International Transportation (67) 2 | 2015 6 Back to reality! The future of autonomous cars Driver assistance, self-driving cars, road transport, urban transportation The decades-long evolution of driver assistance systems in road vehicles will doubtless continue in the future - particularly with regard to passenger safety. In this context it may be possible to achieve driverless movement to a certain degree and under specific structural and traffic-related conditions. Taking into account the reality of the road transport environment and the resulting technological limitations, an approach based exclusively on autonomous cars is fundamentally misleading. It is not justifiable at all, in particular regarding aspects of safety, security and ethics. Author: Andreas Kossak F or years now, forecasts, scenarios and daydreams, development activities, fairs and congresses as well as articles in scientific journals, magazines, newspapers and even in the “yellow press” have been revolving around the imminent drastic changes in individual motorized mobility - catchword: autonomous cars. Some examples: • The transport system concept that won the “Smart Cities Audi Urban Future Award 2012” is based primarily on autonomous vehicles and automated guideway systems [1] (figure 1). The transport system of 2014’s award-winning concept is similar. Comment of Audi CEO Rupert Stadler: “The mobility revolution is the great field of opportunity for the 21st century” [2]. • Scientists at renowned US universities (MIT, Stanford) conducted modellingbased calculations regarding the substitution of all motorized personal traffic (including transit) in the centers of metropolises by “Robotaxis“. According to the case study for Singapore, about 250,000 such vehicles would be needed. However, the average travel time during the rush hours would double, compared to the “conventional” system [3]. • The investment bank Morgan Stanley published a “Blue Paper Autonomous Cars: Self-Driving the new Auto Industry Paradigm”. In this paper it is predicted that “Suburbanization” will be a future megatrend, because when people can do other things while they drive, more of them will live further from the denser parts of urban areas. That is the com- Photo: Daimler AG Figure 1: Vision of urban mobility - “Audi Urban Future Award 2012” Quelle: [2] STRATEGIES Autonomous Cars Autonomous Cars STRATEGIES International Transportation (67) 2 | 2015 7 plete opposite to the megatrend “Urbanization” that sociological discussions hitherto tend to focus on [3]. • According to numerous “future and mobility researchers”, the privately owned car will soon be termed a fossil, because it will be fully replaced by the sharing of conventional or autonomous cars [4, 5]. • The software giant Google announced having developed a fully autonomous car and to be looking for an industry partner for mass production [6]. • The German DOT stated that it fully supports the “development of autonomous driving” [7]. • At the “Consumer Electronics Show” (CES) in Las Vegas in January, several German automobile manufacturers presented their concept vehicles for autonomous driving. A slogan at the conference claimed that, “Self-driving cars are the next big thing” [8]. In various scenarios regarding the future of mobility, people will need neither a car of their own nor a driver’s license to use autonomous cars/ vehicles/ Robotaxis in order to nominally serve any mobility demand in cities as well as in rural areas at- any time and even much better than today. The euphoria in this context is founded on the one hand on statements of- technologyand marketing-oriented representatives of the industries involved, and on the other hand on the conviction of- self-appointed experts, who are fascinated by the supposedly unlimited technological possibilities. From the camp of qualified independent experts, however, increasingly critical voices addressing the manifold hard limitations of operating autonomous cars in the reality of (in particular urban) road transport are making themselves heard [3]. Basics The idea of a road transport system that is based on autonomous cars is far from new. In the late 1950s, for example, stories about “cars of tomorrow” able to drive completely autonomously were in full vogue (figure 2). More than 50 years later, at the 2015 CES, manufacturer Daimler presented a study that looked remarkably similar to the “historical” one - except for the different leisure activity of the passengers [9] (figure 3). The decisive difference is that since the 1950s there have been some far-from-completed, but very useful developments in the area of driver assistance tools. That progress is expected to advance much further in a very short time, resulting in fully autonomous cars. However, this extrapolation neglects fundamental facts. The possibilities and limitations of autonomous cars/ road vehicles can only be described realistically if we define the fundamental characteristics of the road traffic system on the basis of the empirical sciences: Modern settlement and traffic structures are infinitely complex systems that are continually changing as a result of infinitely complex processes. This is in particular the case in urban areas. By definition, such systems can only be modelled within very narrow limits and cannot be comprehensively controlled by technological means. Robert Poole, one of the leading transport researchers in the US, has his own take on this issue. The cofounder and Director of Transport Policy of the Los Angeles-based Reason Foundation, recently summarized his conclusions of some related expert discussions [10, 11, 12]: “...while I see significant potential, the more serious literature I review, the more skeptical I become about the popular media hype of cars without any function for a driver, going anywhere on demand. [...] Let me summarize the main points raised by this collection of experts, as follows: • Automation is inherently brittle and subject to