Although, the automation technology is now intriguing and popular among the public, Google was the first company to produce and test the vehicles on public roads without many regulations. People had no prior knowledge of the technology that engineers had with the vehicles until the developments of them were announced to the people. Google hadn’t considered if the general public would want to purchase this new technology or even have the self-driving vehicles on the same public road as them. However, a recent study was conducted for the purpose of understanding the public’s opinion on their driving abilities compared to the current automation technology and the features that need to be implemented into the technology.
The study was an online sample that was conducted to examine how U.S. drivers perceived their own driving abilities, and understand the desired level of safety that needs to be implemented into the self-driving vehicles before they feel comfortable in riding in one, purchasing one, and letting one drive on the same road as people (Nees, 2019, p. 61). In the study, it asked several questions, and required the participant to answer using a scale that represented a percentile that ranged from 1-99. The results showed that 80.9% of the participants perceived themselves as safer than the average driver (p. 63), 66.1% believed that they are safer than the current self-driving vehicles (p.64), and 68.1% of the sample desired a self-driving vehicle that is safer than they perceived themselves to be (p.64). Figure 1 demonstrates the results of the survey:
Fig. 1. Safety percentile (as compared to a human driver) desired by drivers before self-driving cars are allowed on public roads (top panel), before they would feel reasonably safe riding in a self-driving car (middle panel), and before they would buy a self-driving car (bottom panel). “Never” response indicated the participant would never support the proposition.
With the results of the study, it is clear that the majority of the public wants the self-driving vehicles to be safer than average drivers and the perceived above average drivers before they are comfortable with purchasing, riding, or allowing one to be on the same public road. It is now up to engineers to make proper decisions to satisfy the public in ensuring their safety when it comes to the self-driving vehicles.
With the introduction to the automation technology, there are different levels of the advancement with technology. There are 6 levels of automation that range from the vehicle having absolutely no driving technology to the vehicle having full driving automation. Currently in 2020, consumers can purchase any of the levels of the automation technology. In Table 1 below, it gives a description of the different levels of the automation technology:
Table 1:
Levels of automation technology in vehicles
| Level 0: no driving automation. The whole dynamic driving task (the lateral and longitudinal vehicle motion control, the detection of and response to objects and events) is performed by the human driver. | Level 3: conditional driving automation. The automated driving system performs the complete dynamic driving task under certain conditions. The user has to be able to take over the driving and respond appropriately to a request of the system to intervene. |
| Level 1: driver assistance. A driving automation system performs either the lateral or the longitudinal vehicle motion control under specific conditions; the human driver performs the remainder of the dynamic driving task. | Level 4: high driving automation. The automated driving system performs the complete dynamic driving task under certain conditions, without the expectation that a user will respond to a request to intervene. |
| Level 2: partial driving automation. Under specific conditions, the driving automation system is able to perform both the lateral and the longitudinal vehicle motion control. The human driver has to supervise the system and performs the remainder of the dynamic driving task (that is, detection of and response to objects and events). | Level 5: full driving automation. The automated driving system performs the complete dynamic driving task under all conditions, without the expectation that a user will respond to a request to intervene. |
*Vellinga, 2017, p. 849
Due to the high advancements of the automation technology, federal policymakers must find a way to regulate the testing of the vehicles without hindering the developments of the technology too much, but also keeping the public’s safety above all else. The original regulatory system for motor vehicles is the Highway Safety Act of 1966 and the National Traffic and Motor Vehicle Safety Act of 1966, which created the National Highway Safety Bureau, also known as NHTSA (Mathews, 2020, p.302). NHTSA are required to establish safety standards for vehicles, which are known as Federal Motor Vehicle Safety Standards (FMVSS). However, with the automation of vehicles, they don’t fit the description of the FMVSS, so the NHTSA must conduct rulemaking to amend or create new FMVSS, grant exemptions, or interpret FMVSS (p. 303). Although the NHTSA can exercise this authority, rulemaking is an extensive process, and it requires them to do deliberate research and adhere to the Administrative Procedures Act. In response to this process, NHTSA and the U.S. Department of Transportation issued a preliminary statement of policy and four guidance documents to inform interested parties of issues they see in the development of the automated driving system (ADS) (p. 305). NHTSA requested that companies developing ADSs voluntarily submit a Submit Assessment Letter detailing whether the ADS complies or fails to comply with the guidance areas of privacy, system safety and crashworthiness, or whether the guidance area is inapplicable to the ADS being developed (p. 305). All of the documents are currently published to the entities that are developing the ADSs, and in 2017, two bills, the Self Drive Act and the AV Start Act, were passed, which contains provisions that preempt certain state, local laws and regulations that relate to automated vehicles (pp. 306-307).
For automation technology to be improved, testing the self-driving cars in actual traffic will be required for the development of encoding situational information into the vehicle; however, the testing needs to be regulated enough to ensure public safety, but not so much that overregulation occurs and inhibits data and therefore developments. Therefore, countries like the United Kingdom, and the Netherlands and states like California, Nevada, and Arizona have adopted regulations (Vellinga, p.852). In Division 16.6, Section 38750 of the California Vehicle code, companies must have a Manufacturer’s Testing Permit to test self-driving cars on public roads (Vellinga, p.825). For companies to obtain this Testing Permit, the vehicles must pass all requirements and testing that are set by the Department of Motor Vehicles to operate on any road or weather conditions (Vellinga, p.852). However, there are other states like Nevada, who further regulate testing by requiring companies with a Testing Permit to only test the self-driving cars on certain roads, and specific weather conditions (Vellinga, p.852). Due to these regulations, engineers of companies who produce self-driving vehicles must be more adept in the design of the automation vehicles for them to even test it on various public roads. Thus, by them testing their technology, they can encode the vehicles with more rational decisions when faced with various situations on the road (Vellinga, p.852-853).
