Who needs driverless cars? Advanced driver assistance systems, safety and the law

The World Health Organisation Global Status Report on Road Safety 2023 opens with these sobering words:
‘By the time you have read this page, at least five people will have died in road traffic crashes. Road crashes are the leading killer of children and youth, and they typically strike during our most productive years, causing huge health, social and economic harm throughout society”
Road crashes also cause a multitude of serious injuries, with annual rates of hospitalised injury more than 30 times higher than fatalities. The UN in its Political Declaration of the High-Level Meeting on Improving Global Road Safety in July 2022 ‘[recognized] that road safety is an urgent development priority, a major public health problem, and a social equity issue.’ Even small improvements in road and vehicle safety can have enormous positive impacts.
‘The social cost of road crashes’ in Australia alone has been estimated at least $27 billion annually, and Australian statistics are no less confronting:
“Land transport accidents were the most common cause of death among children aged 1–14 (13%). Suicide was the leading cause of death among people aged 15–24 (36%), followed by land transport accidents (19%).”
Land transport accidents rank 3rd highest cause of death in Australia for those aged 25-44, after suicide and accidental poisoning. There were 1,194 road crash deaths in Australia in 2022 (the most recent year for which data is available), increasing 5.8 % from 2021 after a decade where the number of fatalities remained largely flat.
Human error is estimated to have caused over 90% of road crashes, although Zipper has argued that this ‘deflects attention from … car designs and allows traffic engineers to escape scrutiny for dangerous street designs.’
In 2021 in Australia (the most recent year for which this data is available), more single vehicle crashes (726) caused death than multiple vehicle crashes (468), with ‘single vehicle run-off road’ the most common crash sub-type by far since 2017, and the cause of nearly 50% of deaths described as ‘non-collision’.
Fully autonomous vehicles promise safety benefits that may remove much of the risk that human drivers will err, but that to date, remain largely on the horizon (perhaps even further than appeared several years ago). Some market analysis suggests ‘[c]onsumers still prefer automated driving features where the driver maintains more control’. As fleets are increasingly electrified, conflicting reports are emerging about the crash rates of electric cars , although the concerns have been summarised as
“issues of crashworthiness (i.e. how the structural and weight differences of EVs compared with conventional internal combustion vehicles affect vehicle collision behaviour), postimpact vehicle safety (i.e. the challenges associated with high-voltage circuits, batteries or hydrogen fuel-cells following a vehicle collision) and low noise emission (i.e. the impact on vulnerable road users who rely on auditory cues to respond to approaching vehicles)”
rather than the behaviour of the drivers. As the number of electric vehicles in use grows, it will be important to monitor the data closely.
In the meantime, however, Jessica Truong, Secretary General of Towards Zero Foundation argues that ‘80-90% of fatalities and serious injuries can be prevented by currently available measures’.
Existing vehicle safety features include both passive crash protection features and active crash avoidance features. The latter, usually referred to as advanced driver assistance systems [ADAS] ‘assist drivers with warnings or automatic braking to help avoid or mitigate accidents’, and ‘are focused at addressing a particular task of driving, thereby, reducing driver errors and enhancing safety.’ Although often named differently in different makes and models of vehicles, they include electronic stability control, automatic emergency braking, automatic emergency braking with pedestrian detection, rear automatic braking, rearview cameras, rear cross-traffic alert, blind spot monitoring, adaptive cruise control, following distance warning, lane keep assist, lane departure warning, self-parking, adaptive headlights, fatigue warning, and traffic-jam assist.
ADAS technologies may also influence driver behaviour ‘by making drivers less aggressive and harmonizing the driving environment’ (so presumably safer?) and at times override the driver’s actions or control of the vehicle, although to date, admittedly ‘drivers’ behavior when driving vehicles with ADAS compared to vehicles without ADAS has been meagerly explored’. There’s an important caveat though:
“Most tests on ADAS were performed at safer conditions and with better trained drivers than what normal drivers encounter daily … A few vehicle owners also complained of some advanced features not working properly at nighttime and during ...
None the less, the evidence is growing. As data showing the benefits mounts, understandings about how risks of harm can be reduced and what measures can be taken to increase safety change and develop.
The predecessors of Electronic Stability Control [ESC] features began to be introduced in the 1980s, with second generation systems introduced in the 1990s.
2007 US data showed ESC reduced fatal single-vehicle run-off road crashes for cars by 36% and for light trucks and vans by 70%. It was even more effective in preventing single-vehicle first event rollovers: reductions of 70% of cars and 88% for light trucks & vans. Overall, for all crash involvement, the data was still impressive: reductions of 14% in cars and 28% in lights trucks 7 vans.
In 2009, then Federal Minister for Infrastructure, Transport, Regional Development and Local Government Anthony Albanese announced that ESC would be mandatory on all new passenger vehicles from 2011.
