Written evidence submitted by Mr Christopher Taylor and
Mrs Nicole Taylor (SDV0025)

The development and ‘safe’ deployment of self-driving automated vehicles

Introduction

1.1 The Centre of Connected and Autonomous Vehicles (CCAV) recently revised its estimates for new sales of level 3 (L3) conditional, level 4 (L4) high and level 5 (L5) full automation of vehicles. They predict by 2035 L3/L4/L5 will account for 40% of new vehicle sales in the UK.1

1.2 With legislation for autonomous lane keeping in place,2 it will be possible to move trails into deployment and start to realise the potential £42 billion opportunity by 2035.1

1.3 Whilst large-scale deployment of Automated Vehicles (AVs) may feel little nearer than it did in 2017,3 global sales of Electric Vehicles (EVs) demonstrates this flightpath may be realistic.4 While it took six years for EVs to reach 0.9%, sales increased from 4% in 2020 to 9% in 2021. If that exponential growth continues, by 2023 EVs could account for over 40% of new global sales.

1.4 If adoption of AVs follows the same trajectory, their uptake could also suddenly multiply. It is therefore important to start preparing the regulations that will be required, as the changes needed will be significant (especially as deployment is likely to involve a transition period).

Likely uses, including private cars, public transport and commercial vehicles

Progress of research and trials in the UK and abroad

2.1 The CCAV predictions on uptake are based on two scenarios, termed by the Boston Consulting Group as “highway autopilot with lane changing” – L3 and “the reimagined car” – L4/5.5,6

2.2 At Young Driver Focus, Gary Burnett from the University of Nottingham showed a video of his research.7 The different times taken by volunteers, to hand over and take back control in an L3 scenario, demonstrated the issues related to shared control in the human-machine interface. These issues create legal challenges and serious safety concerns (see 5.3 and 5.4).

2.3 The gradual evolution of level 2 (L2) assisted automation is less risky than L3 and adopting common standards for L2 technology will enable future automation opportunities (see 2.4).8 The EU General Safety Regulation (GSR) and Pedestrian Safety Regulation (PSR) is “a new, game-changing package of technology and regulation” that is available for deployment now and will be “important to the UK’s ambitions for connected and automated vehicles”.8

2.4 Trialling and deploying L2 technology identifies risks, opportunities and potential refinements. The 5-year HelmUK case study identified how the building blocks of existing technology could be applied to enable ‘platooning’.9 Whilst the study concluded the short-term benefits were limited and it posed a potential risk at junctions, this learning will lead to future successes.

2.5 A number of trails have and are being run for niche L4 user cases, often at low speeds in urban areas.10,11,12,13,14 Whilst some of these and other niche use cases look promising, Starsky Robotics’ three vehicle trial encountered too many ‘edge’ cases (overcoming unexpected obstacles - like deer and the weather) and the company closed down.15

Potential implications for infrastructure, both physical and digital

3.1 For an L4/5 AV to be deployed on a single carriageway rural road, they will have to deal with the many and varied ‘edge’ cases faced every day by drivers, including:

potholes, debris, faint white lines and other hazards
responding to the actions of other road users
the weather - snow, ice, rain, flooding etc.
domesticated animals and wildlife
manoeuvres like overtaking stationary vehicles and other road users

especially vulnerable ones like: pedestrians, cyclists, horse riders and motorcyclists

being able to process multiple sensors (and decide which takes priority in an emergency)

3.2 Many of our roads (especially rural ones) do not meet the standards needed for AVs. Speed limits will also be a challenge, 60 mph is too fast for many rural roads and on these roads most drivers choose to drive well below this ‘limit’.16 To guide inexperienced drivers on appropriate speeds for rural roads, Quartix uses data from experienced drivers.17

3.3 Data on appropriate speeds for road conditions is just one example of the digital infrastructure needed for AVs to make decisions. This example raises several questions: How will data like this be captured from experienced drivers? Who will own this data? How will it be shared?

3.4 The use of personal data is one of topics considered by Artificial Intelligence (AI) ethics.17 Another is the need for transparency in how AI make decisions to achieve an outcome. Whilst AI is reliant on human expertise for both discovery and deployment, how it makes decisions can often be opaque and “beyond the interpretive capabilities of human scale reasoning”.18

3.5 The Law Commission report on AVs recommended a safety standard and “safety assurance scheme to provide regulatory oversight of automated vehicles throughout their lifetimes to ensure they continue to be safe and comply with road rules”.19 The role of the Driver and Vehicle Standards Authority (DVSA) will need to be extended to take on this regulation.

