that without having to pay drivers, it will be. But, unless there is a statutary living wage regardless of employment status, once mass automation takes over there's gonna be a somewhat over subscribed job market putting pressure on wages. Perhaps a thriving service economy will replace that? Good luck with that if the economy is stuck in a cycle of unemployment.
What about maintaining the vehicles? Who is responsible if an autonomous vehicle causes injury death or expensive delays due to software or computer hardware failure? Perhaps all the redundant drivers will get employment in the insurance industry? After an accident caused by automation failure because it hadn't been programmed to foresee the situation, will the software developers face courts? Or the vehicle manufacturers? Can we be certain that upgrades will be safe and not hackeable? Or are we just going to say that the driver is still responsible for the vehicle? I'm sure the pilots on that Ethiopian airways flight a few years back that were overruled and killed by the computer in their 737MAX would be eased in their graves by that. Of course, that was only an airliner, not a domestic car or mass produced truck. I bet all the automobile researchers and developers are being far better supported than the engineers at Boeing were during that ill fated airliners development.
After all, it's not like automobile manufacturers have a cost/risk/death assessment for every penny they spent on their vehicles. Oh, hang on....
In summary, it's far too early. I do not trust that the politicians making the decisions understand enough nor are well advised enough to pass these laws yet.

 

[renegademaster]

In particular any "transitional period" where roads are populated by a mixture of human drivers and autonomous vehicles, will produce chaos of the sort not seen since Sweden switched to driving on the right.

You mean like in the numerous American cities with self driving cars already in use? And before anyone says "but Americans have wider roads and less complicated traffic than we do" , the main city for self driving tests is San Francisco and thats hardly McMansion suburbia.

 

[DelW]

You mean like in the numerous American cities with self driving cars already in use? And before anyone says "but Americans have wider roads and less complicated traffic than we do" , the main city for self driving tests is San Francisco and thats hardly McMansion suburbia.

The San Francisco trial hasn't been trouble free though. Cruise (one of the two operators of autonomous taxis there) had its trial suspended last autumn following a series of problems including accidents and obstructing emergency vehicles. I'm not sure whether that suspension has been lifted yet.

They aren't 100% autonomous either. If the vehicle's software can't work out a situation, it contacts a manned control centre where a human operator can take over control. That's only viable for a limited fleet size.

 

[renegademaster]

They aren't 100% autonomous either. If the vehicle's software can't work out a situation, it contacts a manned control centre where a human operator can take over control. That's only viable for a limited fleet size.

Theirs thousands of uber drivers in London, I'm sure some of them can be retrained to drive a car remotely from an office park in Hillingdon

 

[Ianigsy]

It makes a difference for the law to be ahead of the technology instead of ten years behind.

 

[DelW]

Another reference to the American trials not going as well as automated driving protagonists would have us believe:

For self-driving cars, the free ride is over

The probes are flying fast and free.

www.theverge.com

(link to tech website The Verge, partial quote below)

For years, autonomous vehicles have operated in relative obscurity. With few vehicles on the road and a laissez-faire attitude among government regulators, automakers and big tech firms have been free to test — and even commercially deploy — with little oversight.
Well, those days are done. In rapid succession, the National Highway Traffic Safety Administration (NHTSA) has opened investigations into almost all the major companies testing autonomous vehicles as well as those that offer advanced driver-assist systems in their production cars. Tesla, Ford, Waymo, Cruise, and Zoox are all being probed for alleged safety lapses, with the agency examining hundreds of crashes, some of which have been fatal.

In rapid succession, NHTSA opened investigations into almost all the major companies testing autonomous vehicles
The new investigations signal a new — and perhaps more antagonistic — phase in the relationship between safety regulators and the private sector. The government is requiring more data from companies, especially around crashes, in order to determine whether the industry’s safety claims live up to their hype. And the companies are finding that the proliferation of smartphones with cameras is working against them, as more videos of their vehicles behaving unpredictably go viral.

 

[renegademaster]

Another reference to the American trials not going as well as automated driving protagonists would have us believe:

For self-driving cars, the free ride is over

The probes are flying fast and free.

www.theverge.com

(link to tech website The Verge, partial quote below)

All this article says is that their has been 22 crashes investigated by the regulator, and nothing else really of substance. Without per capita comparisons this is pretty meaningless. I imagine at this stage pretty much everything will get investigated. Nobody has claimed they would be perfect, or have a forcefield that will stop others colliding with them but as long as they avoid killing a few people every day like human drivers do they are a positive development. Even if safety was equal to human drivers, which I highly doubt it will end up being, it would still be a net positive to free up people for more socially useful work.

I'm no Luddite, but I think the solution to traffic is not to automate it, but to reduce the need for traffic in the first place.
Yet eg work from home is being cut back, because it doesnt fit with the 'management presence' career bullies.
Too many lorries and vans? Let's cut back on needless consumption, inbuilt obsolescence, fast fashion etc.
Use public transport? Yes, once it's reliable and affordable. The short sighted may suggest that without having to pay drivers, it

Self driving car development isn't going to stop your pet policies, two things can happen at once