An Autonomous Future | From the Archives, April 2015

From the Archives is a series that pulls from a collection of research papers written for a high school STEM program between the years 2013-2015. These papers pushed students to investigate a subject, formulate questions and further developments upon the matter, and ultimately expand knowledge through some form of experimentation and execution. While much of the information and personal opinions presented have changed since the time of writing, I wanted to publicly document my thinking and explorations from that time.


AutoArt1.001.png

Background

As the intelligence and efficiency of modern technology has progressed, people, specifically users, have had to do less in order to accomplish more. Take, for example, digital personal assistants. Before the emergence of technologies like Siri or Google Now, people traditionally wrote to-do items on a piece of paper that they consulted frequently. The creation of smartphones and applications such as Notes and Reminders made things simpler by making lists more accessible and one less thing to carry with you. However, to-do lists and the efficiency of a person arguably reached their peak when lists were able to be created by telling a built-in digital assistant, such as Siri, to set a reminder based on when a certain location or time is reached. In short, people are able to accomplish more because the technology they use daily is able to do things for them, ultimately freeing time and increasing efficiency and sociability.

With this mindset, the next wave of digital automation is coming, this time in the car. Self-driving cars with an autopilot function are being developed and prototyped at an alarming rate, with fully developed technology hitting the markets in nearly five years [13]. While the concept of autonomous vehicles was discussed heavily in the past, as well as the creation of rough self-driving cars laden with video cameras to assist in navigation, the first modern 21st century self-driving car was that was released on the streets and could carry passengers was a fleet of Toyota Priuses modified by Google [17]. To successfully maneuver through busy streets while obeying all traffic laws, these cars use data from Google’s Street View mapping technology, as well as many cameras attached to the vehicle. In all, the seven existing Google cars have driven almost 300,000 miles since their creation with only minor human intervention and without any accidents [7]. Given that this experiment was run in 2012 and enjoyed such a high level of success, it is only natural that more extensive testing and prospects were initiated in the years that followed.

In 2015 alone, two companies and their respective vehicles have headlined the autonomous vehicle topic: Tesla and Mercedes. On October 9, 2014, Tesla CEO Elon Musk announced that the famous Model S vehicle would be shipped standard with a new autopilot function. The Tesla Model S is a vehicle already renowned for including luxury features with an  efficient, all-electric power system. The vehicle requires no gas to run and has a range over 300 miles per charge while driving 55 miles per hour; it takes about an hour to charge the battery depending on speed and miles traveled [19]. The Tesla Model S vehicles that ship with autopilot use “a forward radar, 12 long range ultrasonic sensors positioned to sense 16 feet around the car in every direction at all speeds, a forward looking camera, and a high precision, digitally controlled electric assist braking system” to safely and adequately control the technology [16]. While there is no official early word as to the quality of this feature yet, some reporters have attested to trying the autopilot out in a small capacity [9]. As of March 2015, Tesla has stated that these new vehicles should enter the market by June of the same year [4].

Mercedes' entry into the autopilot game is much newer, with their F 015 Luxury vehicle being shown off at the Consumer Electronics Show earlier this year [21]. This vehicle’s exterior and interior resembles that of a car from the future (release is slated for sometime in the year 2030) and provides insight into how a self-driving car should be designed. The exterior is outfitted with usual autonomous car gear, including cameras and sensors so the vehicle can understand its surroundings better; however, the F 015 also has a multitude of LED lights on the front and back that allow the vehicle to interact with surrounding pedestrians and drivers [8]. The LEDs on the rear of the vehicle can spell words like “slow” and “stop” to alert drivers behind them of traffic situations, and also project a moving walkway onto the ground and encourage pedestrians to cross the street. Such small additions to the vehicle give it a certain personality and friendliness that might be lost with the introduction of a software driver, which may be essential in helping people accept these new vehicles in the long run. 

Connecting the Dots

Being a soon-to-be lone driver soon, cars are high relevant to not only myself, but also the entire teenage population, who are prospective new car buyers. Or at least, this age group was the target market for vehicle sales. According to an article from the Chicago Tribune written in 1990 a poll showed that teens were a major driving force in the car market, with teen employment and earning, and thereby spending, being very high at the time [10]. However, the materials objects valued by teens have changed with the newer, current generation that I am a part of. Evidence from a 2012 article states that teens are “not that interested in buying a set of wheels…[or] even particularly keen on driving” [3]. Two major statistics are presented here: first, the average number of vehicle miles traveled yearly by 16 to 34-year-olds in the U.S. decreased by 23 percent between 2001 and 2009. Second, the number of 14 to 34-year-olds rose from 21 percent in 2000 to 26 percent in 2010. These people are instead using more public transportation, bikes, and even walking in an effort to not only save money, but also help out the environment. This economically and environmentally friendly mindset is one that I, as well as other teens, possess. Furthermore, teens are opting to now communicate via SMS or social media rather than meet in person, and shop online rather than drive to an actual store, thus reducing their need for a car. 

