The is the leading cause of accident death among

The purpose of this memo is to inform you of the safety benefits of autonomous vehicles and convince you that they should be in widespread use as soon as possible.  In 2015 there were 32,166 fatal crashes, and motor vehicle fatality is the leading cause of accident death among young adults (NHTSA,2016). While safer cars and better traffic engineering has improved the fatality rate among drivers, human error remains a deadly killer. The National Highway Traffic Safety Administration reports 80% of accidents are a result of distracted driving (NHTSA, 2016). These accidents are preventable, and the only sure precautionary measure is to take human drivers out of the equation.By providing autonomous vehicles with more exposure to various driving conditions, we can improve their safety and save more lives. This memo discusses how the machine learning algorithms which control the performance of autonomous vehicles are optimized by use on the roads. Today there are only a small number of autonomous vehicles on the roads due to public resistance and lack of proper legislation. I hope to convince you that encouraging widespread use of autonomous vehicles is best for the safety of your city. In addition to the improvement of public safety, I will discuss how autonomous vehicles will provide benefits to your city such as improved traffic flow, better space utilization, and decreased CO2 emissions.  I hope to show you of the merits of widespread use of autonomous vehicles so you can influence legislation encouraging their use.BackgroundAutonomous vehicles, more commonly referred to as self-driving cars, are robotically controlled vehicles that travel without a human driver. The technology for autonomous vehicles is fully developed and they are currently being tested in Pennsylvania by the company Uber. The concept of autonomous vehicles is not new as there have been efforts to create autonomous vehicles since the 1920s (“Phantom Auto…” 1926). Widespread acceptance by the public and supportive legislation are still lacking. Public Safety HazardTraffic accidents are caused almost exclusively by human error. The primary cause of fatal traffic accidents is distracted driving, including texting, eating, or smoking while driving, followed by drunk driving. Pedestrians outside the autonomous cars are safer as well as passengers. Eric Boerer, Bike Pittsburgh’s advocacy director, made this comment while collaborating with Uber to collect data on how autonomous vehicles interact with bicycles. “People did feel much more comfortable riding next to autonomous vehicles than they did next to human vehicles. I mean, autonomous vehicles, they don’t get angry, they don’t have road rage.” (Krauss, 2017). Human drivers are the number one public safety hazard, and one that can be removed from society. Liz Reid of NPR describes riding in one of Uber’s self-driving cars; “The car doesn’t do anything that might be remotely risky…when we come up behind a car that is parallel parking, our vehicle slows down and waits, rather than change lanes to go around it.” (Reid, 2016). This level of caution that ultimately saves lives cannot be forced upon human drivers. How Autonomous Vehicles LearnIt is not possible for programmers to predict every possible scenario an autonomous vehicle may come across while on the road. Thus, programmers use machine learning algorithms so the vehicles can program their own responses based on what has been successful in the past. Machine learning algorithms are computer programs that rely on input observations to build a model from examples, as opposed to following static instructions. This means that the car uses pattern matching with past scenarios to know how to respond to new scenarios.If in the past the car received positive feedback when breaking when a cat ran into the road, it will do the same if a dog runs into the road, as a simple example. The autonomous vehicle will make small changes to its algorithms and will either keep or discard those changes depending on the feedback it receives. This algorithm optimization process is why it is important to expose the autonomous vehicles to the roads as much as possible. While on the surface it seems unwise to put autonomous vehicles on the roads while they are still learning, ultimately, they prevent more traffic accidents than they will cause due to initial program errors.Implementation TimelineThe first graph represents a hypothetical scenario in which autonomous vehicles are introduced into the marketplace in 2035 and fully adopted in 2065. Assuming full safety optimization by 2040, a future with autonomous vehicles would have 0.6 million less fatalities compared to a future without (RAND, 2017). The second graph shows a scenario in which autonomous vehicles are introduced in 2020 and are optimized by 2035. In the second scenario the autonomous vehicles will have saved 1.1 million lives by 2070 (RAND, 2017). According to this projection, developed by researchers of RAND Corporation, putting autonomous vehicles into effect as little as 15 years earlier will prevent half a million unnecessary traffic deaths. One of the researchers commented “Waiting for the cars to perform flawlessly is a clear example of the perfect being the enemy of the good.” (RAND, 2017). Future Implications In addition to the improvement of public safety, the widespread use of autonomous vehicles will improve traffic flow, utilization of parking space, and decrease CO2 emissions. Traffic FlowIt is estimated that Americans living in urban areas spend almost 7 billion hours in traffic, waste 3.1 billion gallons of fuel and lose around $160 billion due to traffic congestion per year (Kearns, 2016). The fundamental cause of all traffic is slow human reaction time. When a light at an intersection turns green, each car accelerates one by one, and this discoordination limits the number of cars that can flow through the intersection. If all cars accelerated simultaneously as soon as a light turned green more cars could make it through the light. Coordination at this level is impossible with human drives but completely feasible with autonomous vehicles. Similarly, traffic that occurs in highways is caused by human errors such as a driver cutting across lanes and the driver behind him breaking too quickly. This causes a chain reaction that leads to the traffic slowdowns on the highway with seemingly no cause. The subsequent increase in commute time would be avoided by autonomous vehicles. Decrease in Greenhouse Gas EmissionsImproved traffic flow decreases the amount of time vehicles spend on the roadway and thus the amount of fuel consumed. They would also drastically reduce the amount of automobile manufacturing, which is an industry that has higher CO2 emissions than those produced by power plants (Kearns, 2016).Space UtilizationParking lots require space for people to get in and out of their cars. This space is unnecessary for autonomous cars as they can be stacked more closely to each other. It is estimated that when self-driving cars park themselves they require 15% less space than human drivers (Kearns, 2016).  This would result in significant savings for urban areas. ConclusionPublic acceptance and supportive legislation of self-driving cars is vital if the benefits of autonomous vehicles are going to be realized in your city. Distracted driving and human error cause countless unnecessary fatal traffic accidents. These deaths can be prevented by autonomous vehicles. Due to the way machine learning algorithms improve the safety of autonomous vehicles, the sooner autonomous vehicles are on the road the more lives that will be saved. I urge you to consider making changes to federal law impacting autonomous vehicles. The sooner their widespread use the more lives that can be saved.