Within an astonishingly short period, autonomous vehicles (driverless cars) have become technologically viable. A fully autonomous transport ecosystem is still very far away, but over the next five years, deployment of some impressive levels of semi-autonomous functionality is highly realistic. Regulators and policy-makers believe they will help achieve to transport goals of reducing car ownership and congestion, and increasing road safety and mobility for the currently immobile, and are actively consulting the industry to facilitate their arrival.
Regulators in the UK have taken a pragmatic approach to the new technology, advocating an evolving program of measures as the industry delivers increasing functionality. However, international regulatory alignment on a range of issues will still be necessary for the industry to be able to develop multi-jurisdictional business plans and invest on the scale necessary to bring the technologies to full fruition. Regulators recognise that they must ensure the insurance and liability rules enable the business and operating models to evolve as their approach will determine the design of transport networks, driving patterns and car ownership.
Judges in the UK are well used to complex cases that require distinguishing between a failure of technology and failure to operate technology correctly
Automation will not entirely eliminate accidents, but it is compelling to transport policy-makers because of the potential for software to help ensure much more compliance with rules and protocols which will greatly improve road safety. For example, where there are no mandated speed limits, a driver should only travel at a speed which is reasonable in the circumstances. Software can help ensure that the vehicle does actually stop within the limits of its braking ability and its onboard camera’s “vision”, adjusted for the weather and the state of the road. In a residential area or near a school or hospital, the software will ensure vehicle speed is reduced sufficiently to avoid collision with vulnerable pedestrians and not “forget” to anticipate that pedestrians may try – and are entitled – to cross other than at an official pedestrian crossing.
Rules of the road
Revised rules will also provide courts with a clearer starting point in determining whether a legal standard has been breached and in this way, fault or liability can be established for other legal and regulatory frameworks.
The Highway Code sets out the basic rules for British road users and most apply equally to semi-autonomous driving. Amending the Code to establish new common rules and best practices for driverless cars should not be difficult to do in proportionate response to the industry delivering increasing functionality. At this early stage, a key question in defining autonomous vehicle operation rules will be the practical extent to which a driver can supervise the autonomous functionality and, effectively, assume control.
As the technology is evolving quickly, it may not be clear whether instructions have been followed and whether there is sufficient evidence of defectiveness versus any negligence on the part of the driver operator. However, judges in the UK are well used to complex cases that require distinguishing between a failure of technology and failure to operate technology correctly, and in any autonomous vehicle context they will assess the facts with the assistance of expert witness evidence.
Software ethics and the trolley problem
Designing software to ensure a vehicle obeys the rules could create a potential ethical issue and a choice between two terrible options. Faced with a child running onto the road in front of it, the computing power of the car could calculate it has a binary choice between likely killing the child who wrongly ran into the road or mounting the pavement and likely killing a group of children. Known as the Trolley Problem (taking one life vs. taking many lives), the algorithmic version of this dilemma should be avoided as much as possible; regulators should mandate that programs simply follow the Highway Code and ignore such subjective choices.
The mass market adoption of driverless cars is necessarily linked to the availability and cost of insurance. Insurers will need to price new insurance products with premiums initially calculated with little empirical data. Given that the product liability component will be recovered from manufacturers, they may need to co-develop new products with insurers, and share risk, in order to help make insurance available and encourage greater adoption of their vehicles.
However, other than who has to hold which types of insurance and the amounts needed, there should be no change to the role of insurance. If an accident was a result of software error, the injured party should be able to claim from the driver (or their insurer) who should seek to recover from the manufacturer (or their insurer). In reality, any other approach would be impractical. It will be important for regulation to ensure a framework where any of the parties that suffer personal injury or damage to property can bring a claim and be in no worse position than they are in now.
An advanced autonomous environment could accelerate the automatic creation, processing and enforcement of obligations and liabilities. These could handle fees for parking, toll road use, secondary carriage and even cleaning. A decentralised system of computers that irrevocably confirms micro-obligations based on predefined protocols is at the heart of the emerging area known as smart contracts, built on so-called distributed ledgers. It is very early to make predictions, but the size and importance of congestion and parking fees to the revenues of local governments make it a potential candidate for an early use case.
Mission difficult, not mission impossible
Our transport regulators deserve credit for being proactive in thinking early and deeply as to how to pragmatically clear the path toward achieving vital transport and environmental goals. In terms of the legal and regulatory environment, in summary, we likely face a driverless evolution rather than revolution. Some problems can be solved by interpretation, and some will require new law and secondary guidance, but all can be solved.
This is a shorter version of an article due to appear in Computer and Telecommunications Law Review (2017) 23 CTLR Issue 1, coming soon on Westlaw.