An autonomous driving technology that makes effO Rts fO R full -scale practical use in various parts of the wO Rld.Development competition has been intensifying since 2010 as a new industry that deducted in the first half of the 21st century, but its development histO Ry is unexpectedly old.
In this article, the histO Ry will be unraveled with the change in development that will lead to the dawn of the dawn of autonomous driving development.
In addition, I referred to the paper by Dr. Sadayuki Tsugawa, a doctO R of engineering, who is familiar with the histO Ry of autonomous driving.Mr. Tsugawa is a leader in the industry in the 1970s who was involved in the development of an autonomous driving system, which is the first in Japan when he belongs to the Industrial Technology Institute of Machinery and Technology (currently the Institute of Industrial Technology).
▼ 60 years of autonomous driving system | Sadayuki Tsugawa https:// www.jstage.jst.Five.jp/article/sicejl/54/11/54_797/_PDF/-CHAR/ja ▼ Automatic driving This old stO Ry | Sadayuki Tsugawa https:// www.JSAE.O R.jp/~dat1/mr/motO R15/mr200235.pdf
Contents of the article
自動運転にいち早く注目した自動車メーカーは、米GMのようだ。1939～40年に開催されたニューヨーク万国博覧会で自社パビリオン「Highways and HO Rizons」を設置し、この中で、20年後、1960年の未来像を具現化した巨大ジオラマ「Futurama（フューチュラマ）」を展示した。「Future」と「PanO Rama」を合わせた造語だ。
The diorama was circular, formed an image of a high -rise building and residential areas with the image of a highway network connected, and the audience was an attraction that was explained while sitting on a moving chair around it.
The highway in this exhibition was "Automated Highway" that allows automatic driving.It seems to be a content that guides the driving car by radio waves and is at the destination.Although the detailed mechanism is unknown, it seems that the current mechanism such as the current ITS (advanced road traffic system) was conceived at the time.
Although it did not seem to be concrete R & D, GM seemed to have gained popularity, producing and exhibiting the "Futurama2", which also renewed the content at the expo that was held in 1964-65.
Autonomous driving development seems to be embodied in the late 1950s.Traffic issues such as traffic accidents and traffic jams have become apparent, and GM has begun joint development under the proposal of the US radio company RCA.
In 1958, GM conducted the first demonstration of automatic driving cars at the Tech Center to introduce the possibility of a built -in guidance system for future highways.It is a method of laying a guided cable on a highway, flowing current, detecting a magnetic field with a guided coil attached to the front of the vehicle, and controlling it.
Although the cost of laying the cable was enormous, it did not spread widely, but was partially introduced as a control technology in the bus stop and the control technology in a closed space.
At this time, it seems that research and development at universities and others began to progress.Around 1960, Stanford Cart, which leads to research on remote control technology for lunar exploration, begins at Stanford University.A graduate student who was involved in the NASA project is conducting research on technologies to control remote vehicles using video information.It seems to be a plan to do.
At the beginning of the development, the camera was equipped with a camera with a simple battery -style four -wheeled vehicle, and it was directly connected to a TV display and a control console that controls the steering and speed with a long cable.It seems that the magnetic tape loop simulates the effects of communication delays, and has been conducting research on the possibility of controlling vehicles while avoiding obstacles with various combinations of communication delays and speed.
After that, President John F. Kennedy announced an inhabitant mission to send an astronaut in the month, and the study, which contributed to unmanned, seems to have fallen temporarily, but a low power TV transmitter and a wireless control link.Development has been continued, such as building and implementing a visual guidance project.
In the late 1960s, the Technology Development of the Federal Country Committee acquired an experimental TV license and controlled carts via a computer based on television images.1.It seems to have reached the stage to automatically follow the high contrast white line in 3 km.
It is said that this development has continued, leading to later participation in the DARPA Urbanca Range.
At Stanford University, research and development of robot "Shakey", which utilizes artificial intelligence in parallel with Stanford Cart, began in the 1960s.
The robot is equipped with a wheeled base, a camera control unit, an infrared triangular measurement meter, an antenna for two directional wireless communication, and an electronic device that contains an on -board computer, and runs through a telephone instruction.is.
It is said that the number of rotations on the wheels has been tracked by tracking his own position, and landmarks and objects are scanned using a camera and a distance gauge.
According to Tsukawa's paper, in Japan, in Japan, under the influence of the United States, the Ministry of International Industrial Technology Institute of Industrial Technology (the current Industrial Technology Research Institute) began research.Like the United States, it seems that the method of laying a guide cable on the running route was the main focus.
