Ensuring the safety of driverless cars in Australia
First published 4 August 2017
Driverless cars should have a game changing impact on the road toll. But it is also inevitable that they will fail from time to time, and fatalities will occur. In managing the transition to increasingly automated vehicles, our regulators need to keep their eyes on the end game.
1,290 people died on Australian roads last year. Globally, the figure is more than 1.2 million, making road traffic injuries a leading cause of death globally. 94% of crashes in the USA can be tied to human choice or error. Automated vehicle technology that corrects for human mistakes, or that takes over responsibility for driving, could dramatically reduce the number of crashes tied to human choice and behaviour. Highly automated vehicles also have a learning advantage over humans. Human drivers make the same mistakes as those who have driven before them, whereas computer driven vehicles can benefit from the data and experience drawn from thousands of other vehicles on the road.
But a computer can also make poor choices, especially if confronted with a situation that it hasn’t been programmed to deal with, or that its sensors can’t see. Roads are complex environments that constantly throw up unique situations that drivers need to respond to. Despite the learning advantage that computers have over humans, it can be expected that the number of crashes caused by a computer failing to respond appropriately will rise, as automated driving systems increasing assume more driving responsibility.
This creates a challenge for government. The community expects government to regulate motor vehicles and roads in a manner that reduces safety risks to an acceptable level. This doesn’t mean governments should regulate to make or roads as safe as possible, as we also have other, competing objectives in connection with the use of our roads, such as acceptable travel times. If government over-regulates automated driving systems, it could delay the achievement of the road safety and other benefits that these technologies offer, including improved mobility for people who can’t drive or afford to own a car, and improved productivity for people who will be able to more productively use time previously spent driving.
Current regulatory arrangements
The regulation of motor vehicles and road safety is presently split between the Commonwealth and state and territory governments. The Commonwealth regulates vehicles standards for new and imported vehicles, as well as nationwide recalls for non-compliances and safety-related motor vehicle defects. The states and territories regulate the licensing of (human) drivers, the registration and in-service roadworthiness of vehicles, the use of vehicles on public roads (via road rules) and compulsory third party insurance. State, territory and local governments also have responsibility for most road infrastructure.
Is more regulation required?
Many people would argue that Australia is already over-regulated, and that we would benefit from less regulation, not more. Do we really need to specifically regulate driverless vehicles?
The Australian Consumer law imposes a number of statutory guarantees upon suppliers of motor vehicles, including that the goods will be safe, free from defects and fit for purpose. Additional obligations may be owed by the supplier, depending on the terms of the contract for the sale of the vehicle. However, these statutory guarantees and contractual obligations are only owed to the purchaser of the motor vehicle, and not to passengers or other road users. Additionally, existing negligence laws will impose an obligation on the manufacturer and/or supplier of the vehicle to safeguard vehicle users against foreseeable risks of injury arising from the use of the vehicle as intended. This duty of care will also be owed to others whom the manufacturer or supplier ought reasonably to have foreseen may be harmed by the vehicle, such as passengers and other road users.
Given these existing safeguards, some will argue that there is no need for governments to further regulate driverless vehicles. However, an important layer in the existing motor vehicle safety assurance system is the requirement that a motor vehicle can only be driven on a public road by a licensed driver that has demonstrated his or her competence to do so. This layer won’t apply in the case of vehicles that don’t require a licensed human driver to watch the road and be ready to intervene if necessary. The absence of this safety assurance layer, coupled with the significant risks to human safety that driverless vehicle pose, the safety-critical role that the automated driving system will have, and the possibility that some manufacturers may be tempted to offer such vehicles for sale before all foreseeable risks have been suitably minimised, provides a compelling case for government to specifically regulate for the safety of automated driving systems.
What form should this regulation take?
There are various options that have been canvassed in a recent discussion paper by the National Transport Commission. In my view, the approach that would work best in Australia is as follows:
1. Government should impose a statutory safety duty on suppliers of automated driving systems that don’t require a licensed human driver to watch the road and be ready to intervene if necessary. The statutory safety duty would require suppliers of these automated driving systems, or vehicles containing them, to ensure, so far as is practicable, that the vehicle is safe if used for its intended purpose. Such a duty would mirror the duty of care owed in negligence by the supplier, but would be actionable by government or any other person that suffers loss as a result of a breach of the duty.
2. Suppliers should be required to certify to government that they have discharged this duty, and provide satisfactory evidence that they have done so (including a comprehensive safety assessment), before they can supply the automated driving system in Australia, or any significant modification or update to it (including software updates).
3. Suppliers should also be required to demonstrate that they have the financial capacity or product liability insurance arrangements to meet reasonable potential liabilities arising out of any defects in the vehicle’s automated driving system. It will be important that those who suffer injury or property damage as a result of a failure by the supplier to discharge its statutory safety duty (or duty of care in tort) are able to recover their loss from the supplier.
4. The supplier’s certification and evidence could be assessed by the Vehicle Safety Standards Branch within the Commonwealth Department of Infrastructure and Regional Development, at the same time as the VSSB assesses whether the vehicle complies with all required Australian Design Rules under the Motor Vehicle Standards Act 1989 (Cth). The VSSB could also assess whether the duty has been satisfactorily discharged in connection with software updates or other significant modifications.
5. Finally, like existing motor vehicles, automated vehicles should be registered (or exempted) by a state or territory road authority before they can be used on public roads. Mutual recognition of registration should continue as per current arrangements.
This approach would be technology agnostic. It would enable risk management processes and tests to quickly evolve as industry standards and best practice emerge, and minimise the risk that government mandated testing could stifle safety related innovation. It would enable manufacturers to complete most testing at their preferred location, and limit testing that needs to be performed (or re-performed) in Australia to that which is necessary to demonstrate that differences in traffic rules and Australian road infrastructure have been addressed.
This suggested approach would align nicely the current division of regulatory responsibilities as between states and territories and the Commonwealth. It would allow safety assessments to be conducted by a single government agency, and minimise the prospect of inconsistent requirements being imposed by each state and territory road authority. It would also preserve for each state and territory the final say that it presently enjoys over vehicles that can be used on public roads within its jurisdiction – something that states and territories will be reluctant to give up.
Most importantly, the above approach would enable Australians to promptly access the safety and other benefits that increasingly automated vehicles offer, without exposing us to unnecessary safety risks.