failures. • Hence, for at least a long time, a driver must be able to take over on short notice. • We don‘t really know how to provide such transitions, and the aviation experience is troubling. • There are ambiguous situations where we may not want the automation to make the decisions.” Selected aspects Limited control As a consequence of “infinitely complex” boundary conditions, mastering the autonomous vehicle concept will be possible only to a very limited degree, even if we presuppose a further massive progress regarding programming and computer capacities. That is why qualified insiders, with good reasons, are marking the limits of the practicability of autonomous cars as follows: “Cars are only able to select from what the programmer has defined” [13]. In particular in cities, cars are closely integrated into a complex multimodal traffic environment. The behavior of pedestrians, cyclists, children and elderly people in such an environment cannot be modelled to an even remotely comprehensive degree by a computer. Even the strongest advocates of maximally autonomous cars/ vehicles concede that such systems are susceptible to failures - the more complex the configuration, the higher the probability of malfunctions. Traffic safety and ethics The development of driver assistance systems focuses on improving traffic safety, among other things. A representative of Figure 2: Driverless cars of the future in an advertisement for “America’s Electric Light and Power Companies” Source: Saturday Evening Post, 1957. Credit: The Everett Collection Figure 3: “Autonomous driving with Daimler” - cars of tomorrow in 2015 [9] Photo: Daimler AG STRATEGIES Autonomous Cars International Transportation (67) 2 | 2015 8 Volvo, a company manufacturing cars that are traditionally known as being extraordinarily safe, recently underlined the company’s vision that “by 2020, no one should be killed or severely injured in a new Volvo car” [14]. That sounds good; however, it obviously pushes aside answers to fundamental ethical questions in this context, such as: • Who decides about the risks for people being involved in an unavoidable accident? • How should car manufacturers decide how to weigh up the deaths of people sitting in other manufacturers’ cars against the risk for people sitting in one of theirs? While the general public is fascinated by the supposedly unlimited technological potential to reduce the number of car accidents and of accident victims to “nearly zero“ and therefore concludes that society is ethically bound to implement a comprehensive system of fully self-driving cars [6], the position of qualified independent experts looks completely different: “But no technology is perfect, especially something as complex as a computer, so no one thinks that automated cars will end all traffic deaths. Even if every vehicle on the road were instantly replaced by its automated counterpart, there would still be accidents due to things like software bugs, misaligned sensors, and unexpected obstacles. Not to mention human-centric errors like improper servicing, misuse and no-win situations […]” [15]. As recent studies of renowned academic institutes have revealed, it is anything but certain that autonomous cars will ever make road transport safer. In particular during the transition phase (mixed traffic) of at least 20 years, the safety standard will even be substantially lower than today [16]. In this light the question must be raised if the whole approach can be justified at all. Vulnerability Transport infrastructure is a fundamental function of our settlements and our economy. This is particularly true for the road sector. Consequently, this infrastructure is of high interest to criminals and thus highly susceptible to criminal activity, as well as to “experimental” cyber-attacks. Digital worlds are (and will be in the future as well) vulnerable to manipulation by hackers, internet criminals and terrorists. As a consequence, a road transport system based on autonomous cars is much more vulnerable than the “classical” system. Legal questions, standardization, interoperability The ethical aspects of traffic safety within a system of autonomous automobiles are closely related to legal questions. This is true for virtually all components of traffic law. It is unlikely that even the key problems in this context will be solved in an acceptable and conclusive way. What happens, for example, in case of a cruel accident that is not caused by any person [14]? Our automobile transport system is composed of vehicles of a great variety of types and makes supplied by an international industry. The share of the transport performance of foreign (licensed) heavy trucks using Germany’s motorways, for example, amounts to about 30 %, the share of foreign licensed cars to around 5-10 %. Hence, a complete system based on autonomous automobiles would presuppose a degree of international standardization and interoperability that would be far more complex than the highly complicated compendium of laws and regulations in force today. Anybody familiar with the related problems within the European Union alone, will easily come to the conclusion that it will take not just years, but decades, to define and put into effect the required basic laws and regulations for the large-scale operation of completely autonomous road vehicles in Europe. Structural conditions Among the main drivers behind the development of autonomous road vehicles are manufacturers and research institutions in the USA. That is not least due to the specific conditions that characterize American urban areas and highways [4]: • The width of the roads in most cities is “comfortable” and the road networks are mostly designed as a rectangular grid, often with a comprehensive one-way regime. • Interurban road traffic typically involves very long distances on limited-access highways. Taking that into account, the conditions for the selective application of road vehicles being equipped with a large