Abstract
This white paper is about the ethics and safety of self-driving cars. It will explain what AI in self-driving cars is, what levels of self-driving cars there are and what the advantages and disadvantages are of having self-driving cars on the road
Author Keywords
Self-driving car; Artificial Intelligence; Ethics; Safety.
Introduction
We live in a world that is advancing every day. Companies such as Google, Facebook, Tesla and many others make use of AI and Machine Learning in their applications. For example Google with their Google Assistant and Tesla with a system called Autopilot (Tesla, Inc.). This white paper is about the last part: self-driving cars. Another example of a company that has been working on self-driving cars is Waymo, formerly known as the Google Self-Driving Car Project. Waymo One is a self-driving car service, where the user, if located in the USA, can request a self-driving car to pick them up and drop them off at their end destination (Waymo, LLC). The cars that Waymo use, can pick up people without a driver in the driver’s seat. But is it safe and ethical to give an AI complete control over a car?
The Six Levels of Driving Automation
According to the Society of Automotive Engineers (SAE) there are six levels of driving automation (SAE International, 2016). Level 0: No Automation, these vehicles are manually controlled. Level 1: Driver Assistance, these vehicles have assistance for steering or accelerating (cruise control) or both. Level 2: Partial Driving Automation, these vehicles can control both steering and accelerating/decelerating but still require a human to sit in the driver’s seat and monitor environment and tasks. Level 3: Conditional Driving Automation, these vehicles have “environmental detection” that allows them to make decisions such as accelerating past a slow-moving vehicle but still require a human to intervene if something goes wrong. Level 4: High Driving Automation, these cars can intervene if things go wrong and require in most circumstances no human interaction but humans can still manually override. Level 5: Full Driving Automation, these cars do not have a steering wheel or acceleration/braking pedals and do not require human attention.
Most of the cars that are being called self-driving cars are actually not self-driving. They are categorized as Level 2. But moving forward, we are likely going to see Level 5 cars in the near future.
The Benefits of Self-Driving Cars
Having self-driving cars on the road has several benefits. In the next paragraphs, we will discuss some benefits.
Reaction Times
In situations where accidents are inevitable, people may think it is the best idea to transfer complete control to the human behind the steering wheel. But there is a slight problem with that. It takes us a long time to switch our attention from the thing that we were doing to another thing that needs to happen. And if it occurs that some kind of accident is happening, we are already too late to intervene. So if we let computers handle the driving of the car, we can drastically decrease the number of traffic collisions due to the faster reaction time of autonomous system’s compared to humans according to Bimbraw, K. (2015), Nyholm, S. & Smids, J. (2016).
Sensors
According to Tesla, Inc and Ingle & Phute (2016), Tesla cars use 8 surround cameras that provide 360 degrees of visibility and have up to 250 meters of range around the car to monitor every move of every traffic contestant and traffic signs around the car. With the information that the car gets from the sensors, the car can predict how fast it needs to go, where all of the other cars are going, if there is a pedestrian crossing and how close it can drive to the car in front of it (Godsmark, P., 2013).
The Disadvantages of Self-Driving Cars
So far, we saw some of the benefits of self-driving cars on the road. But there are also some disadvantages of self-driving cars that prevent them to be deployed as Level 4 or Level 5 cars. That is what we are going to talk about in the next paragraphs.
