R&D Activities In Autonomous Vehicles Using Patent Information

Overview of Automobile Industry

In the automotive industry, CASE which is initials of connected, autonomous car, sharing and electric vehicle has been focused. Above all, autonomous driving is rapidly developing technology by the reason of improvement of computer science. The sales amount of the world's automakers is shown in Figure 1 . Among these manufacturers, TOYOTA, VW, Ford, GM, HONDA, BMW and NISSAN have entered the automated autonomous vehicles industry. Also, the list of impacts by the practical application of autonomous driving is shown in Table 1 . Significant effects are expected such as reduction of traffic accidents, elimination of traffic congestion, response to declining birth rate and aging population, improvement of productivity, and strengthening of international competitiveness.

See Full Size >

Overview of Autonomous Vehicles Industry

Definition of Autonomous Driving and Entry Companies

A definition of the autonomous driving levels proposed by SAE (Society of Automotive Engineers) in U.S. is shown in Figure 2 . There are significant differences between greater than or equal to level 3 and less than or equal to level 2 in the definition. The driver has responsibility for recognizing the driving environment in up to level 2, but the control system is ultimately responsible it in level 3 or higher. Therefore, the core technologies required for automated driving in level 3 or higher are considered control system and software. In addition, significant technological innovation is required for its practical application. The end product manufacturers including Toyota, Nissan, Honda, Mercedes, GM, BMW and etc. have entered from Japan, Europe and the United States. However, it is said that it has been limited to practical operation up to level 2 of automatic operation. In the future, intensified competition is predicted including ICT companies that are not car-made manufacturers (Osawa & Nakamura, 2015).

See Full Size >

(Source: http://cyberlaw.stanford.edu/blog/2013/12/sae-levels-driving-automation)

Technical knowledge required for fully automated driving

As indicated in the autonomous driving levels, the technology for "control" and "system" is required for the realization of fully automatic operation. The technologies related to fully automated driving and its investment ratio are shown Figure 3 .

See Full Size >

Thus, it can be seen the circumstances that the investment in technologies related to "software and control" are high. The control system is more important than the vehicle in itself in the automatic driving field. There is “IoT” in ICT industry. IoT is a technology that acquires information with sensors, stores data in the cloud, analyzes the data with artificial intelligence, and then activates the results. This technology can be applied to automatic driving. Because "prediction, recognition and judgment" are required to operate the vehicle, and these are controlled by software. A concrete representation of the technology required for such operations are shown in Table 2 (Suda & Aoki, 2015).

See Full Size >

The process of operating automatic operation by these technologies is as follows:

• Measure using GPS

• Create a local dynamic map that integrates the map information and the information read by the sensor. After recognizing the driving environment and the destination, the artificial intelligence module understands the environment and makes decision to correct the target driving.

• Perform actual speed control and steering control based on the information from the artificial intelligence module.

This process also requires big data analysis including not only road and traffic information such as inter-vehicle distance and white line distance but also traveling environment sensing and obstacle recognition.

Patent Information and Patent map

Patent documents are an ample source of technical and commercial knowledge. The patent is one of the indicators of capacity for technological development. There are some researches aimed at visualizing and analysing patents, or proposing efficient text-mining approaches for creating patent maps (Arai, 2003; Chen et al., 2018). Therefore, patent information is used as indicator of R&D capability in this research. All Japanese patent information is managed in the Patent Information Platform (J-PlatPat) operated by the Japanese Patent Office and everyone can search, browse and download the patent information. All patents are classified according to each of three classification codes, IPC (International Patent Classification), FI (File Index) and F-term (File Forming Term). In this research, all patent publications related to autonomous vehicle based on both of following conditions from 1990 to 2015 are collected, as in the case of previous research.

((AUTONOMO+ AND CAR) OR (UNMAN+ AND VEHICLE) OR (DRIVERLESS AND (CAR OR VEHICLE)) OR (SELF-DRIV+ AND (CAR OR VEHICLE)) in Claim

B60W in IPC code

There are following three ways to analyze patent information.

• Quantitative analysis: An analysis method to grasp patent information as the number of applications, the number of inventors, the number of keywords, etc.

• Qualitative analysis: A method of analyzing patent information based on the technical content.

• Correlation analysis: A method to analyze the existing relationship between patent information.

