24/06/2019 – Mobility Series / Driverless Vehicles / Global
A core new focus of R&D investment worldwide, the driverless dream is swiftly journeying towards reality. Quite reasonably, much focus rests on how the technology will respond to the real world, yet how human behaviour evolves could prove more problematic.
The number of traffic-related deaths reached a high of 1.35 million in 2016, according to the World Health Organization (WHO) Global Status Report on Road Safety. Now the leading global cause of deaths for children and young adults between 5 and 29 years old, traffic fatalities has also moved up to the eighth leading cause of death for people of all ages, ahead of HIV/AIDS and tuberculosis.
“Road safety is an issue that does not receive anywhere near the attention it deserves – and it really is one of our great opportunities to save lives around the world,” stated Michael Bloomberg, former New York mayor and the WHO global ambassador for non-communicable diseases and injuries, in the report’s foreword.
While progress has been made in certain areas, it is clear that such improvements are not happening quickly enough to meet the UN's goals to halve road traffic deaths worldwide between 2016 and 2020.
Tightening safety regulations undoubtedly plays a key role in improving survival rates on roads. However, in the future driverless cars could prove an even more effective solution, given their promise of identifying and avoiding the situations that lead to crashes. Unlike human drivers, Autonomous Vehicles (AV) do not get tired, drink-drive, look at their phone, or become tempted to speed. As an added bonus, adoption of AVs would effectively reduce congestion and pollution, increase access to public transport, be cheaper, and improve mobility for people with disabilities. The technology’s many advocates worldwide promise all of this, and more. But can the driverless dream realistically deliver?
Certainly, investors think so, with billions of dollars of R&D funding currently being funnelled into making this autonomous vehicle utopia a reality. For example, from 2016–2018 ride-hailing giant Uber poured US$1.1 billion (around 30 per cent of its entire research budget) into its autonomous vehicle unit – the Advanced Technologies Group (ATG). With a fresh injection earlier this year of US$1bn from three Japanese investors (Softbank’s Vision Fund, Toyota, and auto-parts maker Denso), Uber’s self-driving cars unit has now been established as its own corporate entity, valued at US$7.25 billion. Its self-driving trucks are already operating in Arizona, while the firm is also planning to buy "tens of thousands" of self-driving cars from Volvo in what has been described as a “non-exclusive” deal.
Indeed, while the major automotive manufacturers are engaged in the development of AVs, it is software specialists that are the key component. As the major OEMs (most notably Volvo, BMW, Daimler, Continental, Audi, Toyota, Ford, General Motors, Samsung, Volkswagen, Fiat Chrysler Automobiles (FCA), and Tesla) shift into the autonomous zone, they become ever more reliant on the very different capabilities of tech firms – both giants (Huawei, Baidu, Apple, Nvidia and Intel) and smaller players (Waymo, Oxbotica, nuTonomy, FiveAI) – to advance their ambitions in this nascent market. Resultantly, what would hitherto have been deemed surprising collaborations and acquisitions – ones that effectively bridge the realms of automotive engineering and software development – are now becoming par for the course to steal a lead in the AV race. Other firms also engaged in such refreshing tie-ups include ride-hailing companies (Uber, Lyft) and more traditional taxi setups (Addison Lee), bus transport service provider Stagecoach and electrical consumer goods giant Bosch.
Nearly all of the major OEMs and new mobility joint ventures aim to have a level 4 AV on the market by 2021 at the latest, with plenty of real-life trials already in action. German carmaker Volkswagen, for example, already has five of its electric cars – fitted with the latest in autonomous technology – driving on the streets of Hamburg. The city has installed new signals and other traffic management systems to facilitate so-called ‘level 4’ vehicles, which are designed to handle complex urban traffic patterns without help from drivers (although they must be ready to intervene). And software firm Waymo – subsidiary of Alphabet (Google’s holding company) – became the first firm to offer a commercial ‘robotaxi’ using driverless technology when it launched its service in the suburbs of Phoenix, Arizona in December last year.
