cool hit counter Google's unmanned car layout strategy from patent analysis_Intefrankly

Google's unmanned car layout strategy from patent analysis


Author | Li Yue Lv Lin Beijing Kangxin Huayuan Intellectual Property Consulting Co.

(This article is an authorized contribution by Knowlesys and is reproduced with the permission of the author and with prominent attribution. )

(This article is 3462 words and takes about 7 minutes to read)

On December 13, 2017 Google officially announced the establishment of Google AI China Center in Beijing, while Google actively formed an AI China team, releasing signals of its return to China. After the establishment of Google's AI China Center, driverless cars will become a key area of its research.

Google's driverless car project was officially launched in 2009 with the goal of releasing a commercially available driverless car by 2020, and in 2016 Google parent company Alphabet spun off the self-driving car project separately to create a company called Waymo. So far, the project's test fleet has accumulated 4 million miles on the road, holding the lead in actual miles driven by driverless cars [1]. This paper explores Google's layout strategy for driverless cars in China based on Google's driverless car-related patent applications in China.

I. Analysis of trends in the number of applications

The following chart shows the trend of Google's related patent applications from 2009 to the present, which shows that Google has carried out the layout of related patents in China when the driverless car project started, which is enough to show that Google attaches considerable importance to the Chinese market. In 2013, Google's patent applications increased significantly, which may be due to the normal patent output after the accumulation of technology R&D, or may be due to the fact that Google strengthened its own patent layout in China when traditional car companies entered the driverless car field in 2013[2]. In 2014 and 2015, Google driverless cars still maintained a high number of patent applications, indicating that Google's driverless car technology innovation is still ongoing, and subsequent predictions are that Google's patent layout for driverless cars in China will continue to grow (note: patent application data for 2016 and 2017 have not been fully disclosed).

Figure 1 Google China patent application trend chart

II. Technical layout analysis

(one) Technology distribution analysis

Google's driverless car is a prime example of multi-technology convergence, Driverless technologies can be classified in the order of information transfer:1) Environmental awareness technology: Responsible for on-board multi-sensor data acquisition and data analysis process;2) Planning and decision-making techniques: Route guidance and driving decisions based on the processing of sensed data;3) Operation control technology: Automatically control vehicle movement by following the path provided;4) network communications technology: Responsible for communication between driverless cars and other vehicles or networks[3]。 The chart below shows the distribution of Google's driverless car patent technology, It can be seen that Google's patent layout for driverless cars in China focuses on environmental awareness technology。

Figure 2 Google China Patent Technology Distribution Map

(ii) Analysis of trends in technology applications

The following chart shows the distribution of the number of applications for each technology branch of Google's driverless car over time. As the development of Google's driverless car technology advances, the number of related patents laid out in China is gradually increasing, and it can be seen that the number of Google's patent applications in China under each technology branch has been uneven since 2013, which may be due to the fact that Google will break through technical difficulties in different directions every year as the Google driverless car project is developed.

Figure 3 Google China Patent Technology Time Evolution Chart

( three) Focus on difficult technologies analysis

Based on investigating the milestones achieved by Google's driverless car project in recent years and the major technical problems encountered during the project research [4], [5], combined with the driverless car technology decomposition system and Google's publicly available patent information in China, the difficulties in the direction of Google's driverless car technology are summarized as follows.

Figure 4 Google Driverless Difficulty Technology Distribution Map

Among them, the detection of road information is the core problem of environment perception technology, the solution of the "where to go" problem is the core point of planning and decision technology, the control algorithm and the modification of vehicle peripheral control mechanism are the main problems of operation control technology, and the vehicle network communication is the research focus of network communication technology.

The technology development route for Google's driverless car is further analyzed based on the identification of the technical development difficulties.

Figure 5 Google driverless car technology development swim lane diagram

As can be seen from the swim lane diagram above, in terms of environmental sensing technology, for the core issue of road information detection, Google first disclosed a technical solution for vehicle panoramic camera for road information detection in China, and then proposed a technical solution for rotating LIDAR for road information detection in 2014, followed by a technical fusion of panoramic camera and radar utilized in a technical solution for pedestrian recognition requiring protection in 2016.

In terms of planning decision technology, Google initially disclosed a technical solution for path navigation using maps, then in 2012 proposed a technical solution for path navigation using a combination of cameras and maps, followed by a further request in 2015 to protect a technical solution for navigation guidance using a combination of maps and multiple sensors to automatically determine the start and end points.

In terms of operation control technology, in response to the control algorithm and control mechanism modification, Google first made public in China a technical solution capable of automatically controlling the car merging with the road, and in 2015 proposed a fully automated driving and fully automated control without technical solutions on top of that.

In terms of network communication technology, in response to the demand of self-driving vehicle network communication, Google firstly disclosed the technical solution of vehicle information cloud synchronization in China, and then proposed the technical solution of driverless vehicle using network history data for information recommendation in the direction of "Telematics" application in 2017.

Third, Google's four generations of driverless cars in China patent layout

analysis

Google's driverless car has used four generations of prototypes so far. With the change of Google's driverless car R&D ideas in each stage, the focus of Google's patent layout in China will also be different, and the following is mainly based on the R&D characteristics of Google's four generations of prototypes to analyze the patent layout strategy of Google's driverless car in China.

(i) Features of the development of the Google IV prototype

Figure 6 Google's four-generation driverless prototype car

In 2009, Google officially launched its self-driving car project, using a driverless prototype car that was a Toyota Prius conversion. The first generation of Google's driverless car development was characterized by a "prominent map advantage", which was evident in the maps it came with [6].