In the 1970s, research and development in the self -driving fields in Japan also bloomed.The Institute of Mechanical Technology, which promotes research on autonomous driving systems, developed the Intelligent Vehicle, the world's first machine vision, in 1977.
It is a method of placing a stereo camera up and down at the front end of the vehicle and measuring the detection of obstacles in a two -dimensional perspective based on the parallax, and the image processing was made of a video signal from the camera with a hard wired logic.。At that time, when 8 -bit microcesses began to appear, computer processing on the car was virtually difficult, and it was said that the video signal was dualized and processed directly on the IC logical circuit.
In the United States, the development of "ALV (Autonomous Land Vehicle)" for military purposes such as reconnaissance has been developed.The image processing and the rangefinder recognize the surrounding environment, and to achieve automatic driving by making judgments and planning by AI.
The study was subsequently taken over to the University of Carnegie Melon and the US Standard Technical Research Institute, and was applied to research on passenger cars and Mars.
In the early days of development in the 1970s, research by the Mechanical Technology Research Institute was performing a lateral control along the guardrail detected by the vision system, but in the 1980s, the dead reconising function based on the differential Odometa is a vehicle.It is said that it was installed and lattered in a system similar to the inertial navigation.As a result, in 1984, it became possible to automatically drive to the destination at 10 km / h while detecting and avoiding obstacles.
The steering control algorithm was developed based on dead reconging and running path databases.At the time, the car was equipped with a computer body, a CRT, magnetic cartridge tape device, and an XY plotter that recorded the running trajectory.
In the late 1980s, it seems that the attention of autonomous driving has increased in the national project to solve traffic issues.In Europe (EU), the ITS project "Prometheus", which makes vehicles intelligence under the proposal of Daimler, began in 1986.PSA and Renault also participated, and system development has been developed to detect running lanes and obstacles with machine visions and radar.
The "Vita2" developed by Daimler was equipped with 18 cameras to detect 360 degrees around the vehicle, and was able to follow up lanes and change lanes at a speed of over 100 km / h.
In Japan, Nissan and Fujitsu have developed an autonomous driving experimental system called "PVS" as a project of the Machine System Promotion Association.Equipped with a camera for lane marking detection, a camera for detection of obstacles, and an ultrasonic sensor for guarding the guardrail, it is equipped with a dedicated image processing device to be processed at video rate, and controls the knowledge based on the experience of a veteran river.He said he went.
In the early 1990s, Path, the IVHS project in California, launched in 1986, was developing automatic tracking technologies called platoning.It seems that the company was aiming to develop a system that enables high -density cars by using a magnetic marker to enable high -density.
The Robotics Research Institute of the University of Carnegie Melon University in 1995 is an autonomous vehicle called "NAVLAB 5", which remodeled a minivan..C.Out of about 4,800 kilometers from San Diego to San Diego, more than 98 % ran by automatic driving.Although the vertical control, such as brakes, was manual, it was said that autonomous driving was realized in horizontal control.
This study was involved in Takeo Kimide (currently a special professor invited to Kyoto University Higher Research Institute), who had been in the research institute at the time.He is a global authority known as a leading person in image recognition technology.
In Japan, a demonstration of a road coordination -type autonomous driving system utilizing the magnetic marker system was held on the Joshinetsu Expressway, which was led by the Ministry of Construction at the time.
In 2000, the Mechanical Technology Research Institute and the Automotive Electronic Technology Association (currently the Japan Automobile Institute) conducted an experiment and running demonstration of a coordinating system in which five autonomous cars were traveled by cars between cars in 2000.
In 2004, the US Defense Higher Research Planning Bureau (DARPA) implemented the DARPA Grand Challenge (DARPA Grand Challenge), which aims to run long -distance courses using automatic driving technology.15 teams participated in the world's first long -distance and unmanned automatic driving race.
The vehicle that ran 142 miles (about 230 km) in the desert area within 10 hours was awarded, but there were no complete cars, and the Red Team team at the University of Carnegie Melon, which ran about 11 km, was the highest grade.It looks like it is.
The team of the Massachusetts Institute of Technology, who participated, was later founded by Kyle Vogt, who founded CRUISE AUTOMATION.
翌2005年開催の「2005グランドチャレンジ」では、決勝に進出した23チーム中5チームが完走した。1位はスタンフォード大学の「StanfO Rd Racing Team」、2位、3位はカーネギーメロン大学の「Red Team」が占めた。
なお、StanfO Rd Racing Teamのリーダーは、後にGoogle XやUdacityを創設するセバスチャン・スラン氏が務めていた。
The third event, the 2007 Urbanca Range, moved the venue to a closed Air Force base, and was a new initiative to finish about 100 kilometers in urban areas within six hours.
