Autonomous Vehicles

The gradual automation of the driving task and the accompanying shift in performance of the driving task from a human driverto automated driving systems poses the question to what extent this technology can be lawfully used on public roads. This chapter explores this question, primarily focusing on the international framework of the 1968 Vienna Convention on Road Traffic. Discussions primarily focus on the key question whether the notion of “driver” can be faithfully interpreted to permit the operation of self-driving cars. However, the question of who can or should be regarded as driver and the duties and obligations that (should) rest upon him or her are closely intertwined. For this reason, formally amending the Convention by only redefining the notion of “driver” to make it undisputedly consistent with the use of automated driving systems will not be enough to adequately accommodate automated driving. This will also require defining the role and responsibilities of the operator of the automated driving system, as well as considering an alternative system of sanctions in the event of failures or infringements of the rules of the road.

The desire to test autonomous vehicles on public roads confronted the industry with legislation that would not allow the industry to do so. Vehicles without human drivers are often not allowed on public roads. Why did the legislator not simply take out the necessity of having a human driver from the road safety laws or the traffic laws? The cause lies in the technology of the autonomous vehicles. Neither the industry nor academia are able to fully portray the risks, or their absence, of the technology used in autonomous vehicles. To understand the technology in more depth, testing in complex, real life situations, is necessary. Public roads can offer that environment. Without knowing the risks and still faced with demands for testing, legislators did initially not favor hard law instruments to regulate testing on public roads. Soft law instruments existed in various formats. Yet, the more experience the legislators gained with the regulation of testing autonomous vehicles on public roads, more and more hard law instruments started to appear. These hard laws are either allowing the creation of a space to experiment or offering just a framework. Framework providing laws are often supplemented by lower level legislation or guidelines. No matter which approach is taken, the general tendency is that each of these instruments tends to be technology neutral and favor a permit system. In most systems the industry is involved in setting up the parameters of the testing. Insurance is necessary as well as collaboration with public authorities.

Autonomous vehicles are said to bring safety to the roads. Machines are expected not to make the same driving mistakes as humans. Indeed, machines will not drive intoxicated or get too tired to drive. However, the application of adversarial machine learning to autonomous vehicles has shown that the reaction of these vehicles to altered traffic signs may be the cause of unpredictable reactions. Rather than stopping in front of a vandalized stop sign, the autonomous vehicle may speed. This may lead to accidents. Therefore, scholars have developed various liability and compensation schemes to deal with accidents by autonomous vehicles. The following liability and compensation schemes have been suggested to deal with the civil liability of accidents of autonomous vehicles: operator liability, product liability, strict liability, no-fault compensation, and negligence. Each of these schemes are judged against victim and innovation friendliness. The former is being framed as easiness to obtain compensation, while the latter is understood as a burden on the industry. Operator liability, strict liability and no-fault compensation are considered as victim friendly. Product liability and negligence put a burden on the victim to prove either a defect of the product or a fault of the manufacturer. Only by shifting the burden of proof to the manufacturer would these systems be made victim friendly. In terms of innovation, the situation is not obvious. Operator liability, product liability and negligence make it difficult for a manufacturer to anticipate the size of the financial burden in case of an accident. This would be different with strict liability and no-fault compensation. Much of the discussion above is framed in relation to vehicles that are operating autonomously on their own. There is, however, more and more research on infrastructure enabled autonomy. In system, autonomous vehicles will be operating in connection with road side units, cloud services, and other traffic participants. As this will bring together products, services, and behavior, a mix of different liability regimes will make it difficult for the victim to obtain compensation. Therefore, a one-stop window may facilitate obtaining compensation. No-fault compensation could be ideal.