All of the evidence presented above may suggest that there is a dwindling venue for both myself and my peers to connect with the future usage of cars; however, I believe that while teenagers are no longer a market for normal cars, we are the optimal market for self-driving cars. The autopilot feature present in these vehicles can make our lives more efficient and allow me to be even more connected to my friends at the same time. For example, texting and driving is a huge issue - in 2012, 3,328 people were killed in texting/driving related accidents, and any risk for such accidents is increased 23 times by this action [2]. These numbers are alarming, but communication via smartphones is exponentially growing, and both I and those of a similar age group would rather give up driving than the ability to use our smartphones [5]. By purchasing an autonomous car, I would be able to both use my phone safely to check social media and text my friends, and even finish up a presentation for school, while arriving to my destination in an efficient manner with the added horsepower of a car.

Another use case is road trips: my friends and I are planning on embarking on a road trip after graduating. With a traditional vehicle, one of us would have to drive the long distance, incurring physical and mental fatigue, and most of the focus would have to be placed on the road. However, an autonomous car like the Mercedes F 015, which features front seats that rotate to face the rear can not only eliminate the stresses created by driving long distances, but also provide an opportunity to socialize within the cabin and productively plan the trip ahead. In recent years, purchasing and driving a car has personally become less and less interesting to me because of the valuable time it removes from both my productive and social life, but self-driving cars have sparked a new interest in me, and will most likely do the same with my peers. 

Speedbumps

The ability for people to reach their ultimate destination in a timely manner without having to operate the vehicle is not a new idea, for chauffeurs have been doing this for years. However, the idea that a software within the car is driving completely changes the game, and also raises serious roadblocks with the removal of the human element. Two major elements that software lacks are emotion and morality. A current vehicular autopilot system allows the passenger or would-be driver to regain control of the vehicle at any time: such a feature can be both helpful and dangerous. If the autopilot was unable to sense an object laying in the street but the “driver” could avoid it, then it is good that the vehicle can be manually controlled again; however, if a user was under the influence of alcohol and wanted to gain control of his self-driving car, there would be nothing to stop the user from doing so, at which point such privileges become dangerous.

Giving the software full control of the vehicle would stop any unfit drivers from taking control of the vehicle, but would leave the fate of both the passenger and those in the immediate environment prone to the “judgement” of the autopilot. This can be very dangerous if the autopilot cannot detect certain objects, as mentioned above. What happens if an autopilot software is imbued with a sense of emotion and a moral compass and is able to choose whether or not to give a driver control? With this additional factor now in play with autopilot software, the car itself can decide whether it needs the guidance of a driver or whether it is safer to retain control of the vehicle. However, if the autopilot is able to think, it may be relabelled by some as an artificial intelligence, and thereby accompanied by the various social stigmas attached to that term.

While once a thing of science fiction and depicted as a hazard in various stories in the past, such as the film I, Robot, and even more recently with Avengers: Age of Ultron, artificial intelligence is slowly emerging as a real technology in our lives. Artificial intelligence is defined as “the theory and development of computer systems able to perform tasks that normally require human intelligence, such as visual perception [and] decision-making,” a description that very well fits the thinking autopilot mentioned above [1]. AI is already a major point of contention with the innovators of the world; fear about the rise of AI has gained traction within circles of those with the credentials to support their apprehensions. At the start of 2015, an open letter was created that called for safety measures to be initiated against the AI threat, signed by aficionados from Stanford, Harvard, Google, and even notable figures such as Stephen Hawking and Elon Musk [15]. Such actions by Musk and others suggest that while the autopilot capabilities of the Tesla vehicles will continue to evolve for the better, they probably will not evolve into an artificial intelligence. That being said, creating and implementing an artificial intelligence may be a necessary step in the development of these vehicles, but present yet another hurdle when attempting to successfully legalize and introduce them. 

