Abstract
The idea of the autonomous car, originally, comes from science fiction. As the science and technology continues to improve and the flourishing of the artificial intelligence, ideas from science fiction are no longer infeasible. In the past several years, more and more people become interested in high and new technology industries, especially those affiliated with the artificial intelligence. Billions and billions of dollars flow into the industry with more and more scientists and engineers put their effort to create those seemly infeasible items which people treated them as daydream several years ago. In recent year, there are a lot of milestones in the high-tech industry. For example, the Mars Rover, an automated vehicle, was sent to Mars by NASA in order to generate the geography of Mars. The Tesla electric cars invented by Elon Musk, who also sent the Tesla Roadster to space through the Falcon Heavy rocket built by SpaceX. (Space.com) And Sophia the Robot, a humanoid social robot, became the first robot that received citizenship in Saudi Arabia. (Yiwen Xiao) All of those incidents represent the giant improvement of technology in today’s world. Now, a mind breaking development of autonomous car come to the public field of vision. So, what is the autonomous car in reality? And, is it safe enough to be legalized as a new way of transportation under regulation? To answer these questions, investigation and examination of the history of the autonomous car and the progress to legalize it in different countries will be performed. The analysis of the test progress and policies of autonomous cars will expand around recent incidents related to the autonomous vehicles. Overall, the policies regard to the test and actual operation of the autonomous cars are gradually developed to keep the order and efficiency of the traffic. Therefore, the development of the autonomous cars represents a new milestone in science and will bring the public a new transportation method in the future.
Introduction
Although the cars, people encountered every day, are not driverless, most of the cars come with different levels of self-driving capability. Most of the mainstream cars that equipped with “adaptive cruise control, automatic emergency braking, automated parking, and active lane control,” (MIT Technology Review Insights) are known as “Driver assistance” which is the very first level of the autonomous cars. And cars that built with “automated steering and speed control” are known as “partial automation” which is the second level of the self-driving capability. The levels beyond level 2 are known as “Automated driving system monitors the driving environment.” (SAE International.) In other words, cars from level 3 to level 5 are controlled by the automated system which should be an integration of cameras, radars, GPS, artificial intelligence, etc. In order to achieve the driverless cars, most of the giants from artificial intelligence and automobile industry begin their development of completely driverless cars. The most famous leader companies in the autonomous cars area are Google, Uber, and Tesla. Since the technology behind the driverless cars is still in its infancy, the autonomous cars are still in the testing period in order to generate enough safety data and develop related policies.
History
The history of autonomous cars can be traced back to 1969, when John McCarthy, one of the pioneers in artificial intelligence aspect, presented a blueprint of future autonomous vehicles which can drive independently without human control. He mentioned in his essay that an autonomous vehicle should be similar to cars that equipped with “automatic chauffeur” which use “television camera input” to drive on a public road. (Digital Trends.) In other words, the autonomous vehicle should be able to take the users to the destination they inputted without the users’ control, which is similar to the taxi service in reality. However, McCarthy only demonstrated what an autonomous car should be. He did not bring his idea into reality.
Further movement about autonomous vehicles appeared in the early 1990s when Dean Pomerleau integrate artificial neural networks into the footprint of the driverless vehicles. (Digital Trends) The artificial neural networks, he referred, is a technology that allows the artificial intelligence to mimic the way of human learning by finding the connection between what they see and what they will be. In other words, the cause and effect. (Digital Trends.) Based on the use of artificial neural networks, the computer-controlled vehicles should be able to distinguish images between “‘road’ and ‘non-road’ categories.” (Digital Trends.) This technology is very useful in today’s level 1 and level 2 cars which capable of the adaptive cruise control, automatic parking, etc. In fact, these mainstream automobile features were tested by Pomerleau and his fellow researcher 23 years ago in the year of 1995.
More recent development of autonomous vehicle was led by Google in the year of 2009. They began their driverless vehicle project, Waymo, and started to collect safety data. By the year of 2014, Google released their prototype of the autonomous vehicles without “any steering wheel, gas pedal and brake pedal.” (Digital Trends.)
In present days, more and more automobile companies began their development of autonomous vehicles. However, most of the automobile companies like Ford, Mercedes Benz, BMW, Volvo, etc., are focusing on improving the level 1 and level 2 features of driving in order to elevate driver experience instead of completely driverless experience. Different from those automobile companies, the high-tech companies focusing more on the integration of artificial intelligence and the self-driving technology. Besides Google, Uber and Tesla are the other leading companies that began testing on the level 3 through level 5 autonomous vehicles.
Comparison between Top Autonomous Car Builder Companies.
Different with the self-driving system that most automobile companies invented, companies from the high technology industries focus more on the development of the completely driverless system which the driver does not need to react on any incident happened while driving. Due to the similarity of their project, the top autonomous companies, Google, Uber, and Tesla, are competed in their autonomous vehicle’s features, safeties, and the completeness. If treating the autonomous companies as a monopolistic competitive market, just like the dominant position that Tesla has in the electric vehicles industry, whoever finished the development and testing period of the autonomous vehicles will gain the most share of the market. Therefore, the competition between Google, Uber and Tesla is very intense.