It is widely believed that autonomous vehicles will significantly reduce the number of car accidents and therefore save lives. However, autonomous vehicles will also crash. Uber’s fatal self-driving crash in 2018 has attracted attention from all over the world. This article analyzes the Uber crash case from the perspective of Chinese law. If the Uber crash happened in China, both Uber and the driver would be at fault. Uber disabled the test vehicle’s original emergency braking function under computer control, causing a defect in the test vehicle. Therefore, the original car manufacturer could be exempted from liability due to Uber’s conversion of the vehicle. Uber’s aggressive testing attitude has also played a role in this fatal accident. Besides, the safety driver was on the job when the collision occurred. Therefore, her employer Uber, would be liable for damages resulting from the accident. By analyzing the Uber case under the context of the Chinese legal regime, we found that in the event of an autonomous car crash, potential liable parties would be the car manufacturer, the self-driving system developer and operator, the parts manufacturer, the safety driver, the pedestrian, etc. Liabilities should be allocated reasonably among them. It should be noted that, the Uber crash could also be partly attributed to Arizona’s lax approach to regulating autonomous vehicles, which gives all regulatory authorities a lesson of finding a balance between technology development and safety.

A sensor system capable of supporting automated driving functions needs to provide both reliable localization of the vehicle and robust environment perception of the vehicle’s surrounding. The following chapter introduces the working principles and the state of the art of automotive sensors for localization (GNSS and INS) and environment perception (camera, radar and LIDAR), corresponding sensormodelsand sensor fusion techniques. Sensor models will allow for the replacement of conventional test drives and physical component tests by using simulations in virtual test environments to meet the increasing requirements of automated vehicles with respect to development costs, time and safety. Considering the multitude and complexity of possible environmental conditions, realistic simulation of perception sensors is a particularly demanding topic. To increase the performance of a sensor system, compensate for limitations of each sensor modality, and increase the overall robustness of the system, sensor fusion techniques are an important subject in automotive research.

Deep learning has brought many advances to various fields and enabled applications such as speech and visual recognition to flourish. However, recent findings show that Deep Neural Networks (DNN) still have many problems of their own. The many vulnerabilities present in DNNs unable their application to critical problems. Here, some of these vulnerabilities will be reviewed and many of their possible solutions will be discussed. Regarding legislation, a series of practices will be discussed that could allow for legislation to deal with the increasingly different algorithms available. A small overhead for a safer society. Lastly, as artificial intelligence advances, algorithms should get closer to human beings and legislation itself should face deep philosophical questions in an age in which we will be challenged to reinvent ourselves, as a society and beyond.

Multiple studies have established the potential for significant societal, environmental and economic benefits from autonomous vehicles. However, all the current approaches to autonomous driving require the automotive manufacturers to shoulder the primary responsibility and liability associated with replacing human perception and decision making with automation, slowing the penetration of autonomous vehicles, and consequently slowing the realization of the benefits of autonomous vehicles. We propose a new approach to autonomous driving that will re-balance the responsibility and liabilities associated with autonomous driving between traditional automotive manufacturers, private infrastructure players, and third-party players. The proposed distributed intelligence architecture leverages recent advances in connectivityand edge computing. The resulting Infrastructure Enabled Autonomy (IEA) leads to the new business model of “autonomy as a service”.

The automation of vehicles will just be a part of the mobility ecosystem of the future. Connectivity may be another important pillar in that ecosystem. In fact, recent research shows that connectivity can even facilitate the movement towards automation of vehicles. One of the projects contributing to this research is the Australian Integrated Multimodal EcoSystem (AIMES). AIMES is conducted in the city-center of Melbourne and connects vehicles to everything. The aim of this project is to increase safety among autonomous vehicles and other traffic participants and to enhance traffic flows. With the enhancement of traffic flows, AIMES also researched how connectivity can prevent congestions and reduce carbon emissions. Further, AIMES is a living laboratory and thus providing urban transport research that is more innovative, relevant and far reaching than ever.

In an age of fast-paced technological change and hyper-competitive global markets, all firms must focus on putting in place organizational structures and management processes that deliver innovation. This is obviously a technical challenge (developing and gathering together multiple new technologies), but also a “design” challenge (offering end-users a meaningful and frictionless experience). Meeting these challenges and delivering innovation requires a break from traditional hierarchical and centralized approaches to firm organization and governance. In this chapter, we take recent developments in the car industry—specifically the on-going development of automated vehicles—to identify some of the organizational structures, procedures and practices that deliver the best opportunities for meeting the challenge of designing the intelligent car of the future. The chapter suggests that much of the contemporary discourse of corporate governance and compliance (which remains focused on hierarchy, accountability and disclosure) is disconnected from the business realities and needs of firms today (i.e., delivering innovation) and that new frameworks are required. The chapter identifies a number of features of this new approach.