A Study on Short-Term Future Vehicle Speed Prediction using Gaussian Mixture Regression
混合ガウスモデルによる回帰を用いた短期将来車速予測に関する研究
- Delivery
- Available on this site
- Format
- Price
- Non-members (tax incl.):¥1,100 Members (tax incl.):¥880
- Publication code
- 20225206
- Paper/Info type
- Proceedings (Spring)
No.47-22
- Pages
- 1-6(Total 6 p)
- Date of publication
- May 2022
- Publisher
- JSAE
- Language
- Japanese
- Event
- 2022 JSAE Annual Congress (Spring)
Detailed Information
Author(J) | 1) 全 翔澤, 2) 金 淳暁, 3) 眞田 一志, 4) 梅津 創, 5) 西尾 唯 |
---|---|
Author(E) | 1) Sangtaek Jun, 2) Kim Soonhyo, 3) Kazushi Sanada, 4) Hajime Umezu, 5) Yui Nishio |
Affiliation(J) | 1) 横浜国立大学, 2) 横浜国立大学, 3) 横浜国立大学, 4) 本田技研工業, 5) 本田技研工業 |
Affiliation(E) | 1) Yokohama National University, 2) Yokohama National University, 3) Yokohama National University, 4) Honda Motor, 5) Honda Motor |
Abstract(J) | 本研究では,車速とペダル操作量などの個車情報と,制限車速や停止標識などのルート情報をもとに,ドライバーの数秒先の要求車速の予測に機械学習を適用した.個車の走行状況をシーンに分類し,混合ガウス回帰を適用した.変数増減法を利用して入力変数を選定することで,短期将来車速の予測が可能であることを示した. Translation |
Abstract(E) | In this study, machine learning was applied to predict the required vehicle speed several seconds ahead of the driver based on individual vehicle information such as vehicle speed and pedal operation amount, and route information such as speed limit and stop sign. The driving situation of individual vehicles was classified into scenes, and Gaussian mixture regression was applied. It was shown that it is possible to predict the short-term future vehicle speed by selecting the input variables using the forward and backward stepwise selection method. |