arsenal of driver assistance devices, and thus being able to drive autonomously under certain conditions, are by far better in the US than in Europe. Still, skepticism is growing even in the US. Costs Today the costs of equipping a car with the devices necessary to allow driverless travel under special conditions are several times higher than the costs of a car with conventional driver assistance systems [5] (figure 4). Even if we consider that mass production will considerably lower the costs of the different components, the costs would most probably still exceed the amount acceptable for the majority of car owners, especially in view of the limitations on use of the driverless mode in real-world traffic. Only if many, most, or even all autonomous cars are provided by a public or private operator and the individual vehicles are actually used 10 times on average or even more than a privately owned car, the operating costs for a trip may be acceptable. Privacy etc. Privacy concerns are some of the main obstacles standing in the way of automatic road toll collection. How will it be possible to ensure privacy in a world of autonomous cars? What about serving the greatly varying requirements of car users regarding size, comfort, space for luggage, and all their other personal preferences? Large-scale car sharing schemes will raise questions of hygiene and cleanliness. And how do we plan to combat vandalism and robbery, which have been increasing in the motor vehicle sector for decades? Figure 4: The Google concept vehicle for autonomous driving Source: Google/ Raoul Rañoa Autonomous Cars STRATEGIES International Transportation (67) 2 | 2015 9 Substitution of transit The question if public transit services will be replaced by autonomous cars was on the agenda of last year’s annual meeting of the Transportation Research Board (TRB) of the US National Academies of Sciences. Even autonomous-car enthusiasts argued in favor of substituting classical public transit by autonomous vehicles only to the limit of complementing public transit in by-demand areas. The result would be some kind of advanced variant of “Paratransit” [17]. However, even in that regard several questions remained unanswered: What about passengers needing special assistance? Who are the owners and who are the operators? Conclusion The decades-long evolution of driver assistance systems in road vehicles will continue without any doubt - particularly with regard to passenger safety. In this context it may be possible to achieve driverless movement to a certain degree and under specific structural and traffic-related conditions. Taking into account the situation in Europe, this could be possible first and foremost only on limited access roads outside urban areas. In- the foreseeable future, autonomous motorized personal transport will not and should not substitute classical motorized road transport or even substantial parts of it, in particular in urban areas. The individuals and organizations in charge of planning and managing road transport are well advised to rely on the point of view of independent qualified experts in this regard and to concentrate on taking full advantage of the huge not yet utilized technological potential of improving the classical systems. This refers in particular to payment, information and communication systems covering the entire transport sector and related markets as well as to optimizing operations in multimodal traffic environments and the operation of “autonomous rail systems”. ■ REfERENCES  [1] Audi AG: Audi Urban Future Award 2012; in: Die Welt, special issue “Smart Cities”, October 20, 2014  [2] www.audi-urban-future.com  [3] Poole, R.: Progress - and more Questions - on Autonomous Vehicles; in: Surface Transportation Innovations, May 2014  [4] Kossak, A.: Zur Zukunft des Stadtverkehrs; in: Der Nahverkehr 10/ 2014  [5] Wikipedia: Google Driverless Car; accessed on September 8, 2014  [6] Stein, H.: Wir Automobil Barbaren! ; i: Welt am Sonntag, December 14, 2014  [7] Bundesministerium für Verkehr und digitale Infrastruktur: Info- Papier “Automatisiertes Fahren”; Berlin 2014  [8] Grünweg, T.: Kräftemessen der Autopiloten; in: Spiegel Online, January 6, 2015  [9] sv/ sp-x: Lenkrad? Wozu denn? So sitzen Sie im Mercedes von morgen; in: FOCUS Online, January 4, 2015 [10] Cummings, M.L. and Ryan, J.: Who is in Charge? The Promises and Pitfalls of Driverless Cars; in: Transportation Research News, May/ June 2014 [11] Hutton, P.: A Driverless Future? in: Thinking Highways, North America Edition, 2/ 2014 [12] Poole, R.: Important new Thoughts on Autonomous Vehicles; in: Surface Transportation Innovations, November 2014 [13] Pritchard, J.: Die Crux selbst fahrender Autos; in: Hamburger Abendblatt, November 25, 2014 [14] dr/ SP-X: Das chauffieren ohne Chauffeur könnte Jahre dauern - zumindest in Europa; in: FOCUS Online, November 20, 2014 [15] Lin, P.: The Ethics of Saving Lives with Autonomous Cars is Far Murkier Than You Think; published on: wired.com, July 30, 2013; http: / / www.wired.com/ 2013/ 07/ the-surprising-ethics-of-robotcars/ [16] N.N.: Autonomous vehicles may ‘worsen safety’; in: ITS International January/ February 2015, p. 7 [17] Badger, E.: What will happen to public transit in a world full of autonomous cars; in: The Atlantic Citylab, January 17, 2014 Andreas Kossak, Dr.-Ing. Kossak Forschung und Beratung, Civil Engineering Lab, Hamburg (DE) drkossak@aol.com The future of mobility. Berlin-Brandenburg is not a traditional automobile location, but one which is set to make a huge contribution to the future of mobility. As the digital capital, the region has positioned itself as a development site and test bed for the latest transport technologies, where established firms and startups are jointly implementing their ideas. If you‘d like to be part of this, please get in touch: www.businesslocationcenter.de/ its Meet us at ITS World Congress October 5-9, 2015 | Booth A3 210x148_Telematik_ITS.indd 2 18.08.2015 11: 19: 10