Written evidence submitted by Reed Mobility (SDV0008)

Introduction to Reed Mobility

Reed Mobility is an independent research consultancy founded and run by Prof. Nick Reed. Nick was formerly Academy Director at TRL (the UK’s Transport Research Laboratory), responsible for research and innovation across the business. With leadership of a portfolio of connected and automated vehicle projects exceeding £50m, Bosch (the world’s largest automotive supplier) recruited him to lead global R&D activities in urban mobility. Nick left Bosch in 2019 to set up Reed Mobility, working in the public, private and academic sectors around mobility, innovation and strategy. Customers include the European Commission, the Department for Transport, BSI (British Standards Institution), Transport for London, RSSB (Rail Safety and Standards Board) and ARRB (Australian Road Research Board) with project work focused primarily on the safe, efficient and ethical implementation of automated vehicle technologies. Alongside his leadership of Reed Mobility, Nick was appointed as National Highways’ first ever independent Chief Road Safety Adviser and holds a visiting professorship at the University of Surrey, working with the 5G Innovation Centre.

Reason for submitting evidence

I have been involved in some of the earliest projects that demonstrated automated vehicles (AVs) on UK roads and have led AV trials involving driving simulators, test tracks and on public roads. I have seen the hype that has been associated with the promise of such vehicles and consolidation in the industry as some companies failed to deliver on over-ambitious targets. I have long held the belief that the challenges to implementation of AVs can be grouped into three interlinked categories. First, technological – developing the required sensing, computing and actuation systems to enable automated driving. Second, regulatory – putting in place the necessary guard rails to ensure that AVs are operated safely and appropriately. And third, societal – making sure that AVs operate in the way users expect, that other road users can instinctively intuit their behaviour and that they deliver equitable mobility benefits to the widest spectrum of users. I believe my involvement in the AV sector over more than a decade and my perspective as an independent consultant with a background in psychology gives me insights that will be of use to the inquiry. This submission is solely on behalf of Reed Mobility and does not represent the views of any other organisations with which I am associated.

Submission

1. I strongly believe that AVs will have a significant, positive effect on transport and that the UK is well positioned to exploit such benefits. However, it has become apparent that the anticipated benefits are taking longer to achieve and may be less favourable than was expected five years ago.

2. Although, there were some notable experimental trials of self-driving vehicles (including in the UK) in the 20th century, the history of AVs stems from challenges posed by the US military from 2004 to 2007, in which research teams built self-driving vehicles to tackle a variety of tasks with the ultimate aim of producing vehicles that would reduce risk exposure for frontline personnel. The leaders of those research teams and the momentum created around the potential for self-driving vehicles was the spark for the AV industry, fuelled by major investments from venture capital and the automotive industry and the bravado associated with innovation in the technology sector. This led to expectations of rapid and widespread adoption of AVs.

3. However, the driving tasks and regulatory environments for the military trials were markedly different to those that apply when driving on public roads. Although several companies (e.g. Waymo, Cruise) have achieved millions of miles of automated driving on public roads, the majority of these have been accumulated as trials – where the vehicle is either accompanied by a safety operator or not operated as part of a publicly accessible service. Only recently have such commercial services become a reality and only in limited circumstances. This reflects the technological (sensors and software needed to negotiate the infinite variety of the public roads), regulatory (revising and/or replacing long established traffic laws, vehicle/driver certification processes and insurance protocols) and societal challenges (preparing people and businesses for the use of and encounters with AVs) noted above.

4. Tesla is an outlier in the AV industry. Since 2015, its chief executive has been promising the imminent arrival of self-driving capability on its vehicles but deadlines have slipped repeatedly. Tesla’s approach is based primarily on visible-light camera systems and processing of data and behaviour sourced from the hundreds of thousands of Tesla vehicles driven worldwide. This approach differs from other notable AV developers, which use many fewer vehicles for data collection but equip those vehicles with a more sophisticated sensor payload. Ongoing delays to the arrival of Tesla vehicles with genuine self-driving capability, turmoil within the technical team tasked with its delivery and concerns raised over the unique approach taken by Tesla suggest that its AV promises lack credibility. Perhaps more troubling is that such promises may give drivers of these vehicles (sold with a driver assistance function called ‘Full Self Driving’ as an optional extra) undue confidence in the level of automated driving capability that they are able to achieve, thereby risking driver inattention in moments of need. This flaw has been recognised by investigators (National Transportation Safety Board) in several fatal crashes involving Tesla vehicles in the US. The Law Commissions’ recommendation to restrict manufacturers in naming product features that potentially mislead drivers is to be welcomed.