In 2011, Google switched its test vehicle from a Toyota to a Lexus, and its research and development was characterized by a "focus on LIDAR," with Google claiming that the most competitive part of its self-driving technology was its own custom LIDAR, which was later at the heart of Google's lawsuit against Uber [7].

In 2014, Google developed its own driverless model, the PodCar Firefly, and used it as a third-generation driverless prototype, signaling a shift in Google's research and development strategy to "abandon assisted driving and go all in on fully autonomous vehicles. At this stage Google decided to abandon the development of assisted driving features that require human takeover and to devote all its R&D efforts to autonomous driving technologies that do not require human intervention [8].

In 2016, Google selected the fourth generation of the Pacifica driverless car for road testing, a generation of driverless cars characterized by "collaboration with traditional car companies to develop driverless technology." At this point Google abandoned the route of building its own car and instead partnered with Fiat Chrysler to develop a hybrid vehicle for driverless related technology development [9].

(ii) Patent layout of Google's four generations of prototype vehicles

1.The first generation focused on the layout of navigation map-related patents

Google's own high-precision maps with can assist driverless cars in positioning and navigation, and Google's technical solution for using its own high-precision maps for self-driving navigation has also been disclosed and licensed in China (CN201180057954.0), which addresses the specific problem that when a self-driving car encounters a special section of an overpass, the shape and height of the road, lane lines, intersections, speed limits, traffic signals, and other real-time traffic information is particularly complex, which poses a challenge for the self-driving vehicle. Google's disclosed technical solution is to access Google map information in one or more areas, use the sensors on board to identify such real-time traffic information, plan a route to the destination, and drive automatically along that route toward the destination, in addition to warning other passing vehicles by means of flashes and sounds.

2.The second generation focuses on the layout of rotating LIDAR-related patents

Google at this stage said that the most competitive part of its own autonomous driving technology is the self-developed custom LIDAR, this rotating LIDAR technology solution has been disclosed in China (CN201480054147.7), in which the LIDAR LIDAR needs to be able to sense things around it in 360 degrees of direction, so the problem brought is how to provide power, send or receive communication to the LIDAR LIDAR device during the rotating action, Google's disclosed technology solution is to use some non-contact electrical coupling, use the transformer formed between the non-contact electrical coupling and the LIDAR coil to provide power to the LIDAR in rotation, and use the capacitor formed between the non-contact electrical coupling and the LIDAR conductive ring to send or receive communication to the LIDAR in rotation.

3.Third generation steering layout fully automated driving related patents

Google in this stage of full development of fully automated driving car technology program has also been made public in China (CN201580023657.2), according to the final vision of Google driverless car, when passengers enter the car to fasten the seat belt, through the voice to say the destination, press the start button, the car will automatically carry passengers to the destination, when encountering special conditions need to emergency stop, passengers press the stop button can automatically stop, emergency stop button is equipped with a cover to prevent accidentally touching the emergency stop button, in addition, passengers can also voice control the vehicle to pull over.

4.The fourth generation focuses on the layout of vehicle networking-related patents

Google abandoned the route of building its own cars at this stage, opting to partner with traditional car companies for development. As the patent information has not yet been disclosed, the specific characteristics of the vehicle at this stage have not been fully disclosed, but it is speculated from some of Google's patent documents that have been disclosed in China that an important direction of Google's driverless car patent layout in China at this stage is "car networking" related technology, and its technical solutions related to the direction of car networking have also started to be disclosed in China (CN201710281924.7). The patent describes the method of using network historical data to recommend internal parameter settings for self-driving vehicles, and the self-driving car uses machine learning to adjust the car seat or ambient temperature inside the vehicle for a specific passenger.

IV. Summary

To sum up, Google has a strong awareness of patent protection and its patent layout has the following characteristics: 1) simultaneous patent layout with R&D projects; 2) continuous patent layout around key technology breakthrough points; 3) providing product-specific protective layout with the continuous evolution of prototype vehicles.

Google driverless cars will be further automated and intelligent in the future, although the difficulty in getting the car to act like a human to identify and understand various driving conditions and ensure the safety of passengers is still at the level of artificial intelligence, which is one of the main reasons limiting the commercialization of Google driverless cars. We are now 2 years away from the commercialization of Google driverless cars in 2020, so let's wait and see if it will be safe to ride in a Google driverless car then.

Notes.

1.http://www.techweb.com.cn/world/2017-11-28/2611488.shtml

2.http://www.eefocus.com/automobile-electronics/398442/p4

3.https://wenku.baidu.com/view/a1a21c8a6c175f0e7cd137cf.html

4.https://cn.nytimes.com/technology/20150902/c02googlecar/

5.http://www.eeworld.com.cn/qrs/article_2016030426774.html

6.https://www.leiphone.com/news/201610/IJ8U9iR2E4U6vGDA.html

7.http://tech.qq.com/a/20170521/022504.htm

8.https://www.ithome.com/html/auto/269035.htm

9.http://www.sohu.com/a/203732869_120865

Source: Concentrix IP


Recommended>>
1、The strange things in the cryptocurrency world do not look confused I lose
2、The future of AI in all shapes and sizes
3、Differential localization based on mesh networks
4、4 Easy to understand blockchain 51 arithmetic attack
5、Blockone blockchain governance for EOS

    已推荐到看一看 和朋友分享想法
    最多200字,当前共 发送

    已发送

    朋友将在看一看看到

    确定
    分享你的想法...
    取消

    分享想法到看一看

    确定
    最多200字,当前共

    发送中

    网络异常,请稍后重试

    微信扫一扫
    关注该公众号