Budny 3: 00 L07 Disclaimer

Budny 3:00

L07

THE CURRENT STATE OF DRIVING: AN INEFFICIENT, DANGEROUS, AND IMPRACTICAL WAY OF LIFE WITH ONE SOLUTION
Driving is a part of daily life for billions across the world and most people rely on someone to drive in some form or another. This is just an accepted way of life; however, autonomous driving offers a solution to its flagrant problems and some that many people might not even realize they have. Most people fear freak accidents like shootings, terrorism, and stabbings, but don’t think twice before hopping in the car. This is one of the most dangerous things people do daily. Motor vehicle accidents are the leading cause of unintentional injury deaths in America over the past 15 years according to the Center for Disease Control and were the leading cause of unintentional fatal injuries worldwide in 2000 as indicated by the World Health Organization [1][2]. Although motor vehicle deaths have decreased by an impressive 15% over the course of the past 10 years, the fact that it accounts for almost 25% of unintentional deaths annually is awful, especially when there is a solution [1].

There is only so much that can be done to minimize human error, however, the era of autonomous driving will end the need for countless unnecessary deaths worldwide. With the development and implementation of new sensors that can detect and evaluate road conditions within a fraction of a second, autonomous vehicles already have the potential to save hundreds of thousands of lives worldwide. This fact alone is enough to make me truly believe that this is something that should be invested in heavily, but there is a plethora of other benefits to go along with it that deserve autonomous driving to be a focal point of engineering in the world.

Autonomous cars are generally seen as some far-off technology of the future, but prototypes have been around since the 50’s. The first automated prototypes tested by GM used “pick-up coils” that could read the currents in wires embedded into the road and steer accordingly [3]. Although these vehicles could travel fairly large distances unmanned, these cars were not truly autonomous and were more akin to an RC car following a track. These vehicles were a step in the right direction, but they were not aware of the environment around them and were largely untested in real traffic conditions. Due to the limited scope of technology at the time however, most people thought it impossible to create a car that was capable of interacting and operating within real life conditions. Despite this, the first visually-based automated car was engineered in 1977 by Sadayuki Tsugawa and his colleagues [3]. The vehicle was equipped with two cameras that used analog computer technology to process visual ques. This prototype could only travel up to 18.6 mph and used the assistance of an elevated guiderail, but it revolutionized the driving thought process behind automation of motor vehicles and changed the perception of the realm of possibility for automation [3]. 17 years later VaMP with the use of two cameras processing 320 x 240 pixels built a car that could recognize road markings, understand its position in the lane, and the presence of other vehicles, effectively operating at 98.6% autonomy[3].

This was a huge step forward in the autonomous industry. These vehicles no longer had to follow along pre-meditated paths, but could process and interpret the environment around them and act accordingly. The innovation and success of these autonomous vehicles steered research away from cars guided by inductive signals in the ground and instead toward vision-based systems for truly autonomous driving. These experimental cars were the stepping stones that propelled the self-driving car industry to where it is today, and it is impossible to truly appreciate the autonomy of today without first understanding how far technologies have progressed since its first inception.
LIDAR: THE KEY TO AUTONOMOUS DRIVING
Autonomous cars have yet to be legalized for

operation without the presence of a driver inside the vehicle, however, research and technology related to autonomous driving has progressed rapidly. Advancements in every aspect of technology from cameras resolution to computer algorithms and sensors in conjunction with huge improvements in processing power has led to these incredibly complicated and amazing machines capable of driving completely uncontrolled.

Many technological developments have come together to achieve the success of modern day autonomous vehicles, but none was more key to the dynamic and precise interpretation of the environment necessary for safe automation than LiDAR scanners. Standing for light detection and ranging, LiDAR was created in the 1960’s for detecting submarines from aircrafts. LiDAR works by sending a laser beam into a mirror projecting it outward [4]. When the beam hits an object, it is reflected back to the mirror [4]. Measuring the interval of time between the beam being ejected and returning to the mirror, a simple calculation relating this time to the speed of light can determine the distance of that object from the sensor [4]. Sending thousands of these beams per second, it records the information on the distance of hundreds of thousands of these beams, however, this information means little without the sensor understanding its position relative to the surfaces its measuring [4]. Because of this, LIDAR is usually paired with GPS to determine geographical positioning relative to the world and an inertial measurement unit to record the exact positioning of the sensors. Taking in all this information computer algorithms can translate these distance into fixed points relative to the sensor creating a highly accurate 3-d image of the area it passes over at huge distances of up to 2000m [4].

This was a huge engineering achievement that provided a sensor unlike any others before it. As a computer engineer my specialty is combining aspects of hardware and software to produce products. This invention is the embodiment of what I hope to engineer in my ventures as a computer engineer. Taking the inputs of laser-shooting sensors and through computer algorithms creating a physical representation of the areas they touch sounds like something of Star Wars or some alien world, but these engineers were able to make it happen. Similar technologies such as radar and ultrasonic detection systems were not capable of producing anywhere near the same level of image accuracy due to the longer wavelengths of radio waves and ultrasonic sound waves [5]. This accuracy is what allows for LiDAR to provide an exact 3D monochromatic image of its surveyed area [4]. Furthermore, this precision is what is vital to the real-time decision-making of computer processors. The better the computer algorithm can understand its environment, the better it will be able to judge what it should do. This split-second decision-making is what can be the difference between life and death in accidents and is why the technology is being used in the vast majority of self-driving automobiles on the road today.