5. Although lacking the financial support available in Silicon Valley or China, the UK has been admirably progressive in its approach to AVs. As evidence, I would cite:
1. The formation of:
1. a dedicated government policy group (the Centre for Connected and Autonomous Vehicles – CCAV[1]) to help define the UK’s strategy around AVs;
2. an organisation to support the UK’s AV industry ecosystem (Zenzic[2]);
2. The availability of funding (the government has supported a series of collaborative R&D projects including AV trials, cybersecurity, communications protocols and public engagement exercises);
1. These collaborative programmes have energised a wide range of organisations to contribute, including major organisations from the automotive technology, communications, insurance, logistics and energy sectors and encouraged numerous start-ups to grow their innovations.
3. The support for adapting regulations to enable AV testing and deployment including:
1. Code of Practice for automated vehicle trialling[3];
2. BSI 18XX standards series[4];
3. Law Commissions’ review of AV regulations[5];
4. DfT’s CAVPASS (Connected and Automated Vehicles Process for Assuring Safety and Security) programme[6].
6. These initiatives put the UK in a strong position when compared to other countries, a view reinforced by KPMG’s Autonomous Vehicle Readiness Index (2020)[7] – which places the UK in ninth, ahead of Japan (11th), Germany (14th), France (19th), Spain (22nd) and Italy (24th).

7. The steady rather than spectacular progress of AV development suggests a need for pragmatism in government activity. The latest CCAV funding competition recognises this, emphasising a shift to commercially ready AV services[8]. Whilst there is an ecosystem of organisations working on AV technologies, the appetite from industry to use AVs (for the movement of goods or people) is understandably less obvious, given the lack of evidence of economic viability. Successful demonstrator projects are needed to unlock this interest. However, the current immaturity of the technology dictates that these will need to be in highly constrained environments to create the conditions in which AVs can operate successfully in the near term (e.g. AVs using private roads within a factory, port or airport environment or low speed passenger movements on a fixed route with limited interactions with other road users). The challenge then will be to understand which commercial uses cases can profitably apply the technology and how to scale AVs as the systems become more sophisticated and the constraints can be eased.

8. Cars capable of fulfilling the complete driving task for all journeys without a human operator are many years away and are likely to be prohibitively expensive for most private owners. However, private cars are likely to benefit from automated features in the years ahead, enabling drivers to disengage from the driving task for some parts of some journeys. The most likely use cases for such features will be on highways (where the driving task is less complex), in traffic jams (where low speeds mean risks can be more easily managed) and for automated valet parking (where the vehicle and parking infrastructure can work together to manage safe automated operation). The ‘part-time’ nature of the automated driving systems in such use cases creates its own issues. Firstly, the human driver must recognise that the automated features are present and learn how they work. They must also understand what responsibilities they have before, during and after automated driving is in operation and the nature of transitions between human and automated control. Research on these processes is ongoing. However, as these features come to market, vehicle manufacturers will play a crucial role in educating drivers on these topics. This could include training at point-of-sale but also using the in-vehicle display and interfaces to provide education for drivers in how the systems operate (with such training materials only to be viewed when the vehicle is safely parked). There may also be a role for inclusion of the safe use of automation systems in driver training and licence acquisition (at least as an element of the driving theory test).

9. Two key concerns are often raised in relation to AVs. The first is about their effect on employment with fears that the technology will replace human drivers. It will be important to provide reassurance that no-one employed as a driver in the industry today is likely to find their job replaced by an AV. Firstly, the current shortage of professional drivers for passengers and goods vehicles in the UK means that AVs can play a useful role in supporting these sectors in the short-term. Meanwhile, the development and rollout of the technology is likely to be at a pace that means driver retirements and departures from the industry will create gaps that AVs can fill. As the technology matures, it will be important to liaise with professional representative bodies to ensure that drivers’ employment conditions are not unfairly affected by the advent of AVs and that the transition is appropriately managed.

10. The second key concern relates to the availability of safe, convenient, low-cost mobility enabled by AVs. This could threaten recent gains made in uptake of active forms of transportation (walking and cycling) and could increase traffic congestion by encouraging more trips. In response, regulations should ensure that AVs are both objectively and subjectively more safe than equivalent human driven vehicles so that vulnerable road users feel confident sharing the road with AVs. Regulations should also manage the deployment of AVs to ensure that they improve mobility for those currently underserved by existing transport provision but without causing oversupply of vehicles for those already well served.

11. Infrastructure is often cited as a key enabler for AVs. However, I believe the most significant contribution infrastructure providers can make is consistency in the design, implementation and quality of key road features. Not only will this help AV developers by reducing the complexity of the automation task but is also beneficial to human drivers and road users, who will be using this infrastructure for decades to come. Two other infrastructure features are likely to be important. The first feature is good availability of high bandwidth, low latency connectivity. Although AVs must drive safely in the absence of connectivity, this will enable enhanced functions that depend on communications with infrastructure and/or other vehicles. The second feature is improved digital maps with reliable shared information. It will be helpful for AVs if trustworthy, current information about the state of the road network (including e.g. traffic, incidents, weather) were available through a common data exchange. This would improve the safety and effective management of AV operations. National data providers such as Ordnance Survey, the Met Office and National Highways are likely to be important contributors to such a data exchange.