In 2015, an international research team identified that there was a “38 percent overall reduction in rear-end crashes for vehicles fitted with [low speed autonomous emergency braking] AEB compared to a comparison sample of similar vehicles.’
2023 data from the Highway Loss Data Institute and Insurance Institute for Highways Safety shows significant reduction in crashes from a variety of these ADAS systems. As Greenwood, Lenneman and Baldwin (2022) note in,
“Vehicles equipped with forward collision warnings experienced both fewer and less severe crashes (Baldwin et al., 2014, Souders et al., 2020) but vehicles equipped with both forward collision warnings and automatic emergency braking systems (AEB) showed the greatest benefit (Spicer et al., 2018). In passenger vehicles, forward collision warnings alone reduced rear-end struck crash rates by 27% while forward collision warnings with AEB reduced rear-end struck crash rates by 50% (Cicchino, 2017). AEB also reduced heavy truck crashes by more than 40% (Yue, Abdel-Aty, Wu, Wang, 2018). Cross-traffic alerts (Cicchino, 2019) and blind spot monitoring (Cicchino, 2018) also reduced crashes. This evidence shows a significant public health benefit can be realized by ADAS technologies.”
Similar outcomes in Australia equate to saving hundreds of lives annually.
Data demonstrating the benefits of ESC to road safety has now been known for well over a decade. Data demonstrating the benefits of newer forms of ADAS is more recent, but clearly emerging. How should the law take it into account of this evidence, if at all?
Since the earliest days of motor vehicles, the law has had a crucial role in addressing road safety, and in compensating those to whom these new modes of transport caused harm. The law has been used to leverage driver behaviour to improve safety: imposing limits on speed; developing road and vehicle design rules; requiring drivers to be licensed and vehicles to be registered and insured; mandating the use of seatbelts and child restraints; restricting the use of vehicle by type or time of day for learner and provisionally licensed drivers; and prohibiting driving while under the influence of alcohol or drugs.
If the evidence now shows that significant safety gains result from vehicles equipped with ADAS, (even if ADAS may not be 100% effective 100% of the time), then (in a similar way to how the law addresses seatbelts and motorcycle helmets) should ADAS use be mandated, or should its lack of use be reflected in, for example, findings of contributory negligence?
A key difference here may be the complexity of operation of these systems. Even a tool like AEB has ‘limitations known to the manufacturers’ :
For instance, they do not brake effectively when they experience sudden “cut-ins,” “cut-outs,” and “crossing path” scenarios (i.e., an errant vehicle crosses suddenly at right angles) (NHTSA, 2017). A recent study on a closed course found that “cut-out” scenarios in a Level 2 vehicle (as classified by the SAE, 2018) led to crashes into a dummy vehicle on 66% of trials (AAA, 2020). Pedestrian detection systems vary between 30 and 90% in their effectiveness, depending on the situation (Combs, Sandt, Clamann, & McDonald, 2019).
This poses interesting questions indeed in relation to liability for harm for motor vehicle crashes. Can (and should) the law be used to leverage these technologies to achieve even greater safety gains? Does (and should) the data (even if still nascent) change tortious notions of how risk is assessed, what can reasonably be foreseen, what a reasonable person ought to know, and how they would respond to foreseeable risks?
Legal responses such as mandating the fitting of seatbelts in vehicles and their usage by drivers and passengers, recognising their value by findings of contributory negligence when they were not worn and the introduction of common law then mandatory statutory reductions in damages, have all played a very important role in changing attitudes and behaviour and as a consequence reducing injury and deaths. However, while seatbelts may provide a useful analogue, there are very significant differences.
Seatbelts look and operate in the same way across a multitude of vehicles, and are called the same thing. Safety benefits are both easily apparent and easily explainable. No specific training is required in their use, and no technical knowledge is required. They can be used by drivers and passengers whether adults or children.
Differences also arise when considering past legal responses to speed, and driving while under the influence of alcohol or drugs. Drivers, and often passengers too, can easily see whether their vehicle is complying with applicable speed limits. The obligations on drivers in relation to consumption of drugs and alcohol are now well known in the community, and in the vast majority of cases, drivers and passengers will be aware that a driver or passenger may be intoxicated.
By contrast, ADAS in a vehicle may include a variety of different technologies, named differently across different makes and models, and which operate in different ways. Older vehicles may have no or few ADAS, and whether or not ADAS are fitted to a vehicle is unlikely to be clearly visible without searching for it. Even where drivers are aware that the vehicle is fitted with ADAS, they are unlikely to have had any training (professional or otherwise) in how to use it most effectively.
What drivers know about ADAS, how it works and how to use it most effectively varies significantly: ‘ADAS place powerful, yet imperfect, automation in the hands of drivers who hold both misconceptions and reservations about ADAS’ and ‘not all drivers understand the application or limitations of advanced features in the vehicles’.
