Annotated BibliographyMuhammad Hamza FarrukhBag, S. (2017).
Deep learning localization for self-driving cars (Order No. 10259012). Available from ProQuest Dissertations & Theses Global. (1880508671). Retrieved from https://search-proquest-com.
ezproxy.lib.purdue.edu/docview/1880508671?accountid=13360There’s many challenges engineers are facing in developing a true self-driving car and one of them is of localization – ability of a car to accurately identify itself in the world. Using the Global Positioning System (GPS) for such a problem has been common industry practice however it isn’t available everywhere and not all the time. Because of this reason, Bag conducted experiments into the possibility of using visual odometry in conjunction with deep learning (subset of machine learning – using huge amounts of data to train computer models) to either improve or replace traditional GPS localization.
He concluded that using visual data can indeed help improve localization of self-driving cars especially when used with GPS. This source was helpful for my assignment because it showed how new inventions can improve technology that’s been in use for decades for everything from personal to commercial use. Bag was very thorough in explaining his methods and findings and I think this gives creditability to the author because with this much detail his experiments and results can be reproduced with relative ease.
Furthermore, this study is a motivation for someone like me who hopes to research on self-driving cars in the future because it shows room for improvement in the field.Daily, M., Medasani, S., Behringer, R., ; Trivedi, M. (2017). Self-Driving Cars. Computer,50(12), 18-23.
doi:10.1109/mc.2017.4451204The authors of this article analyze the ongoing developments in the field of self-driving cars in Asian, Europe, United States and academia. They discuss the technical implications of government policy on self-driving cars, for example how Asian governments are focused on the reduction of accident rates and pollution.
Finally, they talk about how a cloud platform for these self-driving cars will be revolutionary.Daily, Medasani, Behringer and Trivedi talk about the current status of self-driving cars in different regions of the globe and that is important for my assignment since it shows the audience how widespread this technology is. The main strength of this article is that the authors not only talk about what the current developments are in the specified regions but also what they were in the past and what they might be in the future. The credentials of the authors, all of them having worked with autonomous systems, give creditability to what they talked about in this article.
Gadepally, V. N. (2013). Estimation of driver behavior for autonomous vehicle applications (Order No. 3671348).
Available from Dissertations ; Theses @ CIC Institutions; ProQuest Dissertations ; Theses Global. (1647127404). Retrieved from https://search-proquest-com.ezproxy.lib.purdue.
edu/docview/1647127404?accountid=13360Self-driving cars might eventually replace all human driven cars but until that happens, self-driving cars will have to coexist with human driven cars on the road. This scenario brings its own set of challenges as the self-driving cars will have to consider the relative ‘unpredictability’ of humans on the road especially for collision avoidance systems and this is what Gadepally set out to research in this paper. The author proposed an architecture which consists of a Hybrid State System (a system with both discrete and continuous states) and Hidden Markov Models (a type of graphical models). This architecture uses machine learning along with pattern recognition and tests have shown that it will allow self-driving cars to not only estimate driver behavior but also predict it.
This paper was more useful than I anticipated because of how varied, in terms of complexity, its content was. It has a blend of topics ranging from some that I was already familiar with, some that I am currently studying and some that I have never heard of. This range of topics makes this source unique from the other ones I’ve selected.Monticello, M.
(2016). The State of the Self-Driving Car. Consumer Reports, 81(5), 44-50.Monticello answers general questions in this article like whether a self-driving car can be bought right now, or will self-driving cars make the roads safer? Moreover, the author introduces alternatives to cars which might be available in the future like “Autonomous Aerial Vehicles”. Monticello also explains why we’ll have to wait for “a very long time” before self-driving cars become a norm everywhere and how we’ll start using autonomous taxis within cities long before we’re able to buy one for ourselves.This source answers a lot of questions which the general population will have regarding self-driving cars. Every question that the author answers in this article is supported by a quote from an industry expert and that differentiates it from other articles. This article also explains all the terms it uses in layman words which makes it easier for a large audience to understand what is being conveyed.
Nyholm, S., ; Smids, J. (2016). The Ethics of Accident-Algorithms for Self-Driving Cars: An Applied Trolley Problem? Ethical Theory ; Moral Practice, 19(5), 1275-1290.
doi:10.1007/s10677-016-9745-2In this article, the authors discuss the ethical implications of accident-algorithms in self-driving cars by comparing this dilemma to the trolley problem (a thought experiment in which any decision a person makes results in the loss of life). They discuss how the people programming these cars will have to make these tough decisions which will end up determining the moral and legal responsibilities of the accidents. Furthermore, the authors call a machine’s accuracy into question when risks and uncertainties are added into the equation. This source is very useful for my assignment as it discusses the questions which, I think, are the most important in the field of self-driving cars. It is also the only source I’ve found that considers the ethical complications of this new, revolutionary technology.