Rules of the Road

Legalization is a clear issue that autonomous cars and manufacturers will face. Fortunately, it seems to be a common theme that in the history of technological advancements, be it in the car or for personal entertainment, laws were inevitably broken during the inception and adoption of these innovations. One such innovation that is now commonplace inside of a vehicle that is used heavily by both the driver and the passengers in a car: the media control area. In the past, this hub was simply a small strip that displayed the time and radio channel and was flanked by various dials. In the past half-decade though, such outdated interfaces have been taken over my large, tablet-sized LCD panels that are in many cases touch sensitive and provide access to the weather, the music and text messages on a person’s phone, and even the internet. Such an interface doesn't sound very safe, so surely there must be laws restricting the installation and use of these displays.

Unsurprisingly, there are laws that do prohibit these devices in 45 of the 50 United States. For this example, the regulations in Colorado and California will be observed: Colorado for its direct relevance to the current location, and California because of Silicon Valley’s screen-heavy presence. According to the Consumer Electronics Association, Colorado citizens “18 years of age or older shall not…[engage] in text messaging or other similar forms of manual data entry or transmission while operating a motor vehicle” [14]. California has similar restrictions, yet oddly enough adds that “any person [is prohibited] from driving a motor vehicle if a video monitor, or a video screen or any other similar device that displays a video signal is operating and is located forward of the driver's seat or is visible to the driver.” Both Colorado and California residents, as well as citizens of the other states with similar laws, are breaking such laws every day simply by starting their engines and driving because they have access to and utilize the content-rich displays that are placed in the center console of the vehicle. Despite this, it is uncommon to hear of large numbers of drivers being penalized for violating this law. Why? Such features became nearly ubiquitous in vehicles, meaning that it isn’t feasible for officers to hunt down every person driving safely yet still using an LCD media console. The story changes if an accident is caused because of these devices, yet such incidents are so rare that cars with these features are simply now the accepted norm.

The same concept could theoretically be applied to the “legalization” or simple acceptability of self-driving cars. There will always be people that oppose the introduction of these vehicles for any number of reasons, some valid, others unfounded. But if enough car manufacturers create autonomous vehicles (that do have the option for drivers to regain control) and begin selling them to consumers, who in turn use these vehicles in a safe manner, laws against such vehicles will eventually be enforced less heavily, to the point where they simply become the accepted norm, just as LCD media control hubs did years ago.

Experimentation

Given that autonomous vehicles are not yet present and readily testable by the general public, it was obviously not feasible to experiment with these vehicles to gain further understanding. However, one thing that could be tested was the hypothetical accuracy of the sensors that are used to scan the environment around the autonomous vehicle for obstacles. While not the same sensor, the motion detector I found at the Oregon Museum of Science and Industry’s Physics Lab was nonetheless powerful and helped me come to many qualitative conclusions about the accuracy, or lack thereof, one of the sensors on a self-driving car might possess. 

The objective of the lab involving the sensor was simple: a ball and motion sensor were placed within a glass cube, as seen by the configuration in Figure 1.4. The ball had to be successfully retrieved without turning on a light that would be triggered by any motion captured by the sensor. After trying out many techniques, I deduced two primary observations. The motion sensor recorded information based on a less-than acceptable speed within the surroundings, and an interference high enough on the y-axis of the environment. The speed aspect is simple enough to spot and understand: if an object, in this case my hand, is moving quickly, then the motion sensor will easily pick up its existence. However, the slower my hand becomes, the less accurate the sensor becomes, meaning that my hand is essentially invisible to the sensor if I am inching across the plain. The y-axis component is more complex, yet still follows a pattern. While attempting to retrieve the ball, the sensor immediately detected my hand if it was higher up from the ground. Yet when my hand dipped past a certain point, up to where I was level with the bottom of the cube, I could slip past the sensor undetected. Complexity arose when I tried to correlate these two factors, mainly because there is no connection. The sensor isn’t more accurate if my hand is higher up and moves faster, nor is it less accurate if I move my hand very slowly while it is resting on the bottom of the cube. 

If these observations are applied to the sensors used on an autonomous vehicle, various situations can be created in which the car would and would not perform well. The self-driving car would be able to operate safely in a busy street: the pedestrians are tall enough so that they are high enough on the y-axis, and move at a reasonable pace that can be detected. Surrounding cars are on the same level of the autonomous car and its sensors, and also moves at mostly the same speed as the car, reducing any issues here. The self-driving car may not be as accurate and safe in a neighborhood setting. Here, small children may be using chalk on the road, and could be too low for the sensors to detect them; however, the speed of a running child or a pet may be enough to give the sensors the benefit of doubt. The sensors are most likely to fail, though, if a turtle was crossing a neighborhood road that was otherwise empty, yet the chances of this situation ever occurring are so slim that they may never be realistically considered when determining the safety of the vehicle. But without a driver behind a wheel, maybe the standard of safety has to be higher than necessary for these machines to succeed.


