Related to the question of safety is concern about the legal questions which arise from changed understandings of driving and control of the vehicle. One of the key concerns expressed by Australians surveyed on their thoughts about driverless vehicles is the uncertainty regarding legal responsibility and insurance in case of an accident. This chapter discusses some of the issues and concerns regarding legal and insurance questions, as well as some of the attempts to address those questions.
The Australian Driverless Vehicle Initiative (ADVI) 2016 survey of Australians’ opinions about autonomous vehicles found that the most common concern was ‘being legally and financially responsible if the car is involved in an accident or makes mistakes’, with 92% of respondents identifying this as a concern.
The University of Michigan study mentioned above also identified this area as a barrier to people’s acceptance of driverless vehicles, as did a 2016 survey by the Royal Automobile Club of Western Australia. iMOVE CRC described current regulation as ‘inadequate in the event of a driverless vehicle crash’.
Where does responsibility lie in case of an accident?
The central question revolves around where legal responsibility lies should a vehicle operating with at least some automation be involved in an accident. Currently, even the most advanced vehicles have an identifiable driver who is responsible for control of the vehicle. However, ambiguity arises when the vehicle is not being – and in higher-level vehicles, cannot be – controlled by a person within it.
Thus, the question arises: who is responsible for that vehicle in the case of an accident? The owner, the car’s manufacturer or someone else?
The National Transport Commission (NTC), as part of its work in identifying regulatory barriers to more automated vehicles, noted that:
… in November 2016 Australian transport ministers agreed to reaffirm the existing policy position that human driver remains in full legal control of a vehicle that is partially or conditionally automated, unless or until a new position is developed and agreed. This provided immediate clarity on Australia's interpretation level 3 vehicles while allowing the possibility to update this position in line with technology developments.
In late 2017, the NTC will begin work to develop ‘legislative reform options to clarify the application of current driver and driving laws to automated vehicles, and to establish legal obligations for automated driving system entities’. The NTC, in November 2017, will present to the Transport and Infrastructure Council draft guidelines on the definition of ‘control’ in automated vehicles.
Manufacturer Toyota emphasised that Australian standards underpinning the control of vehicles should be aligned with international ones:
The challenge for legislators will be to ensure that the legislated definitions cater for the different levels of automation in addition to conventional vehicles (i.e. ‘mixed fleets’), and that the clarification in ambiguity on the automated vehicle front does not have any unintended consequences when it comes to defining ‘control’ and ‘proper control’ for the non-automated fleet.
The definition of ‘control’ and ‘proper control’ used in legislation should be based on globally harmonised regulations. WP.29 has a proposal for the definitions of automated driving and the general principles for developing a UN-Regulation, and therefore is a comprehensive framework on which to base local legislation as it takes into account levels of automation, system performance requirements and specific use cases for the vehicles.
Car manufacturer Volvo noted its position on legal responsibility in its submission, and called on other manufacturers to adopt a similar one:
Volvo’s public position on liability is very clear. Volvo will accept full liability for damages or injuries whenever one of its cars is in full autonomous mode. Volvo is confident that the redundant and back-up systems contained in our Autopilot and Pilot Assist technologies will bring a Volvo car to a safe stop. This accords with Volvo’s 20-20 vison that no one will be killed or seriously injured in a Volvo car by 2020.
Volvo believes the Australian government should mandate that all manufacturers who sell fully driverless cars in Australia must accept liability for cars involved in accidents that were in full autonomous mode at the time of the accident.
As Maurice Blackburn Lawyers pointed out, there are serious problems with the idea of holding individual operators responsible for the harm caused by an autonomous vehicle:
A strict liability system would essentially be created, as the operators are powerless to avoid the product malfunction. It would also mean that the manufacturer is not responsible for making vehicles safer, or being made responsible for their negligence. More broadly, it embraces the wrong societal incentives and may disincentivise riders from using autonomous vehicles.
However, the submission further argued, it is important that this system does not create too great a burden on manufacturers and thus discourage innovation.
Academics from the Queensland University of Technology (QUT) noted that considerations of the legal liability issues will also have address the question of autonomous vehicle data:
This confusion [regarding responsibility] is likely to be exacerbated if there is also confusion over whether the data recorded by the vehicle can be readily accessed by consumers and/or insurance companies in the event of a crash.
ADVI also noted that, even where there appears to be some clarity, the actual legal and other implications remain untested:
While some vehicle manufacturers have publicly stated their preparedness to accept full liability for a crash involving one of their vehicles when in driverless mode, the legal basis for this warranty remains unclear, as well as mechanisms for assessing and determining an outcome for property and personal injury claims.
While there are unresolved questions regarding responsibility in the case of a fully autonomous vehicle, the ambiguity increases with vehicles in which neither an identifiable human driver with full control of the vehicle nor a completely driverless vehicle system exists. In such cases, the driver could argue that the vehicle’s autonomous systems should have taken over and averted the accident, while the vehicle manufacturer could equally argue that the driver, absent a completely autonomous system, must bear responsibility for the control of the vehicle.
The submission from Maurice Blackburn Lawyers pointed out the dilemma when a partially automated vehicle causes an accident:
Where the ‘operator’ of a partially automated vehicle is required to actively monitor and be ready to intervene where required, liability will prima facie fall on the operator should an accident be caused by their failure to intervene when they are prompted by the automated vehicle. They may also be liable for negligent decisions in relation to engaging the autonomous system, for example if the instruction manual stipulated that autonomous mode should not be used in certain weather or traffic conditions and they did so anyway, or if they chose to ignore the system commands and caused an accident. On the other hand, the operator may be able to prove defectiveness or negligence on the manufacturer’s part by showing that their failure to respond to the car’s request to intervene was due to the manufacturer not including sufficient driver vigilance controls or because the human-machine interface was poorly designed.