The background of the authors, in philosophy and ethics, gives credibility to the research conducted in this article. Samuelson, R. J. (2017, September 24).
Opinion | Driverless cars may be appealing. But they could be used against us. Retrieved February 16, 2018, from https://www.washingtonpost.com/opinions/driverless-cars-may-be-appealing-but-they-could-be-used-against-us/2017/09/24/f5078cb2-9f9f-11e7-8ea1-ed975285475e_story.html?utm_term=.61ee17e8baeaIn this article, Samuelson writes against the development of self-driving cars without any care for unforeseeable consequences. He mentions how people in the first place might not even buy self-driving cars either because people enjoy driving or because of the cost of self-driving cars compared to regular ones.
A more serious issue he brings up is that of cyberwarfare. It is not hard to imagine that a security flaw in the future might end up completely crippling an entire city’s self-driving transport system.This is the only article I found that is so cautious with this new technology.
However, the author has valid reasons to support his stance. Because of all these reasons, I think this was one of the most useful articles I found for this assignment.Teoh, E. R., & Kidd, D. G. (2017).
Rage against the machine? Googles self-driving cars versus human drivers. Journal of Safety Research, 63, 57-60. doi:10.1016/j.jsr.2017.08.
008Self-driving cars are not yet perfect, and they probably won’t be for a while but how do they compare to human drivers? Last year, Teoh and Kidd conducted research which compared the accident rates of self-driving cars to human drivers. The authors concentrated on Google’s self-driving cars because they have been under tests on public roads since 2009, much longer than any other competitor. Teoh and Kidd found that Google’s cars had much lower accident rates than human drivers and Google’s cars were almost always rear-ended by a traditional human-driven vehicle. The data collected was from 2009 to 2015 and during that period there was only one accident where the Google car shared responsibility.
The analyzation of statistical data to support the relative safety of self-driving cars is something that will be useful supporting evidence for my presentation. The research’s main strength is the reliance on data from public records that everyone has access to. Teoh and Kidd have both done research related to vehicular safety in the past and work for government safety agencies. These credentials give authenticity to the research they’ve conducted in this paper.Wright, A. (2011). Automotive autonomy. Communications of the ACM, 54(7), 16-18.
doi:10.1145/1965724.1965731In this article, Wright talks about the Google self-driving car project and how it came to a reality after the Stanford AI Laboratory’s robot car won a competition held by a U.S. defense agency. This competition involved a car driving 125 miles in a desert without any human assistance. Another project mentioned in the article is the European Union-sponsored SARTRE. This project consists of cars driving in platoons with only the lead car being driven by a human.
This article provides a background into the most tested and, arguably, most popular self-driving car project ever which makes it unique. It is also the only article which described an alternative project which might become a reality before self-driving cars – SARTRE. This source is from a magazine and that makes it much easier to read compared to the other ones I’ve chosen.Yu, S., Jiefeng, C., & Qi, L. (2017).
The Effects of Self-Driving Vehicles on Traffic Capacity. UMAP Journal, 38(3), 323-347.The authors of this article analyze the effects of self-driving vehicles on traffic capacity by extending the traditional cellular automata used for traffic analysis. By running simulations, they found that an increase in self-driving cars on the highway increases the average speed of traffic flow. They also investigate the flaws in self-driving systems that can negatively impact the traffic capacity.The use of computer simulations to answer real world questions regarding the benefits of self-driving cars is crucial to my assignment as I’m in favor of such technology replacing traditional vehicles in the future.
The extensive use of mathematical models in this research paper is its strength. Furthermore, as the authors are from the computer science and engineering department of a university, their credibility in this topic is established.Zakharenko, R. (2016). Self-driving cars will change cities.
Regional Science and Urban Economics, 61, 26-37.Self-driving cars will have a huge impact on urban areas and might end up reshaping cities entirely. Zakharenko researched into this possibility and found that the demand for daytime parking in the downtown of cities will get shifted to outer edges of the city.
This will cause an increase in the density of economic activity in downtowns. Furthermore, rent will increase in inner cities and decrease in outer cities by about 30-40% each.I had not thought about how our cities might change because of self-driving cars and so this source brought up a great point that I had overlooked. The author has thoroughly explained his mathematical equations and workings, and this allows his work to be reproduced. Finally, since this paper was published in a reputable journal like the “Regional Science and Urban Economics”, it further increases the author’s credibility.