競技の結果、1位はカーネギーメロン大学の「TARTAN RACING」、2位はスタンフォード大学の「StanfO Rd Racing Team」となった。
TARTAN RACINGでは、後にAurO Ra Innovationを創業するクリス・アームソン氏がテクノロジーディレクターを務めたほか、Nuro創業者のデイブ・ファーガソン氏、ArFive AI創業者のブライアン・サレスキー氏らが名を連ねている。
In response to DARPA's efforts, it was Google of the United States that started moving quickly.In 2007, he welcomed Sebastian Slan to prepare for autonomous driving development, and launched the Self Driving Car Project in 2009 with one of the leading engineers and others and others.
Private initiatives in anticipation of the full -fledged social implementation of autonomous driving technology that leads to today are greatly emitted in this Google activity.The lobby activity enabled on public road driving on automatic driving cars, indicating the possibility of autonomous cars in the form of many people.
It is also notable that Chris Armsson has produced many engineers, and today's automatic driving development competition with many startups is very large based on DARPA participation engineers and Google.
2010年代に入ると、自動運転技術を開発するスタートアップが続々と立ち上がった。国内では、2014年に先進モビリティ、2015年にティアフォーが設立されたほか、海外では、2013年にCruise Automation、2016年にNuroやArFive AI、Embark Trucks、2017年にAurO Ra InnovationやMay Mobilityなどがそれぞれ創業され、開発競争をけん引している。
Against this background, it is likely that AI (artificial intelligence) technologies such as deep learning (artificial intelligence) technologies, along with the impact of the preceding Google.
The concept of the "autonomous driving level" was born exactly, but the SAE (US Automobile Technology Association), which is currently widely recognized, was formulated in 2016.
Until then, NHTSA (Ministry of Transport Road Safety Bureau) was presented in 2013, and the definitions formulated by SAE in 2014, but when SAE was revised to the current standard in 2016.NHTSA also decided to use the definition of SAE for unifying the definition.
The new standard of SAE has penetrated as the most adopted concept that is adopted around the world, and has reached the present.
▼SAE International Releases Updated Visual Chart fO R Its “Levels of Driving Automation” Standard fO R Self-Driving Vehicles https://www.sae.O Rg/news/press-room/2018/12/sae-international-releases-updated-visual-chart-fO R-its-%E2%80%9Clevels-of-driving-automation%E2%80%9D-standard-fO R-self-driving-vehicles
[Reference] For automatic driving levels, see "What is autonomous driving level? Definition, name, and commercial models? What can you do?"
Public transportation services using autonomous driving buses were started in Switzerland in 2016.Although there are various theories, it seems to be the world's first service for autonomous driving buses in public transportation.
The vehicle was used as a long -term demonstration using the ARMA automatic driving shuttle bus of Navya in France, using a road that can only pass only the permitted vehicles.It seems that they are running in the form of sales operations that collect fees from passengers.
In 2017, the German government revised the Road Transportation Law to enable level 3 on public roads in developed countries.However, at that time, international standards were not united, and the level 3 social implementation was in effect.
In the same year, the Germany Audi announced and released a new "A8" with the "Audi AI Traffic Jam Pilot", which enables level 3 driving, but has forgotten the system based on this situation.。
In Japan, the revised Road Traffic Law and the revised Road Transport Vehicle Law were enacted in 2019, and was enforced in April 2020.It is based on international standards established at the United Nations WP29 in June 2020, and is virtually lifted level 3.
In China, Baidu, a major IT, announced the PROJECT APOLLO = Apollo Plan, an autonomous driving development project utilizing the open software platform.It has become the world's largest entity in which more than 100 developers around the world participate, leading autonomous driving development in China.
Until now, automatic driving buses by gold dragon passenger cars and automatic delivery robots by NEOLIX, and autonomous driving taxis of the first train and Beijing Train Group have reached the mass production stage, and attention will be focused on future developments.
[Reference] For Baidu's initiatives, see "Baidu (BAIDU) Automatic Driving Development Chronology! Promotion of Apollo Planning, Leading the Industry in China".
In December 2018, Waymo, which was spun off from Google's Automatic Development Division, launched the world's first commercial service in Arizona, USA.
Initially, some users provided services under the ride of safety drivers, but gradually expanded, and started introducing unmanned services without safety drivers around December 2019.It seems to be achieved.
[Reference] For Waymo's initiatives, see "Waymo's chronology! Automatic taxi Frontier, Google from Google".