12. Regulation of AV safety will be a critical enabler to the industry. We cannot permit AVs to operate at speed in public environments without data to give confidence that they operate in a manner that is sufficiently safe. For humans, this is essentially achieved in a one-off practical driving test, supported by the assumption (based on millions of years of evolution) that human drivers do not wish to harm themselves or other humans (and understand the potential physical, legal and emotional consequences of doing so). However, we cannot make this assumption for AVs and so will require AV developers and operators to be willing to open their systems to independent scrutiny. They must be able to show that their vehicles promptly and reliably detect relevant hazards, follow the rules of the road in a manner that is compatible with human drivers and respond appropriately and consistently to the situations they encounter. We do not need to know what sensor systems are used or how this information is derived – simply that it exists and is used appropriately. This should therefore enable AV developers to retain their intellectual property on how automated driving is achieved.

13. Where possible, this should be supported by evidence from external sources. For example, CCTV systems used for monitoring roads could supply additional data to confirm the safe operation of AVs. The potentially vast quantities of data from AVs (and supporting data from road infrastructure) will require regulators to develop (or access) significant data science capabilities to be able to manage, manipulate and analyse such data to realise objective AV safety assessment.

14. An important further means of demonstrating AV safety will be simulation. By using computer simulation of AVs to generate risky driving scenarios (and many variants thereof), an AV developer can provide evidence that their vehicles operate safely in the majority of conditions they will encounter in their chosen operating environment (and show how their vehicles will act if conditions change such that they will depart from their expected operating environment, e.g. during severe weather conditions). However, the simulation must have sufficient validity to be of value. Simulations used as evidence of real world safety that do not include the relevant variability of conditions that the vehicle is likely to encounter in the real world (e.g. poor road surface quality, incomplete line markings, glare from the sun, road signs obstructed by parked vehicles etc.) or does not model how hardware and software performance may change over the life of the vehicle or with variable environmental conditions (e.g. fog, glare, dirt etc.) will be of limited value. An important aspect of AV operation will be their interactions with emergency vehicles to ensure that the emergency services can discharge their duties effectively but also to give confidence to the public that AVs are truly compatible with everyday driving situations. While drivers of emergency vehicles are trained to deal with all manner of responses from human drivers, simulation exercises will be useful to build confidence with the emergency services that AVs can respond appropriately when they encounter emergency vehicles.

15. I have emphasised the importance of proving safety because there is likely to be a characteristic difference in tolerance of the public to incidents involving AVs. Even if AVs cause incidents at a lower rate than human drivers, the expectation of infallibility from machines is likely to provoke a response when incidents occur. Furthermore, incidents involving AVs may occur for reasons that are far less intuitive than those caused by human drivers. A fellow human may accept that a driver failed to spot black ice and crashed into oncoming traffic whereas they may be far less understanding of an AV that failed to detect a pedestrian because they happened to be wearing unusual clothing that its systems had not encountered previously. Consequently, it will be vital to have data to show that AVs routinely perform safely and to ensure that when they fail, the causes of incidents can be swiftly identified and rectified and that victims of crashes are adequately compensated. The role of government and regulators is to work with AV developers and the wider automotive ecosystem to establish processes that will generate this data and build confidence in AV operation.

16. In closing, I want to reiterate my positivity towards AVs. I expect them to support safer, cleaner, more efficient, more ethical and more equitable road use. Whilst this is not happening as quickly as was previously anticipated, I still believe that the technology to deliver this brighter future is coming and that the UK is well placed to exploit it. However, AVs are not the only means by which this will be achieved. In determining how the government should support AV technology, due consideration should be given to ensuring that commensurate support is given to active travel and public transport modes, each of which can also contribute significantly (and sooner) to safer, more sustainable travel. Consequently, appropriate priorities for government are, firstly, proportionate support for the growing AV industry; secondly, a focus on commercial use cases that genuinely deliver sustainable, scalable and saleable value and thirdly, ongoing dialogue with the industry and public to ensure the safe and effective use of AV technologies. The competition to create AVs is often characterised as a ‘race’ but I disagree with this position. Mobility is constantly evolving and AVs are another technology in this progression. We should not be racing to an abstract finishing line but continuously evaluating how AV technologies can safely and effectively deliver benefits to our transportation system and shaping regulation to ensure those benefits are equitably distributed. Delivering this may not ‘win’ the race but we will sustainably and positively enhance mobility and thereby improve prosperity for UK citizens and businesses – a far more satisfying victory.

August 2022

Endnotes