Maurice Blackburn also noted that further ambiguity exists in such cases in determining the level of negligence for which the vehicle was responsible:
The standard to which this technology will be held is unclear. For example, in determining issues of negligence, should an automated vehicle be measured by reference to how an ordinary human would act in a given scenario or how a road user programmed to resolutely follow the road rules would have acted?
As noted above, Volvo has publicly stated that it will be liable in cases where a fully automated Volvo vehicle is responsible for an accident. Asked by the Committee about cases of partially automated vehicles, they responded:
I think it stays as it is now. Realistically, if the driver is at fault, the driver is at fault. It runs through insurance. If there is a manufacturing defect with the vehicle, there are policies in place for that. If there is a safety recall, insurance companies would then chase the manufacturer. If the car has not been maintained properly, is it the owner's lack of maintenance of the vehicle or has the person who has maintained it done something negligent during the manufacture? From where we see it, the only change will be in that handover phase when we go to a fully autonomous car; that will then be the issue. Even in that space as well, if the customer or the owner of the vehicle has not maintained it—if the tyres are bald and the car has an accident—there is still a certain responsibility that the driver must maintain the car in accordance with normal use. If it runs out of petrol because the car has not been refuelled and has an accident that way, that is not a manufacturing defect. Although we say we take responsibility, we are not going to take responsibility if it is something you have done negligently or deliberately.
Dr Damith Herath, in a similar vein, noted that the question will change as technology adapts, but also as people get used to the technology and to the idea of changing understanding of vehicle control:
So I think it will evolve in a progressive manner. At this stage probably the best-case scenario we can anticipate is for the human to have the final judgement, for a number of different reasons. One is that we still do not have the full technological capacity for the car to make the judgements that rational humans make. In that instance we want to ensure that the human feels in control. Some of the early studies suggest that having the ability to have the final say in the decision-making process makes it more comfortable for the transition to happen. But as the technology progresses and humans adapt to that, we could ease into giving autonomy to the cars to make the final decision.
The insurance industry in a driverless world
If, as is anticipated, the availability of driverless vehicles results in a substantial reduction in the number of traffic accidents, the vehicle insurance industry will necessarily be changed significantly. This will be multiplied if, as is often argued, the model of car ownership is replaced by more ride-share based options. Approaches to insurance will shift radically if most Australians not only no longer drive a car, but do not own one either.
During the period of the mixed fleet – a range of vehicles sharing Australian roads – ADVI argued that there will be ‘ongoing need for personal insurance, including compulsory third party (CTP) personal injury insurance, property damage insurance and product liability insurance’.
However, ADVI noted that ‘what an insurance industry looks like and how it evolves will inevitably change in line with driverless technology’.
The submission from the Swinburne University of Technology also addressed this question, noting that international studies have suggested that increasingly automated vehicles – particularly passenger vehicles – will alter the insurance industry in a substantial way:
If automation will make crashes far less likely, then why buy vehicle insurance? Some studies have speculated that premiums could be reduced by 75%, especially if drivers are no longer required to get coverage, and liability is shifted from drivers to product liability, manufacturers and technology companies. Under this scenario, insurers might move away from covering private customers from risk tied to 'human error' to covering manufacturers and mobility providers against technical failure.
A Rand Corporation report also predicts that drivers might end up covering themselves with health insurance instead of vehicle insurance. According to a similar report by KPMG, the insurance industry could contract by as much as 60% by 2040 as accident damage payouts and premiums fall.
The Insurance Australia Group (IAG) argued that, while the models of car ownership and insurance coverage may change as a result of the introduction of autonomous vehicles, insurance will still be a necessary part of the landscape:
We believe it is of critical importance, particularly because we cannot say with certainty that there will be no death or injury on the road with autonomous vehicles. We hope that there will be a safety aspect and we actually anticipate and believe there will be a safety benefit. We do not know exactly the full extent of that. And we do not know what will happen in terms of new risks that might emerge—as you previously mentioned—around cybersecurity risk or if there was a mass of systemic failure of the broader system on the road. Certainly, compulsory third-party insurance is a key one going forward. It has challenges, obviously, because it is regulated at a state-based level. So it is different in every state.
The Insurance Commission of Western Australia argued that the shift in responsibility for accidents from human drivers to autonomous systems should result in a corresponding shift from Compulsory Third Party insurance to insurance coverage of vehicle and software manufacturers:
Manufacturers and suppliers should have insurance that is appropriate and sufficiently broad to cover a number of risk areas, including public liability, product liability and cyber risk. People injured in the event that automated vehicle technology fails should be easily able to claim on that insurance. This cover should extend to all people injured in any crash including passengers, cyclists and pedestrians.
Professor Robert Sparrow and Dr Mark Howard of Monash University made the same point, arguing that ‘eventually insurance to protect against the cost of motor vehicle accidents will be primarily purchased by manufacturers rather than individual owners, thus radically reshaping the current motor vehicle insurance industry’.
DIRD noted that, as part of the insurance industry’s response to the changing environment of vehicle use, ownership and safety, some firms have been involved in trials of driverless vehicles or begun releasing new insurance types in light of technological advance.
The Committee recognises that the availability of driverless vehicles, especially of passenger vehicles, may necessitate a substantial shift in the way vehicles are insured and in the current understanding of legal liability. This topic should be discussed as part of the engagement with the public, as is recommended in chapter eight of this report.
The Committee notes the ongoing work of the National Transport Commission in identifying regulatory and other barriers, including key issues such as the definition of ‘control’, to the introduction of automated vehicles in Australia, and considers this work to be a priority. The Committee is pleased to see that the NTC currently has a range of projects relating to automated vehicles and looks forward to seeing the results of those.