Sharing the Road: Challenges and Strategies

Abstract

The idea of autonomous cars has been around for decades, but the recent advancements in automation, robotics and communication technology have given sharp rise to the prospect of self-driving/autonomous vehicles technology. With the recent acceleration in research and development in this field, the dream is now turning into reality and soon autonomous vehicles (AVs) and human-driven vehicles (HVs) will be sharing the road. This chapter presents an insight into the possible challenges and hurdles that need to be addressed in order to make this co-existence possible. Considering all possible scenarios and circumstances is crucial to develop the right technology and infrastructure for future transportation systems. The chapter further discusses the strategies and solutions suggested and developed to overcome these challenges.

1. Introduction

With the recent advancements in vehicle technology, the idea of driverless cars, which was once a dream, seems to be turning into reality, and several companies have started to invest into self-driving cars, and trials are already in progress by some well-known companies. This large-scale research, development, investment, and trials suggest that time is not far when autonomous cars and human-driven cars will be sharing the road. As much as the idea sounds fascinating, it brings about several challenges and considerations, such as safety, regulation, dealing with traffic flows and congestion, to mention a few. This chapter aims to the challenges and barriers in large-scale adaption of AVs. The chapter is organized as follows: Sub-Section 1.1 elaborates the classification of vehicle automation which is important to understand the development of autonomous vehicles on different levels. Section 2 mentions the challenges and considerations that need to be addressed while adapting AVs on a large-scale and allowing them to share road with HVs. Section 3 describes some of the strategies and possible solutions suggested in order to overcome these challenges. Finally, Section 4 concludes the chapter.

1.1 Levels of vehicle automation

Depending upon the level of involvement of human in the driving process of vehicle, levels of autonomy are defined for autonomous vehicles, elaborated in Figure 1. The lowest level, Level 0, represents no automation thus all tasks are performed by the driver. Next level, Level 1, has some driver assistance available, e.g., Automatic Braking or Electronic Stability Program. Level 2 is the partial automation level, where some combined automated features are present, i.e., lane-keeping, and adaptive cruise control.

Figure 1.

Levels of vehicle automation (source: Created by author).

The driver still has to be involved in driving and must monitor the environment. Next level, Level 3, is conditional automation level where driver can stop controlling some of the important functions of the car in certain conditions but must always remain ready to take control of the vehicle without any advance notice. Level 4 is a high automation level where the vehicle can fully execute all driving functions. Level 5 is the full automation level where the vehicle can perform all the functions involved in driving under all conditions and circumstances.

2. Challenges and barriers

As mentioned earlier, research and technology advancements in the field of autonomous vehicles have led to the well-known companies racing to develop and trial self-driving cars, and eagerly competing to bring their AVs up and running on the road. No doubt the idea of self-driving car seems fascinating, and this technology also promises to completely revolutionize the future transport by improving road safety and better connecting the communities. While earlier this year, Mercedes became the first officially certified Level 3 autonomy (conditionally automated) car company in the US [1], recent developments in UK suggest its roads could see fully autonomous vehicles rolled out by 2025 [2]. With all these recent developments, it seems that time is not far when we will be seeing AVs on the roads [3, 4, 5], yet there are still several concerns and challenges that need to be accounted for. Researchers have widely discussed possible barriers and challenges in literature [6, 7, 8, 9, 10, 11, 12] and are also working towards the strategies to counter them. Some of the most prominent challenges are listed below:

2.1 Safety and reliability

As soon as we think of a self-driving car, the first concern that comes to our minds is safety and reliability. Despite all the progress, development and excitement, safety and reliability still remain the biggest challenges when it comes to self-driving cars, especially the fully autonomous ones.

According to Mcity ABC Test, presented by University of Michigan, testing the safety of autonomous cars includes three main components: Accelerated evaluation, Behavior competence, and Corner cases [13]. Figure 2 demonstrates the components and sub-components of the ABC Test. As shown in the figure, accelerated evaluation focuses on lane change, car following, and left turns. Behavior competence is about testing the performance in rigorous scenarios, such as, weather and lighting conditions. Finally, corner cases include those cases that are on extremities of the test conditions. For example, zig-zag motion of joggers on the street and detection of dark-colored cars in dark surroundings [13].

Figure 2.

Mcity ABC test components (source: Created by author).

2.2 Testing and validation

A comprehensive and rigorous testing and validation process is required to be established before letting the AVs on the roads. Bringing a driverless car on the road is not only about driving and controlling the vehicle, but also about decision-making, and that too in all possible circumstances. It involves many factors, such as, the number of miles the car is driven during test, the interaction with real traffic that includes decision-making as well, and special circumstances i.e., congestion, traffic flow, weather, lighting conditions and other scenarios mentioned in 3.1. Without considering all these factors, an AV, especially an L4/L5 AV cannot be fully validated [14]. For reliable testing and validation, ground truth data must be used. Test tracks are usually designed to create and re-create critical scenarios in a controlled environment. The disadvantage, however, is that only a limited variety of scenarios can be tested in a short period of time, lacking the exact vehicle and traffic dynamics and possible congestions.

