Autonomous Vehicles 101
In 1885 the car was created and for over a century we have come to increasingly depend on efficient methods of transportation such as the car and plane as we move away from more traditional mediums of transportation such as animals. Ever since the innovation of the car, humanity has also begun to look for ways in which the car could function properly without the large assumed risk involved with driving a car; these factors are called human risk factors. As reported by the National Highway Traffic Safety Administration, almost “94 percent of serious crashes are due to dangerous choices of errors people make behind the wheel”, therefore it is important to think about how we can potentially eliminate the need for humans for these types of processes.
The Future of Driving
When people begin to talk about AV — Autonomous Vehicles — they are generally talking about the more complex levels of autonomous driving on a scale, also known as the SAE levels of automation graph. The graph reflects the decisions created by the Society of Automotive Engineers for how to define the different levels of automation which is a great benchmark for most companies to base their current cars or future cars upon.
Level 0 — The level before even partial autonomy is the level that cars would have been categorized in, a little over 10 years ago. This was the state of my 2008 Honda CRV because it had no control of the breaks or acceleration nor my steering wheel which is a significant determinant for the next levels of autonomous driving.
Level 1 — The first level of autonomy is the level of most current day mid-priced cars. This is also the state of the super-popular Honda civic 2019 which is the car that I currently drive. This car has the capability of sensing cars through some kind of depth perception and is able to take control of the brakes if the situation becomes dire. This also usually has a lane detection system as well which may take control of the wheel.
Level 2 — The second level is the level a lot of Tesla vehicles are currently in. The commercial Tesla vehicles are able to take control of the gas pedal, steering wheel or the brake but not all at the same time. This level still requires the driver to pay full attention to the car’s surroundings and to be ready for any situation in which the autopilot system may fail and need help or backup.
Level 3 — The third level is what the current industry innovators are trying to achieve as this is a stepping stone for the final 2 levels. Uber, Waymo (a subsidiary of Google), BMW, GM, and many more companies are currently looking to make their cars completely capable of taking control of both the steering and the pedals at the same time. The driver must still be ready to take control of the vehicle but not required to pay full attention to the road and environment at all times.
Level 4 — Level 4 is the level that most people like to call “Fully autonomous”. At this point, the vehicle is able to completely drive itself and requires no interjection of the driver. At this point, humans are not even required for driving but there is still access to a fully functional cockpit. The use of the cockpit is in the rare case that a human may want to intervene and take control of the car. The only reason why this is not currently in use is because of the limitations of current laws and legislation that restricts many companies but will soon be incorporated everywhere. Some companies have already reached this point with their cars but there are still important ethical dilemmas that work through and more software to refine.
Level 5 — The vehicle takes full control and does not need a driver to do anything. At this point, you can literally take out a functioning cockpit from the car and it would work just fine. This is what is shown in most science fiction movies or what you would see at the back of limo where all the seats are facing each other because it is irrelevant to look at the road unless for self-entertainment.
But…How Does It Work?
Well, we are still currently in the innovation phase and trying to find the next best and cheapest technology to make the cars more efficient and in general better for social safety. Current technology in use for autonomous vehicles includes sensors, lidar, radar, and photographic systems in order to make a car drive perfectly on a road. One limitation to current technology is the price to manufacture and consume these sensors for cars, but as demand increases price generally decreases through more production which will make it easier, faster and cheaper to make completely autonomous cars.
The separate uses of each of these sensor devices are extremely unique and complicated so you may want to consume dedicated content for that. But the general use of these technologies is to create a virtual environment for a computer system to better understand and decide certain actions for driving. Putting together all of the gathered information from the sensors, a computer can create a whole 360 degree animated simulation for itself and make super accurate and fast decisions.
Instead of the use of an actual human in these extraordinary cars, it uses a super interesting function in artificial intelligence called deep learning. This is a software that is able to learn through real experiences that the car is put through — like driving on real roads and reacting to unique and difficult circumstances — and uses this information plus past coded algorithms to drive effectively. Because of the use of deep learning, it makes the potential of these cars almost infinite times better than hard coding all the rules because it will create its own rules and continuously learn and fix any issues by making sure to never make the same mistakes again.
The final technology used to make the dream of a fully autonomous car possible is the technology used to send information from one car to another or one car to its computer. With the soon to come new 5G networks, we know that its effect on the automotive car industry will be groundbreaking because of the amount of information we could potentially send, and its speed. Our predictions on its influence on the current industry are that it will make cars a lot more reliable by allowing for more decisions to be made at small intervals of time. With the limitations of current networks, it makes sending information to a computer which will make decisions for a car extremely hard because of the amount of information accumulated by all of the sensors at the same time. In the future when this technology becomes a viable option for data transfer, it will allow for the car to communicate with its entire virtual environment including other cars and even the traffic light at a much faster speed which will make autonomous vehicles more viable.
Controversy
Despite all of the upside of these extraordinary cars, they are still obviously being looked at with a keen eye by society. People have in fact created thought experiments to bait a viewer or reader to be against the use of these cars because of their infinitely faster decision-making times. They are basically exploiting the ability for a car to make decisions faster than humans and rationalizing the human's wrongful decision making by calling it a natural reflex. Which means that society is attempting to reinforce human flaws. Asking people if it is okay for a car to choose to hit a person on a bike or a child crossing the road is an unreasonable question to ask and a rare case to consider which shows how scared people can be to the massive advantages of self-driving cars. Despite this fact, it will still take years of technological progression and the creation of laws, legislation, and rules to be accepted before anyone is able to own or use one of these cars on a daily basis with no limitation.
In the future, with hope, an open mind and research, we might just be able to eliminate humans altogether from the equation of driving, making transportation safe and easy. By teaching cars how to drive themselves, in the next decade we might just be able to live our science fiction dreams.
TLDR
There is a new type of car around town, that’s the self-driving car. With all the hype around it, especially Tesla, people have no clue about the potential of self-driving cars and how they actually work. Through the use of an assortment of sensors such as lidar, radar, and cameras, computers are able to use an Artificial Intelligent software called deep learning to slowly but effectively teach a car how to drive itself in the most vigorous and demanding situations. Through limitations set by the government and disagreement with some members of society, the self-driving car industry is also very limited in growth but with hope, an open mind and research, anything is possible!