Despite its prominence in the current autonomous car industry, LiDAR as we know it today did not exist a decade ago and was not even implemented into cars until the mid-2000’s. When the winner of the 2005 Grand DARPA challenge—a multimillion dollar competition for American autonomous vehicles—implemented 5 LiDAR sensors in their winning vehicle, this revolutionized the formula for autonomous vehicles [3]. The obstacle with using the LiDAR sensors at the time was that they only recorded space in a 2-d plane and thus required tilting stages to record a segment of space and record a 3-d image [3]. This inspired the creation of a 3-d version of LiDAR the very next year [3]. Capable of scanning and interpreting the environment in 360 degrees, this scanner laid the foundation for all future 3-d LiDAR scanners and was recognized by the Smithsonian as a foundational breakthrough for autonomous driving [3]. Using technology and applying it to create a whole new purpose is exactly what engineering is all about. Ideas like this are what I hope to achieve as an engineer and are what is necessary to advance the world.

Current LiDAR systems can send millions of beams in 360 degrees by up to 90 degrees of elevation to create huge maps 3-d maps of the surveying area [6]. Using this map, with advanced computer software the car can predictably navigate the environment, detect objects like people and stop signs from hundreds of yards away, and even track speed and direction [6]. This was the combined effort of engineers from all aspects the practice. From the electrical engineers that designed the power grids in the system to be able to fire in all directions, to the computer engineers and programmers that wrote the programs to relate the perceived information to a physical model, this was an incredible feat of engineering that showcases human innovation.

This has already had a huge impact on the world of autonomous driving. The ability to detect and differentiate objects among its environment means that the car can prevent crashes that human reaction time and attentiveness wouldn’t have. 94% of crashes in the US involve human choice or error [7]. This could mean the end of preventable motor vehicle accidents for good. That’s 1.2 million lives saved every year and 20+ million car crash related injuries prevented [8]. Furthermore, the prevention of car crashes means more efficient travel, safety of mind, and no more problem of intoxicated drivers.
UBER’S AUTONOMOUS CAR: A GLIMPSE INTO THE FUTURE OF AUTONOMY
The company Uber currently uses LiDAR as an integral part of its auto-navigation. Utilizing a top-mounted LiDAR system to create a map of its surroundings, LiDAR modules in the front, rear, and sides of the car to detect objects in blind spots of the main module, 20 cameras to further look for obstacles, a colored camera to put the LiDAR map into color, and GPS to track the car’s position, Uber has created a vehicle that operates in the real world and serves to transport people every day [9]. Currently limited to the Pittsburgh area, Uber is providing the option to experience a driverless taxi. This could potentially pave the way for a whole new idea of travel. As technologies improve and prices drop, these taxis could become the main form of travel. Instead of making a 25-minute drive to work, you might decide that its simply more efficient to grab one of these driverless taxis and get some work done on the way over or relax and eat some breakfast. The average commute time in the US according to WNYC is 25.4 minutes so taking this new form of travel would mean effectively adding 50 minutes of productivity to the day and over the course of a month that’s over 17 hours of productivity gained [10]. Extrapolate this over the course of years with the entire population of drivers and the benefits are immense. If projections follow current vehicle technology trends, autonomous capable vehicles will dominate the majority of the automobile market by the 50’s and be mandated by the 60’s-80’s. Autonomous driving is here to stay, and I hope to be a part of it. The benefits are clear, and it is only a matter of time before they take over and the world will be changed forever.
SOURCES
[1] “Unintentional Injury Deaths.” Center for Disease Control. Accessed 10.25.17.

https://webappa.cdc.gov/cgi-bin/broker.exe

[2] “Injury: A leading cause of the global burden of disease, 2000.” World Health Organization. Accessed 10.25.17.

http://www.who.int/violence_injury_prevention/publications/other_injury/injury/en/

[3] T. Vanderbilt. “Autonomous Cars Through the Ages.” Wired. Accessed 10.25.17.

https://www.wired.com/2012/02/autonomous-vehicle-history/

[4] “How does LiDAR work?.” LiDAR UK. Accessed 10.27.17.

http://www.lidar-uk.com/how-lidar-works/

[5] M. Lapedus “Radar Versus LiDAR.” Semiconductor Engineering. Accessed 10.27.17.

https://semiengineering.com/radar-versus-lidar/

[6] “An Introduction to LIDAR: The Key Self-Driving Car Sensor.” Voyage. Accessed 10.28.2017.

https://news.voyage.auto/an-introduction-to-lidar-the-key-self-driving-car-sensor-a7e405590cff

[7] “Technology.” Waymo. Accessed 10.30.17.

https://waymo.com/tech/

[8] “Annual Global Road Crash Statistics.” Association for Safe International Road Travel. Accessed 10.25.2017.

http://asirt.org/initiatives/informing-road-users/road-safety-facts/road-crash-statistics

[9] S. Gould, Yu Han, D Muoio. “Here's the tech that lets Uber's self-driving cars see the world.” Business Insider. Accessed 10.30.17.

http://www.businessinsider.com/how-ubers-driverless-cars-work-2016-9

[10] “Average Commute Times.” WNYC. Accessed 10.28.2017

https://project.wnyc.org/commute-times-us/embed.html#5.00/42.000/-89.500

[11] T. Litman “Autonomous Vehicle Implementation Predictions.” Victoria Transport Policy Institute. Accessed 10.28.2017.

https://www.vtpi.org/avip.pdf

ACKNOWLEDGMENTS

• 
  Thanks to Colin for helping me out with some formatting and giving some ideas of what to write on this paper
  
• 
  Thanks to Han for letting me shadow you and giving me inspiration to become a computer engineer
  
• 
  Thanks to Elon Musk for helping bring automation to the consumer level
  
• 
  Thanks to my parents for helping inspire me to pursue computer engineering
  

University of Pittsburgh, Swanson School of

Engineering

Date of submission: 10.31.2017