End product manufactures, component suppliers and the ICT industry have entered the automated driving industry. Therefore, we investigate trends in the number of patents and technology development by each industry using heat maps and changes of core technology using HITS algorithm in this research. Specifically, we will clarify the characteristics of technological development and changes in core technology in these three industries. The list of companies in each industry is shown in the Table 3 .

HITS Algorithm

The HITS algorithm is a method to extract highly important information by performing link analysis (Kleinberg, 1999). The Authority indicates what has important information and the Hub, and the Hub indicates what is linked to the Authority. The Authority Score and Hub Score are extracted by repeating the following equations.

$$Auth\left(p\right)=\sum Hub\left(q\right)$$

$$Hub\left(p\right)=\sum Auth\left(q\right)$$

In the above equation, q means the score before the evaluation, and p means the score after the evaluation. Therefore, there are relation that a good authority is linked by multiple good hubs, and a good hub is linked by multiple good authorities. In this research, we apply this HITS algorithm to a patent citation network to extract core technologies that are attracted attention from other companies in the industry.

An Approach Based on the Number of Patent Publications

As our first approach, the numbers of patent publications related to R&D of autonomous vehicle in Japan and U.S. are shown in Figure 4 . The horizontal axis represents the year, and the vertical axis represents the number of patents based on application year in the figures.

See Full Size >

The number of applications for Japanese patents increased around 1997 and then decreased, but it can be seen that the number of patents increased sharply with the development of AI and image processing technology in 2015. Looking at the change in the number of English patents, it can be seen that the number has gradually increased since around 2005, and has grown rapidly in recent years. By these figures, it can be understood that interest in automatic vehicle development has increased.

An Approach Based on the Research Fields

Research Fields in Japanese Patent

Heat maps that is represented the R&D fields based on patents by each industry in Japan are shown in Figure 5 . The region where the color is dark red in the heat-map indicates that the technology development factor is highly focused in the figures. The vertical axis in the figures uses F-term with a large number of occurrences in their patents. The degrees of emphasis in the all industry are increased in 3D241 (related to control of drive devices and motor control of vehicles), which is an essential element in automatic driving technology, in the 2000s. It can be seen that end product manufactures have begun a wider range of technology areas than other industries by the industry comparison. In addition, there is a feature that the start area differs according to each industry.

See Full Size >

Research Fields in U.S. Patent

Heat maps that is represented the R&D fields based on patents by each industry in U.S. are shown in Figure 6 . IPC is used in the figures because F-term deals with only Japanese patent documents. As a result, it can be seen that there is no involvement in the R&D for autonomous vehicle in the 1990's without end product manufactures. However, the R&D fields are expanding in the parts supplier industry and the ICT industry unlike the heat map of the Japanese patent.

See Full Size >

An Approach Based on the Core Technology

Autonomous Level of Core Technology

The automatic operation levels of the top 30 core technologies in every year that is applied HITS algorithm to citation network are shown in Figure 7 . The core technology that corresponds to the level 1 of automatic operation is red, the core technology that corresponds to the level 2 is yellow, and the core technology that corresponds to the level 3 automatic operation is green in the figure. Patents for driving assistance, steering wheel operation, emergency accident prevention, navigation in order to cope with level 2 (system supports both steering operation and acceleration / deceleration) are ranked in almost every year. Also, there are some patents for toll road charging on freeway and IC merging devices since 2000. Therefore, there is a possibility that autonomous driving level 3 (the system operates all at a specific place except emergency) can be realized on a freeway.

See Full Size >

Transition of Leading Companies

The trends in the number of patents by each industry in the top 30 major technologies are shown in Figure 8 . It can be seen that almost all major technologies were generated by end product manufacturers. In addition, the technologies produced by the ICT industry fell temporarily in the 2000s, but their technology constructed around 2015 has gained much attention.

See Full Size >

Discussion

Autonomous vehicle technology involves various technologies. So, not only companies of automotive manufacturers, but also automotive suppliers, and ICT companies can be major in certain areas of autonomous vehicle technology. Using HITS algorithm, we find most of the top thirty companies of automotive driving patent ranking are automotive manufacturers. Autonomous vehicle technology involves not only automotive technologies, but also various technologies. But it is said that the most major cost of building cars is factory of automotive manufacturers, not materials. And for development of products, technological development of manufacturing is needed. So, even in era of autonomous vehicles, the importance of technology of automotive manufacturers will be still high.