Google.
Google, the first high-tech company, began their development of autonomous vehicles, dominate the development of the autonomous vehicles with a large number of safety data collected in the past nine years. They announced that “its autonomous cars have collectively driven 300,000 miles under computer control without one single accident occurring.” (Digital Trends.) With such successful data, the US National Highway Traffic Safety Administration (NHTSA) has recognized Google’s autonomous driving system as official drivers last year. NHTSA also declared that the standard of Google’s driving system meets the standard of an official driver in current US law. (Dezeen) In other words, the recognition of Google’s system can be treated as a milestone that pushes the US government to develop policies that regulated accidents caused by autonomous vehicles and to legalize the entire autonomous vehicle industry.
Uber
With the success of Google, Uber, a company that provides taxi service through mobile phone applications, begins their invention of the autonomous vehicle also. Different from Google, they intend to integrate their autonomous vehicle with their taxi service. According to the Dara Khosrowshahi, the CEO of Uber, they started the invention of self-driving vehicles at the end of 2014. (Bloomberg Businessweek.) Besides developing an entirely new car, they integrate their autonomous driving system with Volvo’s XC90. And the test of their self-driving vehicles was initially operated in Pittsburgh and San Francisco. When the Uber users called an UberX, the Uber self-driving cars will be assigned to request randomly. (TechAZ) Now, the test of Uber’s self-driving vehicle is operated in a new location, Arizona, in order to receive more driving data to guarantee the safeness of Uber’s self-driving vehicles.
Tesla
Besides Google and Uber, Tesla started their development in the autonomous driving system as a feature of their vehicles, the autopilot feature. Tesla mentioned that the autopilot feature would let the cars “drive autonomously in some situation.” In order to achieve the self-driving feature of their cars, cars equipped with the autopilot feature come with eight cameras that can see up to 250 meters away. (Business Insider) Moreover, besides the Autopilot feature, Tesla also start their program to develop completely autonomous vehicles. Different from Google and Uber, their test drive program keeps opaque from public’s attention. According to Tesla’s report to California DMV, “Tesla confirmed that it didn’t test fully autonomous vehicles on state roads in 2017.” (Electrek) In other words, the progress of Tesla’s autonomous vehicle is still a mystery to the public.
Autonomous Cars Around the World.
As the autonomous vehicles become a trend of future transportation, countries besides the United States also started their own programs to develop and test the autonomous cars and to formulate related policies regards to this. Although companies, like Google, Uber, and Tesla, declare that their autonomous vehicles have collected enough safety data to be tested in the real road, there are some test drive incidents happened in the U.S. Besides U.S, countries like China also preparing for their autonomous vehicle testing. In this section, the test drive related incident will be exanimated by countries with a comparison of related policies.
United States.
As the pioneer in the autonomous vehicles industry, United States companies began their test drive on the real road since 2015 when the Tesla introduced the Autopilot feature of their vehicles. However, this is only a level 2 version of the autonomous vehicle. If only count the computer-controlled vehicles, Waymo, a Level 4 autonomous vehicle from Google, is the first one that started its test drive on real U.S roads in Arizona without a driver in 2017. Based on the data collected from Tesla in the year since 2015, date from Google since 2017, and data from Uber since 2017, questions about the safeness of autonomous vehicles are aroused dramatically. Although Elon Musk, the CEO of Tesla, said, “Cars are dangerous because of human drive them,” (Fortune.) the cars drove by the computer are not as safe as people imagined. In other words, these autonomous cars related accidents proved that the technology of the complete autonomous vehicles is not in its adultness but its infancy. The transportation law regards to the autonomous vehicles are needed to be drawn immediately.
Test Drive Incidents.
Before Google tested its cars in real city road without a safety driver, they test it with a driver. However, there is a minor accident happened on February 14, 2016. The accident happened while the Google’s autonomous vehicle tried to pass the sandbags to make a right turn, it was hit by a public transit bus from behind. The Google’s autonomous vehicle believe that that bus hit it from behind should stop and wait for it to finish its right turn, whereas the public bus did not stop. Since the speed of Google’s autonomous vehicle and the speed of the transit bus is slow, there were not any injuries. (Wired.) Through this incident, it can be seen that although the artificial intelligence built in the autonomous vehicle is smart enough to avoid the sandbag during its route to turn right, it still lacks enough awareness of human behavior. Different from machines, human’s behavior is more complicated. This accident could be caused by the inattention of the bus driver or the overconfidence of the self-driving program from Google.