References

[1] "Artificial Intelligence Definition." Merriam-Webster. Merriam-Webster, n.d. Web. 28 Mar. 2015.

[2] "The Dangers of Texting While Driving." The Dangers of Texting While Driving. Federal Communications Commission, n.d. Web. 28 Mar. 2015.

[3] Florida, Richard. "Why Young Americans Are Driving So Much Less Than Their Parents." CityLab. The Atlantic Monthly Group, 10 Apr. 2012. Web. 28 Mar. 2015.

[4] Hardawar, Devindra. "Tesla's Model S Is Getting a Self-driving 'autopilot' Mode in Three Months." Engadget. AOL Inc., 19 Mar. 2015. Web. 28 Mar. 2015.

[5] Kingkade, Tyler. "Millennials Would Rather Ditch Car Than Smartphone Or Computer: Zipcar Survey." The Huffington Post. TheHuffingtonPost.com, Inc., 1 Mar. 2013. Web. 28 Mar. 2015.

[6] Kumparak, Greg. "The Tesla Model S Will Get “Autopilot” Mode In About Three Months." TechCrunch. AOL Inc., 19 Mar. 2015. Web. 28 Mar. 2015.

[7] Lardinois, Frederic. "Google’s Self-Driving Cars Complete 300K Miles Without Accident, Deemed Ready For Commuting." TechCrunch. AOL Inc., 7 Aug. 2012. Web. 28 Mar. 2015.

[8] Lee, Nicole. "Riding in Mercedes' Luxurious, Self-driving Car of the Future." Engadget. AOL Inc., 25 Mar. 2015. Web. 28 Mar. 2015.

[9] Lowensohn, Josh. "My Lap of Terror in the Tesla D." The Verge. Vox Media, Inc., 10 Oct. 2014. Web. 28 Mar. 2015.

[10] Mateja, Jim. "Poll Shows Teens A Driving Force In Car Market." Chicago Tribune. Tony W. Hunter, 18 Mar. 1990. Web. 28 Mar. 2015.

[11] "The Mercedes-Benz F 015 Luxury in Motion." The MercedesBenz F 015 Luxury in Motion. Daimler AG, n.d. Web. 28 Mar. 2015.

[12] Miller, Claire Cain. "When Driverless Cars Break the Law." The New York Times. The New York Times, 13 May 2014. Web. 29 Mar. 2015.

[13] Sathe, Gopal. "Technology for Autonomous Cars Just 5 Years Away but Policy May Take Longer: Ford." NDTV Gadgets. NDTV Convergence Limited, 4 Mar. 2015. Web. 28 Mar. 2015.

[14] "State Laws for Electronics Use in the Car." CEA. Consumer Electronics Association, n.d. Web. 29 Mar. 2015.

[15] Strange, Adario. "Elon Musk, Stephen Hawking Warn of Artificial Intelligence Dangers." Mashable. Pete Cashmore, 13 Jan. 2015. Web. 28 Mar. 2015.

[16] The Tesla Motors Team. "Dual Motor Model S and Autopilot." Tesla Motors. Tesla Motors, 10 Oct. 2014. Web. 28 Mar. 2015.

[17] Vanderbilt, Tom. "Autonomous Cars Through the Ages | WIRED." Wired.com. Conde Nast Digital, 6 Feb. 2012. Web. 28 Mar. 2015.

[18] Weber, Marc. "Where To? A History of Autonomous Vehicles." Computer History. Computer History Museum, 3 Apr. 2014. Web. 28 Mar. 2015.

[19] "Your Questions Answered." Tesla Motors. Tesla Motors, n.d. Web. 28 Mar. 2015.

[20] Ziegler, Chris, and Dieter Bohn. "Live with Elon Musk in Detroit." The Verge. Vox Media, Inc., 13 Jan. 2015. Web. 29 Mar. 2015.

[20] Ziegler, Chris. "The Mercedes-Benz F 015: This Is What Tomorrow's Self-driving Cars Look like." The Verge. Vox Media, Inc., 05 Jan. 2015. Web. 28 Mar. 2015.