2020年に入ると、自動運転関連スタートアップなどの新規上場が相次ぎ始めた。Velodyne Lidar、Luminar Technologies、Innoviz Technologies、AEyeといったLiDAR開発企業が上場で先行していたが、2021年に入るとTuSimpleやAurO Ra Innovationといった自動運転システム開発企業の上場も始まった。
今後もArFive AIら有力スタートアップの上場が続く見込みで、今後の動向に注目が集まる。また、上場手法としてSPAC（特定買収目的会社）を選択する企業が多く、ソフトバンクグループのように受け皿となる買収用企業を用意する動きが出ている点にも注目したい。
In March 2021, Honda launched the new Legend as the world's first mass -produced car with a level three functions.Although it was sold limited to 100 units by leasing, it seems to have opened a new history.
The "traffic jam pilot", which enables level 3, enables automatic driving during traffic jams on highway.Considering the practical area, there are still many restrictions, but the knowledge obtained from these social implementation contributes to the sophistication of the system.We look forward to future evolution.
[Reference] For Honda's initiatives, see "Honda's autonomous driving strategy summary of the world's first level 3 mass -produced vehicles".
In May 2021, the German Federal Congress passed the Road Traffic Law revision bill to enable level 4.The technical requirements for autonomous cars should provide, driving permissions, and obligations of those involved in driving are clearly specified.It seems to be enforced in 2022.
Details are omitted, but unlike the revision of level 3 o'clock, level 4 implementation is possible.The United States and China, which are currently ahead of level 4 services such as autonomous driving taxis, are providing services with their own states and cities, and the state has not yet been developing legal development.
In the future, there is a possibility that development companies from around the world will gather in Germany, which expects level 4 services and services, so we will continue to watch the trends.
[Reference] For the German autonomous operation method, see "Thorough explanation data of German" autonomous driving method "is also required to preserve 13 items.
The following flow introduced in this article is as follows.
| 年／年代 | 概要 |
| 1939～40年 | 世界博覧会でGMが近未来の自動運転構想を披露 |
| 1958年 | GMが誘導ケーブル式の自動運転技術を披露 |
| 1960年ごろ | スタンフォード大学が遠隔制御技術の研究に着手 |
| 1960年代 | 人工知能搭載ロボットの開発も |
| 1960年ごろ | 日本でも自動運転の研究がスタート |
| 1977年 | 機械技術研究所が世界初のマシンビジョン自動運転車を開発 |
| 1980年代前半 | 米国でAI自動運転車の開発進展 |
| 1984年 | 機械技術研究所の研究がさらに進展 |
| 1980年代後半 | 交通課題解決に向け自動運転の注目度高まる |
| 1990年代初期 | カリフォルニアで高密度車群走行システム開発 |
| 1995年 | カーネギーメロン大学が自動運転車で大陸横断 |
| 1996年 | 上信越高速道路で自動運転デモ |
| 2000年 | 自動運転車5台の隊列走行デモ実施 |
| 2004年 | DARPAグランドチャレンジ開催 |
| 2009年 | グーグルの自動運転開発プロジェクト始動 |
| 2010年代 | 国内外で自動運転スタートアップが続々設立 |
| 2016年 | SAE自動運転レベルが世界スタンダードに |
| 2016年 | スイスで自動運転シャトルバスの本格実証スタート |
| 2017年 | ドイツでレベル3可能にする道路交通法が改正 |
| 2017年 | 「プロジェクトアポロ」始動 |
| 2018年 | 世界初の自動運転タクシーの商用サービススタート |
| 2020年 | スタートアップやベンチャーの上場ラッシュ |
| 2021年3月 | レベル3を実装したホンダ「レジェンド」発売 |
| 2021年5月 | ドイツでレベル4可能にする道路交通法改正案が成立 |
Although the autonomous driving method has changed significantly, it has been found that research and development utilizing computer visions, which leads to modern times, has been promoted early, and Japanese researchers were also active in Japan and overseas.
In addition to Stanford University and Carnegie Melon University, institutions that have high achievements in automatic driving development have been accumulated for a long time, and nested engineers have been working on DARPA (US Defense Higher Research Planning Bureau) and Google today.It is also very interesting that the competition of autonomous driving development is driven.
80 years after the GM's Automatic Driving Initiative, more than 40 years have passed since the system development utilizing computer visions, the concepts and future technologies depicting the future have become reality, reaching a practical use.In the 2020s, legislation has begun to progress.
In particular, due to the progress of AI technology, the development speed since the 2010s has been tremendous and continues to evolve in the daily routine.Attention is continuing to the initiatives of the autonomous driving industry, where developments that repain their history every day.
[Reference] As a related article, refer to the "List of List of Automatic Driving (the latest version in 2021)".