2.3 Traffic congestion and unexpected encounters

Another important challenge for autonomous vehicles is encountering traffic congestion and other unexpected circumstances e.g., a traffic warden guiding the traffic to pass through a red light [15]. Although, by using Advanced Traffic Management Systems (ATMS), on-street traffic data can be collected and used to predict traffic flows and patterns [16], but still the possibility of unexpected traffic flows and congestion can almost never be eliminated. In addition to traffic flow prediction, autonomous car should be able to take adaptive decisions according to the surroundings objects and conditions. Similarly, driving an autonomous car in tunnels or on mountains, on bridges or intersections, can be challenging.

2.4 Technology and infrastructure

Most of the AVs will be relying on Artificial Intelligence (AI) and Machine Learning to process the data obtained from the sensors and to help in decision-making. These algorithms are going to detect objects, classify them, and take actions accordingly, e.g., buildings, other vehicles, traffic signals, road signs, streetlights, and pedestrians. AI is undoubtedly taking over the world in autonomous vehicle technology as well as in other disciplines of life, there is still a long way to go. AI cannot understand real-world scenarios, for instance, if the AV sees a plastic bag flying in front of the car or if it senses a flock of birds sitting on the road, it will stop unnecessarily. Unlike human drivers, AI cannot understand that birds will fly away as the vehicle moves forward [16, 17, 18].

In terms of infrastructure also, major changes will be required. Clear traffic lane markings and traffic signs are required. If the vehicles run on electricity, a robust charging network is needed. Network providers will have to ensure a seamless connectivity in order to avoid communication and connectivity issues. Technology and infrastructure upgradation requires huge investment [16, 17].

2.5 Sensors

Autonomous vehicles use a wide range of sensors to “see” the environment around them, collect the information and feed this collected data to the control system so that decisions can be made and required action can be taken. Cameras help to view objects; LiDAR (Light Detection and Ranging) sensors use light in the form of a laser to measure the distance between the object and the vehicle; RADAR (Radio Detection and Ranging) sensors detect the objects, measures their speed and the direction of their movement [17, 18, 19, 20]. Ultrasonic sensors are also used that help to measure short distances at low speeds. They are independent of color, and work well in bad weather conditions and dusty environments [13].

In a fully autonomous car, sensors should be able to detect objects, distance and speed under all weather conditions and environments without the need of human intervention. The accuracy of the sensing capability can be negatively impacted due to adverse weather conditions (fog, snow, or heavy rain), traffic signs with graffiti, heavy traffic, and low light. Cameras and sensors are not able to track lane markings if they are covered by water, ice, oil, or debris [15, 17, 18, 19, 20]. Figure 3 shows the sensors, cameras, and other basic components of an autonomous car.

Figure 3.

Sensors on autonomous car (source: Created by author).

2.6 Complex 3D route map creation and maintenance

Autonomous vehicles heavily rely on pre-defined maps in order to navigate and reach the desired destination. With the help of sensors and maps, AVs are able to detect the obstacles and follow the right path. For AVs to perform accurately, three-dimensional (3D) maps need to be created. This can be done by driving the vehicle on all routes, capturing images and categorizing them into intersections, driveways and fire hydrants etc. It is a time consuming and complex process in terms of coverage and efficiency. If, however, the user wants the car to go to a new location that is not already saved in the maps, or traffic signals are changed, or a new construction work takes place, 3D maps will not be able to help [15, 17].

2.7 Artificial versus emotional intelligence

Emotional Intelligence (EI) refers to the ability to recognize, identify and manage one’s own emotions and the emotions of others. It includes characteristics such as self-awareness, self-regulation, motivation, empathy, and social skills. Driving a car is not only about skill and control, but it is also a social process where human drivers interact with each other and with pedestrians, reading their facial expressions and body language, predicting behaviors, and making quick decisions [15]. This is something that seems almost impossible to achieve as human instincts, behaviors and emotions cannot be replicated or replaced by machines.

2.8 Cybersecurity

Cybersecurity and data privacy remain another challenge for practical implementation of autonomous vehicle systems. In past, there have been scenarios where conventional HVs have also experienced cyberattacks. If even the conventional cars are vulnerable to cyberattacks, the risk of being vulnerable to cyberattacks is much higher for AVs [15]. According to the literature, cyberattacks may target infrastructure sign, machine vision, GPS (Global Positioning System), in-vehicle devices, acoustic sensors, RADAR, LiDAR and other in-vehicle sensors, odometric sensors (accelerometers and gyroscope etc.), electronic devices and maps [21]. Considering this, more robust security protocols need to be developed to protect the cloud-based communication system of car so that the risk of cyberattacks can be minimized [17].

