"AI Congestion Prediction" From July 28, 2022 [E14] Started demonstration experiment on Keiyo Road
-Delivery time is one hour ahead of schedule and delivered at 13:00 every day-
- Corporate Top
- Press Room
- Press Release Headquarters
- "AI Congestion Prediction" From July 28, 2022 [E14] Started demonstration experiment on Keiyo Road
July 27, 4th year of Reiwa
East Nippon Expressway Co., Ltd.
NTT DoCoMo, Inc.
East Nippon Expressway Co., Ltd. (hereinafter referred to as "NEXCO EAST") and NTT DoCoMo, Inc. (hereinafter referred to as "DOCOMO"), CA Tokyo Wan Aqua-Line Expressway (hereinafter referred to as Aqua Line) and E17 Regarding the *1 "AI traffic jam prediction" that has been implemented so far in Kan-Etsu Expressway (hereinafter referred to as Kan-Etsu Expressway), From Thursday, July 28, 2022, we will start a demonstration experiment at E14 Keiyo Road.
In addition, DOCOMO's newly developed technology that speeds up the process from the creation of demographic data to the prediction and distribution of forecast information by AI will advance the delivery time of the forecast by one hour from the same day to 14:00 to 13:00, and the service will be improved so that you can plan the afternoon schedule earlier.
1. 1. What is AI Congestion Sight?
ドコモが持つ携帯電話ネットワークの仕組みを利用して作成されるリアルタイム版モバイル空間統計※2（以下、人口統計）と、NEXCO東日本が保有する過去の交通量・渋滞・規制などの実績データをもとにドコモが開発した「AI渋滞予知」技術、さらにNEXCO東日本の交通工学的知見・ノウハウを掛け合わせることで、当日の人出から所要時間や交通需要※3を予測しドライバーに配信しています。
「AI渋滞予知」は、約9割のお客さまから高評価をいただいています（別紙【PDF：885KB】参照）。
2. Why Keiyo Road?
The Keiyo Road is a heavy traffic route that connects central Tokyo and Chiba Prefecture, and is one of the routes that causes a lot of traffic congestion even within the NEXCO EAST jurisdiction.
- Feature (1): Since the user characteristics are different on weekdays and holidays, the traffic congestion situation changes in a complicated manner.
- Weekdays ... Mostly used for commuting and business purposes
- Holidays ... Mostly used for leisure and sightseeing purposes
- Feature (2): Interchange intervals are short, and inflows and outflows to and from general roads occur frequently.
- Feature ③: There are inflows and outflows from other lines in the same Chiba prefecture, such as the Higashi-Kanto Expressway and the Chiba-Togane Road.
NEXCO EAST and DoCoMo have been considering expanding "AI traffic jam prediction" to Keiyo Road, hoping that AI technology will be used to improve the accuracy of traffic jam prediction and help customers avoid traffic jams. This time, by expanding the technology of "AI congestion prediction" provided by Aqualine and Kan-Etsu Expressway-etsu Expressway and analyzing the increase and decrease of demographics in more detail, the technology to predict traffic demand even under such complicated conditions. We have succeeded in newly establishing the technology and realized the introduction of "AI traffic jam prediction" on the Kan-Keiyo Road.
In addition, "AI Congestion Prediction" on Keiyo Road applies to "Soga IC-Shinozaki IC" on the In-bound line.
Prior to this demonstration experiment, when comparing the estimated travel time with the actual travel time for the past (979 days) traffic conditions from the Keiyo Road Soga IC to the Shinozaki IC, the number of days was an error of 30 minutes or more. Is 14 days (1.4%) for "AI congestion prediction" compared to 115 days (11.7%) for "conventional forecast" * 4, which is a significant improvement in accuracy (decrease in the number of days with an error of 30 minutes or more). Was confirmed.
Comparison of maximum error in one day
| Maximum error in one day | Conventional prediction | AI traffic jam prediction |
|---|---|---|
| 50 minutes or more | 6 days (0.6%) | 1 day (0.1%) |
| 40 minutes or more | 27th (2.8%) | 3 days (0.3%) |
| 30 minutes or more | 115 days (11.7%) | 14th (1.4%) |
| 20 minutes or more | 311 days (31.8%) | 71 days (7.3%) |
| 10 minutes or more | 767 days (78.3%) | 287 days (29.3%) |
- The maximum error in one day was the maximum on the day by comparing the estimated required time every 30 minutes after 14:00 with the actual driving time based on the past traffic conditions between the Keiyo Road Soga IC and Shinozaki IC. Refers to the value of the error
- Time required when traveling between Soga IC and Shinozaki IC at legal speed: Approximately 29 minutes
- Evaluation target: January 1, 2017 (Sun) -February 28, 2022 (Mon) (979 days excluding accidents and regulations)
3. 3. 1 hour ahead of delivery time
DoCoMo's newly developed "Real-time AI Social Infrastructure Technology * 5" will speed up the process from creating demographic data to forecasting by AI and distributing forecast information so that you can plan your afternoon schedule earlier. So, the delivery time of the forecast will be advanced by one hour.
We would like to ask our customers to cooperate in avoiding traffic congestion through distributed use, such as adjusting departure times and adding places to stop by, referring to the forecast information in "AI Congestion Prediction".
| until now | Deliver forecast information every 30 minutes after 14:00 on the day at 14:00 [Provided routes] Aqua Line, Kan-Kan-Etsu Expressway |
| from now on | 13:00 Deliver forecast information every 30 minutes after 14:00 on the day [Provided routes] Aqua Line, Kan-Etsu Expressway, Keiyo Road |
Four. How to use AI congestion prediction
Please use the two-dimensional barcode on the right to access, or click the banner from the top page of NEXCO EAST 's website "DraPla".
[NEXCO EAST | NTT Docomo] AI traffic jam prediction
The NEXCO EAST Group has positioned the period from 2021 to 2025 as "a period that contributes to the achievement of the SDGs and transforms into a new future society," and is making various efforts.
In addition, DOCOMO is working with partners to create new value in the new DOCOMO Group Medium-Term Strategy.
Both companies will continue to consider further utilization of "AI congestion prediction" and work to resolve traffic issues.
- [November 30, 2017 Press Release]
NEXCO EAST, NTT DoCoMo, and CA Tokyo Wan Aqua-Line Expressway start traffic jam prediction demonstration experiment by "AI traffic jam prediction"
[December 18, 2019 Press Release]
Predict traffic congestion on the E17 Kan-etsu Kan-Etsu Expressway with "AI Congestion Prediction"! - Domestic distribution statistics (real-time version), which is one of the lineup of mobile spatial statistics. This information indicates the number of people in a group for each area or attribute, and cannot identify individual customers. The demographics used in this experiment comply with the "Mobile Spatial Statistics Guidelines," which summarizes the basic items for creating and providing mobile spatial statistics, in order to strictly protect the privacy of our customers.
Mobile Spatial Statistics Guidelines - Traffic demand is the number of vehicles that potentially pass through the Expressway at each time of day, and corresponds to the amount of traffic when there is no limit to the amount of traffic (traffic capacity) that the road can carry.
- The conventional forecast is a traffic jam forecast created and published by NEXCO EAST based on past traffic jam records.
- Technology for high-speed large-scale AI processing by compactly compressing large-scale demographic data over a wide area with high efficiency and using the compressed data directly for processing.
To view the PDF file, you need the Adobe Systems plug-in software "Acrobat Reader (Japanese version)". If you do not have Download from here (free)Please use it.