Engineers improving the automated driving systems argue that overregulating the testing and production of the automation technology will hinder the developments of it or can even stop the process entirely. By policymakers and the NHSTA not having established and solidified regulations yet, and them having the authority to change the regulations, it could potentially harm the developments. Although, policymakers and the NHSTA do have the authority to amend or change a regulation, they give a prior notice to the engineers of the changes being made. Also, the changes to prior regulations aren’t too drastic that will hinder or completely halt the developments of the technology. Policymakers are assuring that guidelines are being followed to protect all parties involved, which are the engineers, car manufacturers, and the general public.
Although the future of the automation technology in vehicles seems to be promising, there have been many obstacles and tragedies that engineers and manufacturers have faced with automated vehicles. For example, in May 2016 in Florida, a car collided with a tractor trailer killing the driver, and though this seems to be a typical fatal accident, it was completely different (Vellinga, p. 848). The vehicle that was involved in the accident was a Tesla Model S, in which the vehicle was in an automated car feature called ‘Autopilot’ and the driver not paying attention to his environment lead him to his death. Another example is in March 2018, Uber had been testing its automated vehicles, until one day, Elaine Herzberg was killed by a self-driving vehicle, who was in autopilot mode, while she was crossing the street at because the vehicle failed to detect her, and the safety driver conducting the testing was distracted, and was unaware of the roadway (Matthews, p. 296). These accidents display the dangers of the automation technology when there is a defect in the programming and the unawareness of the potential drivers of the vehicles.
Unlike humans, who have moral principles in their mind, automated vehicles have to be programmed with an ethical decision-making process that must be able to react when it is faced with different scenarios where the accident is unavoidable. Nyholms and Smids posed two questions when it comes to this decision making process, which is “should autonomous vehicles be programmed to always minimize the number of deaths” or “ should they perhaps be programmed to save their passengers at all costs?” (2016, p. 1275). Both questions are valid because situations will arise where the vehicle must make the decision. A philosophical theory that was created when thinking of various ways to program the vehicles is called the trolley problem. The theory is if there is a runaway trolley and the only way to save 5 people on the tracks is to sacrifice one person (p. 1276). Recently philosophers started liking the idea of accident-algorithms compared to the trolley problems. Philosophers, Wendell Wallach and Colin Allen, wrote:
…could trolley cases be one of the first frontiers for artificial morality? Driverless systems put machines in the position of making split-second decisions that could have life or death implications. As the complexity [of the traffic] increase, the likelihood of dilemmas that are similar to the basic trolley case also goes up (Wallach and Allen 2009, 14).
With multiple writers, they believe that the problem with programming the automation technology for different accident-scenarios is like the trolley problem (p. 1276). This suggest that when programming the system with an ethical framework, it should focus on accidents that are considered risky and unavoidable. In Table 2, it summarizes the points that are made with the accident-algorithms and trolley problem:
Table 2
Accident-Algorithms and Trolley Problem for Autonomous Vehicles
| Accident-Algorithms | Trolley Problem | |
| 1a. Decision faced by: | Groups of individuals/multiple stakeholders | One single individual |
| 1b. Time-perspective | Prospective decision/ contingency planning | Immediate/ “here and now” |
| 1c. Numbers of considerations/situational feature that may be taken into account | Unlimited; unrestricted | Restricted to a small number of considerations; everything else bracketed |
| 2. Responsibility, moral and legal: | Both need to be taken into account | Both set aside; not taken into account |
| 3. Modality of knowledge, or epistemic situation: | A mix of risk-estimation and decision-making under uncertainty | Facts are stipulated to be both certain and known |
*Nyholm & Smids, 2016, p. 1287
Other questions that should also be asked are “who should have the authority/jurisdiction to the ethical decisions that are programmed into the autonomous vehicles, would it be the engineers developing the system, or the policymakers that regulate the technology?”, “should there be set guidelines for whoever makes the decision of the ethical framework of the autonomous vehicles?”, and “who should be held accountable for accidents when an autonomous vehicle is involved?” All these questions should be answered and resolved before car manufacturers mass produce these vehicles and consumers are able to purchase them. NJ Goodall makes a point that “since robots and humans have such a difference in ethics, the autonomous technology has to face greater challenges when making decisions on public roads” (Goodall, 2016, p. 58).
Engineers could argue that since they are developing and improving the automation technology, they should be the ones that have full control on the ethical framework that is programmed into the vehicles. However, even engineers, must follow regulations and laws that are set by the government for which they are under. The government should set a criterion of ethical guidelines by which the engineers have to work under those specific policies to produce and release their innovations with the autonomous vehicles.
Although, engineers and policymakers must make critical decisions when it comes to the autonomous vehicles because of the current disadvantages to the technology, the proven advantages and the different possibilities of improvements and developments to it will soon triumph. For example, with the use of autonomous vehicles, car accidents will significantly decrease, they can reduce congestion on roadways, it can become another source of transportation for the public, it can also create jobs for the public, etc. (Litman 2016). Those are a few possibilities with that autonomous vehicles can help resolve within countries that have high populations and a need for transportation.
A Genius Thought (JPB) 