3.6 Both these recommendations have implications for the digital infrastructure. How software is regularly updated is an important consideration for this infrastructure. The AV will need to be connected to a network to ‘securely’ send and receive data and software updates, it may also need to communicate with other vehicles. AVs are likely to adopt an Internet of Things (IoT) edge computing model.20 Decision making is better done locally in real time, processing and analysing large amounts of data is resource intensive and better done centrally.

The regulatory framework, including legal status and approval and authorisation processes

4.1 Other Law Commission recommendations in their report on AVs were a two phase approval and authorisation process and new legal roles for users, manufacturers and service operators, with removal of criminal responsibility for the person in the passenger seat.19

4.2 This change involves a paradigm shift for criminal prosecutions, civil claims and the insurance industry. Prosecutions for corporate manslaughter “are rare and the evidential hurdles difficult to reach”.21 The Law Commission’s report recognised, whilst balance was required, there were gaps in corporate manslaughter, fraud and health and safety legislation for AVs and recommended “holding manufacturers and service operators criminally responsible for misrepresentation or non-disclosure of safety-relevant information.”19 Alongside the safety standard/safety assurance scheme, this will be a new regulatory role for the DVSA.

4.3 With roads providing the “tracks” for AVs, another paradigm shift may be required, the regulatory model for railways will become ever more applicable and appropriate to highways. The gaps identified by the Law Commission for corporate manslaughter, fraud and health and safety for AVs also apply to highways: compromising safety standards, accountability gaps, need for active involvement by senior management, asymmetry with how drivers are dealt with, asymmetry in access to data and moral blame in the face of serious wrong doing.19

4.4 The Office of Road and Rail (ORR) is responsible for the regulation of railways (including health and safety).22,23 This regulatory role does not exist for highways and criminal prosecutions cannot be brought using the Highways Act.24 Road deaths are not a priority for the Health and Safety Executive, as the Police take the lead on road collision investigations.25 However, their expertise and therefore the focus of their investigation relates to the Road Traffic Act.26

4.5 Whilst the ORR provides oversight of National Highways,27 following the abolition of the Audit Commission,28 there is no equivalent oversight for Local Highways Authorities (LHAs).

4.6 To prepare for the deployment of AVs the ORR’s remit needs to be expanded from oversight of National Highways to regulation of highways (including health and safety). The ORR’s remit will also need to be expanded to both licence service operators and promote competition, The regulation of AVs that travel on our highways will be a role for the DVSA (see 3.5 and 4.2).

4.7 “The Automated and Electric Vehicles Act 2018 (AEV Act) introduced new legislative provisions to smooth the path to compensation for those injured by automated vehicles (AVs). Broadly speaking, each listed AV must carry insurance. The insurer must then pay compensation for any damage caused by a vehicle while it is driving itself. The insurer may then bring a secondary claim against any person or body responsible for the incident.”18 There are several contributory factors insurers will want to consider: the user of the AV, other road users, vehicle design and maintenance, road design and maintenance, speed limits, post-crash response etc.

4.8 In 2021, 1,417 people were killed on the roads in England and Wales,29 and there were 720 prosecutions for roads deaths.30 For other fatalities the Inquest is the only opportunity to investigate their death. 285 Preventing Future Death reports with a category of road (highways safety) were submitted to the Chief Coroner's Office between August 2013 and April 2020.31 17% had vehicle and other contributory factors, the other 83% had environmental contributory factors with road design or maintenance and speed limits being a factor in all but 2 of them.

4.9 The DfT recently announced the establishment of a Road Safety Investigation Board (RSIB),32 who will “provide vital insight into safety trends related to new and evolving technologies, which could include self-driving vehicles…to ensure the country maintains some of the highest road safety standards in the world and exciting new technology is deployed safely.”32 Like the Coroner, rather than apportioning blame, the RSIB will be seeking to prevent future casualities.

Safety and perceptions of safety, including the relationship with other road users such as pedestrians, cyclists and conventionally driven vehicles

5.1 The Safe System is an holistic approach that applies five pillars to reduce road danger: safe roads, safe speeds, safe vehicles, safe road use and post-crash response.33 AVs “could improve road safety by reducing human error, which contributes to over 85% of accidents.”1 This would reduce the focus on road use and increase it on the other pillars.

5.2 Smart Motorways (designed to improve traffic flow and reduce the risk of shunts) resulted in an increased risk to the occupants of stranded vehicles.34 This experience demonstrates the need for vigilant monitoring and being able to take quick corrective action, as the deployment of new technology can have unwanted (though not necessarily unforeseen) consequences.

