Dawn of autonomous vehicles: review and challenges ahead
Unemployment
Unemployment is a major problem that can arise from automation advances. Autonomous vehicles will threaten the job of professional drivers (see e.g. CNBC (2016)) and change the required skills for workers linked to mobility systems. Taxi and other on-demand services drivers may be the first to experience this threat, as corporations already began driverless experiments. The UBER experiment in Pittsburgh is one such example. Truck drivers may come next, as the sedentary and predictable driving style makes it “a job ripe for disruption” (The Guardian, 2016a). Also, as will be discussed below, automation allows the efficient technique of platooning of heavy weight vehicles, whose fuel efficiency gains may further encourage trucking companies to go Autonomous vehicles. Even if the more demanding task of driving on national/municipal roads is, at a first stage, better done by a human driver, it is likely these become completely expendable as time goes on. Companies related to vehicle repair and maintenance may experience a reduction in demand for services, due to less accidents (see below)
Accident reduction and insurance premiums
About 90% of all accidents are due to human-error (Bengler et al. (2014)). AVs have the potential to drastically reduce accidents, as AV driving is not subject to distraction, bad driving behaviour, and slow human reaction times. Consequently, it is expected that AVs have lower insurance prices (The Guardian, 2016b). Insurance companies will have to face up to new challenges such as accident liability. Schroll (2015) suggests the elimination of liability for any accidents involving self-driving cars, and recommends the creation of a National Insurance Fund to pay for all damages resulting from those accidents. In addition, other risks will arise, which will need to be evaluated by these companies, such as cyber risk and system failures.
Technological barriers/developments
Prior to massification, AVs need to be able to operate competently in the heavily constrained 2D environment that is traffic, especially urban traffic. Reaching navigation proficiency in this environment is much more complex than in the case of airplanes, which is the main reason auto-pilot appeared for airplanes much sooner than for cars. AVs use various sensor’s systems and digital maps to scan their nearby environment. All information is combined in an on-board computer system that uses sophisticated algorithms to determine if the vehicle can move to a new position, in a continuous process which makes decisions many times per second. This navigation system must be reliable in all weather conditions and environments for that, there is still a lot of work to do. Vehicle map position must be precise and reliable in real-time, as well as environment information. Some situations have proved very challenging, mainly because technology is based on optical systems. For example: hidden lane markings, night-time, bad weather conditions, bridges, blinding light from the sun, obscured lights, unusual signage, four-way junctions, hand gestures, head nods and hand gestures, blocked GPS signal, etc. In addition AVs will need to be able to recognize and deal appropriately with unforeseen situations. Overcoming these situations reliably requires improving maps, sensors and computer algorithms. Deep learning systems (TechCrunch, 2016) may help dealing with unpredictable environments, as they can pass human-like decision making patterns to vehicles. This requires however a long learning period, which may delay time-to-market. Instead, some improvements can also be made to fix the infrastructure in order to be as predictable as possible (WIRED, 2016). In the quest for high-precision maps and GPS data, which are essential for AVs, the Japanese government and the European Union plan integration of their GPS satellite constellations (NIKKEY Asian Review, 2016). An alternative to GPS (or complement to it) was developed by a team of USA researchers, exploiting existing environmental signals such as cellular and Wi-Fi (TECH i.e., 2016). Detailed street-level maps of cities using vector-based graphics have also been developed (Road Show, 2016). With this technology, AVs determine their position by calculating their distance to known objects, instead of using GPS. Other technological developments include: software for vehicle guidance without GPS by Oxbotica (Popular Science, 2016); a localizing ground-penetrating radar (LGPR) that works well in all weather conditions, day and night, developed by the MIT Lincoln Laboratory (MIT News, 2016); LIDAR technology, which allows AVs drive in the dark as in daylight (Fortune, 2016) (electrek@, 2016). This technology has become increasingly cheaper and smaller (Autoevolution, 2016).