‘Misuse and abuse of ADAS is seen in recent NTSB-investigated fatal crashes in ADAS-equipped vehicles in which drivers deliberately did not supervise the automation and the automation failed (NTSB, 2017, NTSB, 2018, NTSB, 2019a, NTSB, 2019b). Disuse of ADAS is seen in evidence that drivers avoid purchasing certain ADAS components (Barry, 2018) and turn off others (Reagan & McCartt, 2016). A recent survey by Consumer Reports found that even though backup crashes are much less frequent than crashes in the forward direction, drivers shopping for a new car valued backup cameras more than AEB. Similarly, even though blind spot monitoring systems are not as effective in preventing crashes (reduced lane-change crashes by 14% (Cicchino, 2018a) as AEB systems (reduced rear-end struck crashes with injuries by 45%, Cicchino, 2017a), drivers shopping for a new car valued blind spot monitoring over AEB systems (Barry, 2018). This evidence suggests drivers are not aware of the relative effectiveness of currently available ADAS for crash mitigation.’
In the US, the NHSTA gives this warning:
‘It’s important to understand that some driver assistance technologies are designed to warn you if you’re at risk of an impending crash, while others are designed to take action to avoid a crash. Be sure to review your vehicle’s owner’s manual for more information on your vehicle’s technology and safety features. Understanding how the technology works and how it can better protect you, your passengers and others is key.’
Interestingly, some reports now suggest that
‘[c]ar buyers understand and want advanced driver assistance systems (ADAS) features that provide safer driving - and they expect them to be standard in new luxury and mainstream vehicles. But the same feelings do not apply when ADAS is applied to autonomous driving systems.’
Given this complexity and variability, what is it reasonable to expect that drivers know or ought to know? How should the level of skill and behaviour expected of a reasonable qualified experienced driver be assessed to determine?
Broader questions of equity also arise. Newer cars and cars with ADAS cost more than older cars without those features. The Australian passenger fleet has an average age of over 10 years. ‘[B]uyers may really want safe cars, but they will resist paying extra for safety features.’ Many car owners (especially younger and older drivers – who may pose particular risks) cannot afford to either purchase a newer vehicle or upgrade their existing vehicle.
And as every lawyer knows, the calculus of negligence balances not only how the risk should be assessed, and what precautions are reasonable in response to that risk, but also the cost of those precautions and social utility considerations. Any legal response in relation to liability and use of ADAS will need to consider both intended and unintended consequences carefully.
Mandating use of vehicles with the highest levels of ADAS will impact different sectors of the community in different (and unequal) ways. In only one of what no doubt could be many more examples, those routinely who travel long distances on roads with little traffic (especially in rural regional and remote areas) might benefit differently from driver alert systems (which might reduce single vehicle crashes as a result of inattention of falling asleep) from those who drive primarily in busy urban areas (where forward collision warning systems might be particularly useful).
As new ADAS continually develop and appear first in high end (and the most expensive) vehicles, how often should vehicle owners and users be expected to upgrade? Should passengers have an obligation to check a vehicle’s safety features before accepting a ride? Should drivers receive particular training in how to use these technologies most effectively before they can be licensed? Should it be acceptable for drivers and passengers to continue to use vehicles that pose significantly greater risks than those with ADAS, or should this risk be acknowledged when liability for harm attributed? What happens if the ADAS don’t perform in the way the driver or passenger expects – even where that expectation might be misplaced? What broader costs might be involved – fleet replacement, insurance premiums, accessibility to transportation, and impacts on community mobility more generally? How should benefits to individuals and communities be weighed against those costs?
While work is already underway in Australia to support the “nationally consistent regulatory approach to enable the safe operation of automated vehicles when they are deployed on Australian roads” and the “Automated Vehicle Safety Law is anticipated to commence in 2026”, humans will still be driving motor vehicles in Australia for many years.
‘The Australian Design Rules (ADRs),… national vehicle standards for vehicle safety, anti-theft and emissions, … apply to new vehicles manufactured in Australia and to imported new or second-hand vehicles being provided to the Australian market.’ The ADRs already require various ADAS for passenger and/or omnibus vehicles including ESC, Brake Assist and now (as of 2024) Advanced Emergency Braking for Passenger and Light Goods vehicles. The next few years will see these and other ADAS features become even more common in vehicles on Australian roads.
These safety features already available in vehicles currently on roads offer significant safety gains – reducing in errors by human drivers. Even small reductions in road crashes can have huge impacts – lives saved, injuries minimised or avoided, costs reduced or not incurred at all.
What impact could (or should) all this have on legal liability for road crashes and trauma? Or on what risks the community views as acceptable?
The evidence is pointing more and more clearly to what makes vehicles, drivers, passengers and our roads safer. Is it time to think differently about how the law can be used to leverage currently available measures to prevent ‘80-90% of fatalities and serious injuries’ and work more intentionally ‘towards zero’?
We don’t need to wait for fully autonomous vehicles to start driving towards that goal.


Further reading Law and tech collide: foreseeability, reasonableness and advanced driver assistance systems