Study Cases

Companies participated in the research and development of the autonomous car in

24

increasing numbers. In terms of the industry and advantage, we divided those companies into two categories: technology company and car manufacturer.

The autonomous car was composed of an “autonomous system” and “car body”.

The technology companies had the advantage of technological ability and data. Google, for example, had the advantage of having user behavior data and road data from its network service and map service. Another example was Uber, which had the advantage of having passenger/driver behavior data and driving data from its transportation service. The various and numerous driving data were the key data for training the artificial intelligence of the autonomous car.

By contract, the car manufacturers had the advantage of having car-related technology including automobile manufacturing technology and driving assistance technology such as advanced driver assistance systems (ADAS). In addition, many autonomous car companies were being founded, which focused on the research and development of artificial intelligence and autonomous car technology. Due to their advantage, there was cooperation or acquisition among the various industrial companies, such as GM and Lyft, Ford and ARGO AI.

We listed some companies that participated in the research and development of the autonomous car in Table 10.

Table 10. The Companies Researching and Developing the Autonomous Car

Industry Advantage Company

Technology company

Technological ability, information

Google, Amazon, Baidu, Letv, Apple, Samsung, Uber, etc.

Car

manufacturer Automobile technology

Tesla, BMW, Ford, Mercedes-Benz, Audi, Volvo, GM, Toyota, Nissan, NuTonomy, drive.ai, Mobileye, ARGO AI, etc.

We took a close look at the research and development philosophy for each case.

Most of the car manufacturers sought for a practical application that they hoped would progressively improve autonomous driving technology rather than directly develop fully automatic driving technology; thus, car manufacturers are currently aiming to preferentially develop partially automatic driving technology, which adopts incremental improvement of autonomous driving.

We chose Tesla and Google (or called Waymo) as our study cases. Why we chose these two companies as the representative? Google was the first to start researching and developing the autonomous car in 2009 and kept the “driverless” philosophy which was

25

totally different from other companies; moreover, Tesla kept a research and development philosophy like most car manufacturers but was the earlier companies whose autonomous system was installed in their commercial car and actually used when driving on the road in 2014.

Regarding the incremental improvement and “self-driving”, we chose Tesla for close consideration. Tesla focused on the self-driving system and its continuous update to meet commercial needs and emphasized a philosophy that allowed the driver to use Autopilot to drive the car but still had to continuously monitor the driving situation, just as a pilot monitored an aircraft on autopilot. In contrast, Google’s philosophy was totally different. Google eliminated all controls from people and pursued making it easy and safe for people and things to move around. It was oppositely focused on the

“driverless” mode of autonomous driving and implemented a kind of disruptive design of the autonomous car (Hoag, 2015; Cunningham, 2016).

From these cases, we found that the word “autonomous” had two meanings: self-driving and driverless. “Self-self-driving” meant that the autonomous car could drive autonomously but there was still a driver; in comparison, “driverless" not only meant automatic driving technology but also emphasized that there was no driver. Because of the different philosophies and for the abovementioned reasons, we chose Google and Tesla, one of the car manufacturers, as our study cases. We summarized the differences in Table 11.

Table 11. The Comparison of Study Cases Google / Waymo Tesla pursued making it easy and safe for people and things to move around.

Tesla allowed the driver to use Autopilot to drive the car but still had to continuously monitor the driving situation, similar to a pilot monitored an aircraft on autopilot.

Focus Car and Autonomous Driving System

Google X, the predecessor of Waymo, was the Google driverless car project that was initiated in 2009. In December 2016, the driverless car department was independent

26

of Google and a subsidiary of Alphabet, which was Google’s parent company, and renamed Waymo (Etherington, & Kolodny, 2016). Waymo was an autonomous car development company, which took over the Google driverless car project.

Initially, Google started a driverless car project called Google X, which was led by Google engineers and Stanford University’s Artificial Intelligence Lab, in 2009.

They used the Google Street View Car for research and developed a driverless car.

Moreover, they cooperated with Roush Industries, a car manufacturer in Detroit.

Google was devoted to the fully self-driving car, and even emphasized “driver-less”, because Google believed the partial-automatic driving mode may make drivers complacent, so that they might be unable to immediately switch back to the manual mode in an emergency (Chang, T. -Y., 2016). At first, Google equipped the cars with self-driving technology including Toyota Prius, Audi TT and Lexus RX450h (Lavrinc, 2012). Due to the previously stated reason, in May 2014, when Google customized their own new prototype vehicle from the ground up, they removed some things from the car.

There was neither a steering wheel, brake pedal, gas pedal nor a driver in the car. The only action the driver could perform was to start and stop the car by pressing a button, which was 100% autonomous. Google attempted to realize a real fully automatic car that operated without human intervention, and we focused on the innovative prototype as the study object.

In 2015 June, the Google autonomous car became the first to complete the test of a self-driving car with no steering wheel, no braking and no gas pedal on general urban roads. According to the Google Self-Driving Car Project Monthly Report (Google, 2016), Google’s fleet of 58 autonomous cars traveled a record monthly total of 170,000 miles, 126,000 miles of which were autonomously driven. The test data on the Waymo official website showed that the Google driverless car had accumulated 3,000,000 miles driven on the road as of May, 2017 (Waymo, 2017, retrieved 2 June 2017, form https://waymo.com/ontheroad/). In 2017, Waymo introduced a new driverless car prototype with the Chrysler Pacifica Hybrid minivan and launched the Early Rider Program on its official website so that the public could apply for a test ride in the new driverless car (Waymo, 2017, retrieved 2 June 2017, form https://waymo.com/journey/).

Tesla

Tesla, founded in 2003, was the biggest automobile and energy company in America and its headquarters was in Palo Alto, California. Tesla produced and sold products including electric vehicles, solar panels, and energy storage equipment. In addition, the Tesla autonomous car was the first to be tested and driven on the road.

Tesla improved their car and developed the neural network technology of self-driving artificial intelligence based on their road data and experiences.

27

Tesla focused on the self-driving car’s software instead of the entire car. Tesla Autopilot was the Tesla’s driver assistant software. By over-the-air technology (OTA), Tesla could wirelessly update software embedded in the Tesla electric car to add self-driving features such as adaptive cruise control and lane departure warning. In other words, Tesla intended to simultaneously develop and apply Tesla Autopilot, which was first installed in Tesla Model S in 2014, and gradually perfected the self-driving system.

In the process of design and development, Tesla required the driver to monitor the car all the time, similar to a pilot monitoring an aircraft on autopilot. In terms of this principle, the driver had a critical role in Tesla’s self-driving car.

However, the first deadly crash involving a self-driving car occurred in May, 2016 in the U.S.A, as the car was driving on a Florida highway. At the time of the accident, the Tesla Model S car was in the “Autopilot mode”. Due to the crash and subsequent disputes, Tesla and Mobileye ended their partnership in 2016.

Up to 2016, approximately 90,000 Model S and Model X Tesla electric cars had Autopilot software installed and in use (Stewart, 2016). As long as the cars continued driving on the road, the Autopilot software could send data such as road conditions back to the Tesla database. Moreover, Tesla had the Autopilot system continuously learning and improving by big data analysis. It was as though there were 90,000 cars assisting the testing of the Tesla self-driving car. Moreover, Tesla Motors said that it expects to enable full self-driving by the end of 2017 (Ayre, 2017; Lambert, 2017).