Legal, liability and ethical issues
Along with regulatory legislation on how AVs are to be used, the Highway Code and certification standards will need to be revised. The Gear 2030 (EC, 2016) presents a review of the EU legislation related to AVs, with special attention to the challenges that such vehicles will pose. Of particular importance is liability in accident cases. If an accident occurs, who is liable? The car’s owner, or the automaker? Some automakers (Mercedes, Volvo and Goggle) said they will accept responsibility and liability if their technology is at fault once it becomes commercially available (Jalopnik, 2015). Hevelke and Nida-Rümelin (2015) present a discussion about who should be held responsible for accidents of fully AVs from a moral standpoint. According to them, automakers’ responsibility should be limited to not obstruct AVs improvements. As to ethical problems, these may arise before imminent crashing, with algorithm behaviour having to decide which humans to endanger (Goodall, 2014). Cybersecurity and data privacy Like with all electronic devices, AV cybersecurity is a serious issue. In 2015 two hackers remotely took control of a Jeep Cherokee (WIRED, 2015) and this year a team of hackers did the same for a Tesla Model S (The Guardian, 2016c), raising fears that a large-scale attack could bring a city to a halt. As a result, some carmakers and service companies resorted to crowdsourcing, rewarding hackers who find bugs in their software. A set of automotive cybersecurity best practices was also published (AUTO-ISAC, 2016). AVs collect massive amounts of data as they operate, data which can be “foodstock” for business opportunities (Financial Post, 2016). Associated to this are data privacy and security issues. Authorities are becoming aware of these issues and draft regulations are starting to appear (DMV, 2015)
Conclusions and future:
The usually hard task of anticipating the future becomes even harder when a large change, such as the appearance of autonomous vehicles, is looming. Many consequences of their appearance are, at best, nebulous at the present stage, along with how deep they might reach. What seems to be consensual is the fact that AVs will bring, sooner or later, a paradigm change in transport. Whether this change will come from widespread adhesion to new car usufruct models, traffic efficiency, electric powering, other factors or any combination of former is not clear. Neither is what practical implications it will have on cities, especially if one considers that cities change due to many factors. This research summarised the possible impacts of AVs and the state-of-the-art with respect to academic research on the subject. It also argued that AVs can make alternative car usufruct models more affordable, possibly to an extent large enough that these can subsequently shape the city and traffic. However, technological hurdles must be surpassed before any of these come to fruition. Recent Tesla auto-pilot crashes prove even the more basic AV functions need vital upgrades, leaving harnessing of all AV potential still far away. An AV-based society requires a different collective mind-set. People will need time to adapt, and it is likely that some die-hards will never give-up being at the helm. Nevertheless, with big companies so committed to AV development, it is only a matter of time before some of the aforementioned impacts unveil. Some forecasts as to what the future may bring can be found in the recent work of (Litman, 2015). With research on AVs impacts still at its infancy, the field is ripe for exploring and modelling them in a quantitative way, anticipating their extent and side effects. Research on usufruct model changes, traffic demand and congestion, and how to best integrate AV with normal traffic rank on top of this to-do list, given their potential to shape the city and its urban way of life. Extending the study presented in this article in section 3 to more cities and countries may help estimate short- to mid-term regional penetration rates.
P O S I T I V E
STRENGTHS INTERNAL
Accident reduction
Cheaper insurance policies
Equity (mobility for all)
Increased traffic efficiency
Increased on-demand services
Reduced travel costs
Reduced parking demand
OPPORTUNITIES EXTERNAL
Change vehicle ownership paradigm
Use of smaller vehicles
Fosters use of electric vehicles
Requalification of parking space
Congestion reduction
Emissions reduction
N E G A T I V E
WEAKNESSES INTERNAL
Lack of regulation/certification
Reliance on technology
Costly technology
Technological issues to be solved
Legal & liability issues
Ethical issues
THREATS EXTERNAL
Unemployment
Urban expansion
Increased traffic demand
Data privacy
Data security
Reference: https://repositorioaberto.uab.pt/bitstream/10400.2/7050/1/driverless_postprint.pdf