2.9 Standards and regulation

With all major companies competing hard to launch fully autonomous vehicles, it is highly important to develop standards and regulation for correct and safe use of this technology. There are some recently proposed regulations e.g., for automated lane keeping systems [22], for safety for evaluation of autonomous products [23], and for safety of the intended functionality [24]. These standards, however, do not address several issues such as, sensors and machine learning etc. [15, 17, 18]. Table 1 summarizes some of the main standards and regulation developed related to safety, testing, connectivity, cybersecurity, design, verification, and validation.

Standard/Regulation	Description	Year
ANSI/ITSDF B56.5–2012	Safety Standard for Driverless	2012
ISO/TC 22/SC 33	Vehicle dynamics, chassis components and driving automation systems testing	2014
ASTM Committee F45	Committee on Driverless Automatic Guided Industrial Vehicles	2014
Code of Practice for Testing (UK)	Legal requirements for conducting public trials of automated vehicle technologies and service in the UK	2015
ISO 16787:2017	Intelligent transport systems — Assisted Parking System (APS) — Performance requirements and test procedure	2017
IEEE 1609.2a-2017	Standard for Wireless Access in Vehicular Environments	2017
Automated and Electric Vehicles Act 2018 (UK)	Rules on safe use of automated vehicles on GB roads	2018
ISO 21717:2018	Intelligent transport systems — Partially Automated In-Lane Driving Systems (PADS) — Performance requirements and test procedures	2018
IEEE 1609.2b-2019	IEEE standard for wireless access to vehicular environment	2019
ISO/PAS 21448:2019	Road vehicles — Safety of the intended functionality	2019
PAS 1880:2020	Guidelines for developing and assessing control systems for automated vehicles	2020
ISO 21202:2020	Intelligent transport systems — Partially automated lane change systems (PALS) — Functional / operational requirements and test procedures	2020
ISO/TR 4804:2020	Road vehicles — Safety and cybersecurity for automated driving systems — Design, verification, and validation	2020
UN Regulation No. 157 - Automated Lane Keeping Systems (ALKS)	System for performing dynamic driving tasks under certain conditions	2021
ISO 22737:2021	Intelligent transport systems — Low-speed automated driving (LSAD) systems for predefined routes — Performance requirements, system requirements and performance test procedures	2021
ISO 21448:2022 (Revised Version)	Road vehicles — Safety of the intended functionality	2022

Table 1.

Summary of standards and regulations.

3. Solutions and strategies

Although the challenges seem to be massive, researchers and developers are working hard towards the development of solutions to overcome these challenges and to make it possible for AVs to hit the roads soon. This section covers the strategies and solutions suggested in literature and the latest technologies developed that can eliminate or reduce the challenges that once seemed impossible to overcome.

3.1 Emerging technologies and potential solutions

The emerging technologies for autonomous vehicles are edge computing, software defined networking (SDN), network function virtualization (NFV), vehicular cloud computing (VCC), and named data networking (NDN). Their role in AVs is as follows [25]. Edge Computing helps to improve storage and provides real-time data processing, thus results in rapid decision-making. SDN offers flexibility and scalability to vehicular networks. In the same way, VCC helps in efficient road traffic management, dynamic traffic light management as well as improvement in road safety by instantly using the vehicular resources. NFV helps to improve efficiency and allows network functions to be distributed. NDN fixes networking issues such as those related to IP architecture and provides secure data sharing among AVs. These technologies clearly have a potential to improve connectivity, memory, and efficiency in autonomous cars.

Moreover, quantum cryptography and block chain enabled security algorithms can help to overcome cybersecurity issues related to sensors. AI and deep learning-based solutions can improve real-time data analysis and complex dataset management. Currently used radar technology lacks collision detection and collision avoidance algorithms. Also, when multiple radar sensors operate in overlapping frequency bands in the same vicinity, interference occurs. Therefore, radar interference management is needed with improved identification and classification of obstacles and enhanced algorithms [25]. All these technologies combined with high-speed wireless networks such as 5G and 6G can help to reduce and eliminate several challenges related to the technology, functionality, efficiency, connectivity, and security of autonomous vehicles.

Safety and security can be improved with the help of Responsibility-Sensitive Safety (RSS) framework. It is a safety standard developed by Intel that employs reinforcement learning technique [26, 27]. In the same way, V2X (vehicle to everything) protocols help autonomous vehicles communicate with their surroundings such as, vehicle-to-infrastructure (V2I) communication and vehicle-to-vehicle (V2V) communication. V2I helps to exchange data with the surrounding infrastructure such as speed limit, signs, and traffic lights, whereas V2V helps to communicate with other vehicles that results in collision avoidance and safer operations, especially during unexpected traffic situations and congestion [26].

4. Conclusion

The idea of autonomous vehicles has been around for years, but it seems to be turning into reality in the recent years with companies making huge investments and technology such as AI doing wonders. There are, however, several challenges and barriers that need to be accounted for, and a deeper insight is required to work towards this goal. This chapter highlighted the current barriers and challenges that must be considered and accounted for so that AVs and HVs can practically and safely share the roads. It also discussed some of the emerging technologies and possible solutions and strategies that can help overcome these challenges in order to revolutionize the autonomous vehicle technology and current transportation system.