Different from Google’s accidents, which is minor and without any injuries, the accidents from Uber and Tesla are severe. Since the autonomous driving test from Tesla keeps opaque, accident records generated are mostly from its Autopilot feature. It was an accident happened on January 22, 2018, when a Tesla Model S hit the back of a firefighter’s truck. The Model S was under the autopilot mode. When the accident happened, the fire truck was working on a freeway accident and hit by that Model S at a speed of 65 mph. (Electrek.) Although the Tesla is moving at a very high speed, because of the large crumple zone of the Model S, there were not any injuries. Although the spokesman from Tesla indicated that the Autopilot feature of Model S requires the drivers’ attention, it proves that the autonomous driving system is not smart enough to handle all the information generated even though it has an extraordinary vision which can identify moving items from 250 meters away.
The accident from Uber is a huge topic these days because of it is the first autonomous vehicle that killed someone while test driving. This is an incident that happened on March 18, 2018, when a self-driving Uber vehicle hit a “49-years old women while she walked her bicycle across a street in Tempe, Arizona.” (CNN Money.) The policy report released these days indicated that although there was a safeguard sit behind the steering wheel, he did not have a chance to try to stop the car since the accident happened so rapidly. And the autonomous vehicle was moving at a relatively fast speed, 40 mph. Since there was not any sign that the autonomous vehicle was intended to stop, police were investigating this accident to try to determine the responsible person. And Uber has stopped all the testing of the “self-driving vehicles throughout the United States and Canada.” (CNN Money.) This incident is an alarm to the autonomous industry. Because of the overdrive of the autonomous program, companies like Google, Uber, and Tesla are overambitious about their program. They put too much effort in test driving the autonomous vehicles to collect safety data instead of keep improving their self-driving programs.
As more and more accident data collected through the test drive of the autonomous vehicles, it is time to stop and reflect on the reason why the accidents become more and more severe each time. And how to determine the responsible person within the accidents that involve the autonomous vehicles.
Policy.
The most important problem for the future complete autonomous vehicle is the determination of the responsible person within accidents. The “autonomous driver” is a system without human emotions. It is hard to determine whether the victim from the autonomous vehicle involved accident is hit by the users’ intention, the mistake of the autonomous driving system, or the manufacturers’ intention. To legalized or to regulate the autonomous vehicles, the determination of responsible person is crucial. However, since the autonomous vehicles industry is still in its infancy, the development of the related guidelines is still on its way of developing.
At the end of 2017, the NHTSA released the guideline called “A Vision for Safety 2.0,” which lowers the barrier for the autonomous vehicle to begin self-testing. (TechCrunch.) According to this document, the determination of the responsible person within the autonomous vehicle accidents is still undeclared. There is a great chance that the accident of Uber happened in last Sunday will become a trigger to help demonstrate some of the new policies regard to the autonomous vehicle accidents.
China.
Different from the U.S., China began its autonomous vehicle development recently. The top companies that are focusing on the autonomous vehicles are Baidu and LeSEE. Although the start time for China to invent the autonomous system is late, the achievement of the Chinese autonomous vehicles system is quite a big deal. During the Chinese Spring Festival Gala in 2018, the director of the Gala show used more than a hundred electric cars developed by a Chinese automobile company to present an autonomous driving show. And those autonomous driving systems were from the Baidu, a high-tech company in China. Since the strict test drive regulation of autonomous vehicle in China, there was not that much incidents happened while test driving.
Test Drive Incident.
Although the test driving cases are lacking, there was an interesting incident happened during the annual artificial intelligence creator event in 2017, Yanhong Li, the CEO of Baidu, live streamed a video that showed he was driving the self-driving vehicle to the city road without the permitting from the Beijing Traffic Management Bureau (BTMB). Because of AI creator event was also on live streaming, the audience reported this incident to the BTMB, the police investigated this incident and charged Li with a certain amount of fine due to the violation of traffic law. Since the time he drove his autonomous vehicle was during the rush hours, it is dangerous to let a self-driving vehicle under test go to the busy city road.
Policy.
Similar to the U.S., the draft of related traffic law regard to the autonomous vehicles are still under development. However, an official of China’s Ministry of Industry and Information Technology mentioned that the responsible person of the autonomous vehicle involved accident could be the driver of the vehicle. (Forbes) In other words, the manufacturer should take the responsibility of the accident. However, due to the lack of formal laws, the Chinese policy regards to the autonomous vehicle accidents are under construction without any formal policy released.
Conclusion
Based on the rapid growth of technology and artificial intelligence, the development of autonomous vehicle has gradually become a mainstream topic. Through years of development, most of the companies that design the autonomous vehicle have reached to the testing period. In order to finalize their project as soon as possible and win the competition within the industry, some companies might become overdrive with their product. Moreover, with the increasing number of the autonomous vehicle involved accidents during the testing period, people gradually aware that the autonomous vehicles could be very risky. It is not as safe as people imagined. Therefore, it is time to pause the overambitious testing and reflect on the reasons why there is an increasing number of accidents happened. And how to regulate the autonomous vehicle accidents with related policies.