On a national level, the Chinese government has inaugurated its far-reaching ‘Made in China 2025’ initiative, which aims to transform the country into an innovation hub that could enable it to dominate the rising Internet of Vehicles (IoV) sector in the years ahead.
Elsewhere, autonomous driving forms a key part of Singapore’s Smart City initiative, under which the government has authorised numerous AV trials as well as encouraging both R&D and investment in Internet of Vehicles (IoV) applications. Notably, the island state’s Land Transport Authority is partnering with Volvo and local Nanyang Technological University to launch the world's first 12-metre autonomous bus.
Such growing momentum worldwide will see the autonomous vehicle market expand from US$54.23bn in 2019 to reach US$556.67 billion in 2026, predicts Allied Market Research.
The AV promise paradox
So, billions more dollars will be poured into the R&D of AV technology worldwide in the years ahead, into advancing the tech’s ability to detect and avoid collision and thus protect human lives (see our Science & Technology feature on MIT’s latest research innovation in that realm, p16). To date, however, less money has been funnelled into how humans will actually respond and adapt to a future world in which driverless vehicles are the norm.
An alignment with Asimov’s First Law of Robotics – that is, that ‘a robot may not injure a human’ – means autonomous vehicles are essentially being built to ‘act flawlessly’. While this sounds good on paper, human’s innate adaptability and propensity to leverage on the advantages thrown up by new environments opens a distinct possibility that the programmed virtuosity of the driverless vehicle could become a point of exploitation by other road users.
For instance, on a city highway saturated by autonomous vehicles, the threat of being run over by a car disappears. Resultantly, you (or, indeed, any other relatively intelligent creature) would swiftly learn that oncoming traffic posed no danger at all. In the process, the dynamics between vehicle and pedestrian (or between an AV and an actual traditional driver) would be skewed to such an extent that the human confronting the driverless car could effectively form a strategy to ‘go’ at all times, striding (or driving) into the road and forcing the safety-conscious AV to stay put. Now replicate that scenario many thousands of times over and you can easily envisage how a city’s traffic might be brought to a grinding standstill.
As Gemma Reed and Jason Thompson posit in an article for the World Economic Forum, “Part of the promise of autonomous vehicles is their proposed safety through deference to human life. But, if the point of transport systems is to enable efficient movement of people and goods for the benefit of society, this strength of AVs might prove to be their ultimate weakness as a viable mass transport mode.”
It is easy to see how autonomous technology could hold the solution to many of the challenges we face in transport. It also continues to hold many fears: An autonomous Uber car achieved a grim milestone for the industry in March last year when it became the world’s first reported fatal crash involving a Level 3 self-driving vehicle and a pedestrian. The vehicle (a refitted Volvo) killed a woman in the street in Arizona, which has become a test-bed for the ride-hailing firm’s AV ambitions since August 2016. The fact the car’s capabilities were classified as Level 3 – that is, kitted out with an autonomous driving system that occasionally requires a driver to take over control – opens the possibility of the human-AI interplay as a weak spot, and the potential that truly autonomous vehicles (Level 4 and 5) would reduce errors and thus collisions.
Nonetheless, bolstered by such tragedies, the vision of an errant, driverless car mowing down passengers is a common nightmare scenario amongst the public. Many people are also uneasy about AVs being given the power to decide who lives or dies in a crash, should it come to that. Whether – and indeed why – the ethics of a panicked human driver (armed with the occasional errors and misjudgements that fallible humans make) should be favoured over that of a robot (programmed to “act flawlessly”) is an emotive dilemma and something that societies must grapple with before widespread adoption of AV. While such fears are understandable (indeed, only human) – the wholesale uptake of AVs could result in a far less envisioned (and far less dramatic) scene: one of passengers stuck in gridlocked, flawlessly-acting AV traffic for hours on end, as pedestrians – liberated from the dangers of unpredictable drivers – start to take liberties themselves.