1
Introduction

This chapter provides an introduction to the whole book. After a section on motivation and introduction, a brief history of automated driving is presented, followed by how Advanced Driver Assistance Systems (ADAS) naturally evolved into autonomous driving functions. Some past and current autonomous driving architectures are presented using examples from the field. A literature review section where the key papers and more recent developments in path planning and robust path-tracking control for autonomous road vehicles, also including the relevant literature on cybersecurity, and how it relates to autonomous vehicle path planning and tracking, are summarized next. This is followed by a section on the scope of the book, briefly detailing what is covered in each chapter. The chapter ends with a brief summary and concluding remarks.

1.1 Motivation and Introduction

The race toward series produced autonomous road vehicles has been rapidly progressing during the last decade. Most automotive OEMs and technology companies had promised or forecasted autonomous driving models by the year 2020, two years before the publication date of this book. This obviously did not take place. While we do not have truly autonomous driving vehicles that the public can currently buy, the currently available lane keeping, adaptive cruise control (ACC), emergency braking systems, traffic jam assistants, and their extended versions in some vehicles allow an almost autonomous highway driving experience under ideal conditions [1]. Autonomous shuttle service has been successfully deployed in a lot of different geofenced areas worldwide [2-4]. Large-scale autonomous taxi service is about to start in several countries in Asia soon, using drive-by-wire vehicles retrofitted with sensors and control systems [5]. Autonomous vehicle races have also been increasing around the world [6]. Autonomous delivery vehicles and autonomous truck platoons are also technologies with many successful, limited-scale deployments [7,8]. Automotive OEMs were planning to introduce autonomous products for the fleet market first, before making them available to the general public. Introduction of autonomous vehicle fleets that can also be used as ride hailed taxis is now expected by the year 2023 even though there may still be delays considering the failed predictions of the recent past. The current technology of traditional and nontraditional automotive OEMs and technology companies like Google's Waymo, and similar ones is sufficiently advanced for nearly full driverless operation in well-mapped environments under ideal conditions. The relatively smaller percentage of nonideal conditions and uncertain environments make it difficult to implement full-scale autonomous driving under arbitrary conditions and environments.

Figure 1.1 Categories of autonomous driving according to SAE.

Autonomous road vehicles have been categorized into six categories by the Society of Automotive Engineers (SAE) as shown in Figure 1.1 [9]. Currently available automated driving technology in series produced vehicles falls under Level 2 which is partial automation. Level 2 partial automation is achieved in series production vehicles with lane-centering control for steering automation and ACC and collision avoidance for automation in the longitudinal direction. L3 partial automation is characterized by all driving actuators being automated and the presence of a driver who can intervene when necessary. Recently introduced autopilot systems for cars are examples of conditional automation where the car takes care of driving in some driving modes like highway driving but the human operator is always in the driver seat to take over control if necessary. The Highway Chauffeur is a Level 3 autonomous highway driving system in which almost all highway driving functions are carried out autonomously, but the driver is needed to take over if something goes wrong or might go wrong like a lane change maneuver [10]. The Highway Chauffeur is currently available technology for series produced vehicles and uses an eHorizon electronic map to take care of driving on the highway until the chosen exit is reached. The Highway Pilot is a Level 4 autonomous driving extension of the Highway Chauffeur [11]. The driver is still in the driver's seat but the vehicle can perform highway driving completely autonomously without the need for driver interaction. Highway Pilots are expected to enter the market after 2022 [12].

In Level 5 driving automation, there is no need for a driver as the vehicle takes care of all driving tasks autonomously. It is clear that SAE Level 4 and Level 5 autonomous vehicles have to be capable of making their own decisions based on situational awareness using perception sensors and decision-making algorithms to satisfy the fixed mission of following the highway between initial and final destination locations. This includes planning their route once the destination point is specified and taking care of path planning, path-tracking control, and collision avoidance maneuvering, if needed, autonomously. This same approach is also needed for the lower speed autonomous driving in urban city environments which is a much more complicated situation due to the many other actors like vulnerable road users being present and more unexpected situations being likely to occur. This book treats path planning, path tracking control, and collision avoidance maneuvering for both urban and highway autonomous driving and also treats pedestrian collision avoidance of autonomous driving in the context of the urban application.

Automated driving shuttles in smart cities that are used for solving the first-mile and last-mile problem are other well-known, emerging examples of autonomous road vehicles [11]. These shuttles operate at relatively lower speeds which definitely improves safety levels while also creating a traffic bottleneck around them. In comparison to limited access highway operation, these shuttles operate in significantly less-structured environments with unpredictable interaction with vulnerable road users such as pedestrians, bicyclists, and scooters. The roads they use involve pedestrian crosswalks, intersections with or without traffic lights, roundabouts, and sharper turns as lower speed of operation is possible. Successful applications of these low-speed autonomous shuttles exist in fixed routes. The whole route needs to be mapped in advance and extra landmarks in the form of signage have to be added in some cases as scan matching of the recorded map is used for localization of these autonomous shuttles. Level 4 like autonomous driving of these shuttles is achieved during the segments of the route without intersections and unexpected interactions with other road users. The safety driver takes over control of the vehicle in intersections and during unexpected events. This is called assisted autonomy and is currently necessary for safe operation. True Level 4 autonomous driving capability of these low-speed urban environment autonomous vehicles is expected to be realized in the near future.

The most fundamental task of an autonomous vehicle is the ability to plan and follow a path while avoiding collisions. Path planning is optimized to make sure that the resulting trajectories have comfortable motion with limited acceleration and jerk. Uncertainties in environmental conditions, vehicle dynamics, vehicle load, and load distribution and the range of required speed from very low speeds for urban driving to highway driving speeds require the path tracking and collision mitigation controls to be robust. The motivation of this book is to contribute to this very important area of autonomous driving by presenting recent research results in path planning and robust path-tracking control. Robustness is achieved through two different approaches. The first one is regulation of the path following dynamic model to reject the uncertainties and disturbances and to handle the variable time delays that are present. The second approach is to use a robust feedforward and feedback controller combination to achieve guaranteed performance. The presence of static or moving obstacles such as other cars, pedestrians, and bicyclists is also treated by presenting methods for modifying the path to avoid such collisions in realistic applications. The methods presented in the book are applicable in real life, having been tested in a realistic hardware-in-the-loop simulation environment and in road testing with a research-level autonomous vehicle in addition to the usual model-in-the-loop simulations.

1.2 History of Automated Driving

Contrary to popular belief, the origins of autonomous driving and automated vehicles go back all the way to the 1920s. Radio-controlled cars were the novelty in the 1920s while 1960s and 1970s saw the emergence of cable-controlled cars, actually and unknowingly taking a step backwards. Computer-controlled cars resembling today's autonomous vehicles started emerging in a very rough form in the 1980s and 1990s. In the first driverless car experiments of the 1920s, an antenna was mounted on the car which was driven by an external operator using radio signals, much like radio-controlled toys. It should be noted that this remote operation forms the basis of some current driverless vehicles being...