One of the final bills passed by the outgoing Sunak government was the Automated Vehicles Act (2024), designed to make provisions for the possible rollout of autonomous vehicles in the future. If the new Starmer government is to continue this work, then lessons need to be learnt from the US, where the deployment of autonomous vehicle passenger services (AVPS) – often referred to as ‘robotaxis’ – has been on a rocky path since the first public tests of such vehicles in 2020. Here, Dr Sam Hind offers recommendations for how a new regulatory framework for autonomous vehicles might work in the UK.

• The ‘move fast and break things’ ethos of big US tech firms leading the development of robotaxis brings wider social risks.
• An incrementalist approach to the development and deployment of autonomous vehicles is favoured by municipal transport authorities in the US, where tests have already taken place.
• Regulatory oversight of autonomous vehicle testing and deployment should be holistic and open the ‘black-box’ of autonomous vehicle operation to manage social risks.
• Supporting a fledging autonomous vehicle industry in the UK should involve fostering a ‘homegrown’ tech industry, less reliant on US firms and expertise.

 Permission to ride

On October 24 2023, the California Department of Motor Vehicles (DMV) – the body responsible for regulating vehicle licenses in the state – suspended the operations of a robotaxi firm, Cruise, that had been running an ‘autonomous’ taxi service in the city of San Francisco for two and a half months. Cruise’s suspension was understood to be the result of its failure to cooperate with an investigation into a road traffic incident one of their vehicles was involved in. According to reports, a pedestrian who had been hit by an ordinary vehicle (driven by a human) had subsequently ended up underneath a Cruise robotaxi (being driven ‘autonomously’).

Many such stories involving autonomous vehicles have hit the headlines in recent years, most notably the death of Elaine Herzberg, killed by an autonomous Uber vehicle being tested in Tempe, Arizona. The incident that led to the suspension of Cruise’s robotaxi permit in October last year, however, was significant because no vehicle operator was behind the wheel at the time. Cruise and Google/Alphabet’s autonomous vehicle division, Waymo, had been granted permission to operate across the entirety of San Francisco without geographical (previously downtown only) or time (i.e. outside of rush hour) restrictions in July 2023.

Permission lay with a different body to the California DMV – the California Public Utilities Commission (CPUC), responsible for issuing public utilities licenses, such as for those wishing to run taxi services like Cruise and Waymo had desired. For this purpose, CPUC had set-up a tailormade permissions process for prospective robotaxi firms seeking to pick-up actual passengers. These were dubbed their ‘Drivered Pilot’ and ‘Driverless Deployment’ Autonomous Vehicle Passenger Service Programs.

Lessons to learn?

In August 2024, the Parliamentary Office of Science and Technology (POST) published a research brief explaining what automated vehicles are, how they work, and the policy considerations for any future development of them in the UK. In my contribution to the brief, based on my research findings, I suggested that the UK should learn from the bumpy rollout of robotaxis in the USA – with a specific focus on how this has happened in the state of California.

US big tech firms and similar such companies operating within the autonomous driving/ robotaxi space still appear to be driven by a ‘move fast and break things’ ethos – the phrase allegedly coined by Meta’s Mark Zuckerberg – whereby products should be put in consumers hands as fast as possible, without regard for the merit or rationale for typical systems of governance. While this ethos has allowed operators to deploy robotaxis (in one form or another) on public streets relatively quickly, it has created a series of subsequent risks and concerns for wider public bodies, public-facing workers (such as bus drivers or those in road maintenance), and users of public transportation services.

In response to Cruise and Waymo’s request to expand their robotaxi operations, key Californian public transportation authorities, responsible for millions of transportation users across the state, banded together to propose a more responsible approach to the deployment of robotaxis. They called this the principle of ‘incrementalism’. In short, that any prospective robotaxi operator seeking to gain permission for running services in San Francisco or anywhere else across California should need to pass specific, standardised performance milestones before further approval or expansion might be granted. Though this may appear an obvious suggestion to outsiders, this was not the permission process undertaken by these companies, who had refused to hand over certain kinds of operational data of their robotaxi vehicles, claiming commercial confidentiality rules.

Policy recommendations

Any regulatory environment for the development and deployment of autonomous vehicles and/or robotaxis in the UK should require firms to submit relevant operational data such as the frequency of unplanned stops and so-called ‘vehicle retrieval events’ (VRE) – key indicators of the performance and reliability of underlying autonomous vehicle, AI, and machine vision systems. Any assumed regulator must be able to open the metaphorical ‘black-box’ of autonomous vehicle operations, to help them make decisions about whether to approve or refuse subsequent operational permissions. In addition, this work would benefit from being housed under one regulatory roof, to avoid a dance between respective regulators of motor vehicles and taxi licensing like in the California case. One such option would be assign responsibility to the recently-announced Regulatory Innovation Office, tasked with speeding up public access to new technologies. A close partnership with the Department of Transport on such matters would be wise.

If the UK is committed to supporting a fledging autonomous vehicle industry in the UK, it should not be reliant upon the US tech sector to do so. As the semiconductor chip crisis (2020-23) and the subsequent EU Chips Act (2023) and US CHIPS and Science Act (2022) illustrate, global technological supply-chains are increasingly brittle and precarious, subject to monopoly control over key materials (like lithium), technologies (such as photolithography machines), and assembly plants (so-called ‘chipfabs’). If the UK is serious about investing in the future of a technology-intensive industry like autonomous vehicles, it must act similarly to attract investment and talent to the UK. Failing to do so might mean a homegrown autonomous vehicle industry runs out of road before the journey has even begun.