5.3 L3 will introduce similar risks. The Law Commission’s report on AVs, considers whether the failure by the user to take back control should be an offence and what actions should be taken by the automation.19 The one suggested, coming to a controlled stop, would create a situation very similar to a stranded vehicle on a Smart Motorway.

5.4 The Law Commission also considered the user needing to take back control in an emergency and was concerned there would be insufficient time to fulfil this task.19 In the video shared by Gary Burnett (see 3.1) the more engrossed the volunteer was in non-driving activities, the longer it took to take back control.7 A key finding in the VENTURER trial was transition was “achievable in optimum conditions, but not necessarily desirable in safety terms.”12

5.5 Shared control in the human-machine interface could be more challenging and less safe than a clear demarcation (i.e. as in L2 or L4/5). The gradual evolution of L2 would be a safer option. The inclusion of additional assistance will provide safety benefits, if the latest technology was adopted it is estimated if may be more beneficial than safety belts.9 The wide-scale adoption of proven technology, like Autonomous Emergency Braking, may also pave the way for L3, as this technology will reduce the risk associated with unexpected behaviour by an AV.

5.6 While rural roads carry 40% of traffic, they account for 60% of fatalities and with 795 fatalities in 2020 were 10 times more dangerous than motorways.35 A key finding of the VENTURER trial was AVs need to demonstrate more cautious behaviour than the average human driver.12 Some L2 assisted technology (adaptive cruise control, lane assist etc.) is not suited to narrow, winding single carriageways and trials of AVs do not yet provide the confidence needed for their deployment on rural roads. 10,11,12,13,14 Focus needs to be on first extensively proving L4/L5 at low speeds on urban roads using a service operator model.

The role of Government and other responsible bodies, such as National Highways and local authorities; and potential effects on patterns of car ownership, vehicle taxation and decarbonisation in the car market

6.1 The government needs to progress the recommendations in the Law Commission’s report on AVs and extend the role of the DVSA to deliver the regulatory framework required for AVs.19

6.2 The Law Commission did not consider the regulatory framework required for the highways, however, similar gaps to the ones it identified for AVs apply to highways.19 The ORR regulates the railways,22 its role needs to be extended to deliver the regulatory framework required for highways, including health and safety, and covering the operations of the National Highways, LHAs and AV service operators.

6.3 In the model outlined in 6.2 the ORR will regulate the highways managed by National Highways and LHAs, this role will ensure checks and balances are in place to avoid situations like those encountered with the deployment of Smart Motorways.34 The ORR would be responsible for regulating novel L2 automation (e.g. platooning)8 on the Strategic Road Network, licencing L4/L5 service operators and regulating the LHAs where L4/L5 AVs operate.

6.4 There are significant safety concerns about the deployment of L3 AVs (see 5.3 and 5.4). To keep our lead in AV development, the government needs to focus on L2 and adopt EU GSR and PSR.8

6.5 Whilst, the initial deployment of L4/L5 in urban areas would be low, the government could accelerate this by running pathfinders in a cities or urban boroughs. In an urban environment, if a service operator model was widely adopted by the residents, it could drastically reduce the need for car ownership, radically change the cityscape (with fewer vehicles parked on-street) and deliver benefits in decarbonisation. Adoption by residents could be accelerated using levers like: charging for on-street parking,36 funding the scrappage of non-EV vehicles and replacing vehicle excise duty with road pricing based on distance travelled, emissions and road safety.37 Innovations in passenger transport and automated deliveries using smaller vehicles could also reduce the number of larger vehicles on urban roads and encourage active travel.

6.6 For safety reasons the focus for L4/L5 needs to be on urban roads at low speeds, with conventional vehicles operating on rural roads and motorways. The government will need to develop a roadmap for AVs, identifying transition states to achieve an end state of L5. The L2 conventional owner model (with increasing levels of assisted technology) is one transition state and the L4/L5 service operator model (gradually being deployed in urban areas) is another transition state. How these models will interact with one another and how the deployment of L4/5 on motorways and rural roads will happen is not yet entirely clear.

Conclusion

7.1 The safety issues encountered with deployment of SMART Motorways must not be repeated.34 The safety standards and safety assurance scheme for AVs needs to be regulated by the DVSA and our highways and the licencing of service operators needs to be regulated by the ORR.

7.2 AVs are on the verge of being ready for deployment.1 Due to the significant safety concerns about L3, the focus needs to be on increasing levels of L2 assisted technology and extensively proving L4/L5 at low speeds on urban roads using a service operator model.8,11,13,14

7.3 The opportunities L4/L5 offers are very exciting, decarbonisation will make cities more pleasant and reduced car ownership and on-street parking will release urban spaces for other uses.