Effect of Input Data on Model Accuracy in NARX Driver Model
NARXドライバモデルにおける入力データがモデル精度に及ぼす影響
- Delivery
- Available on this site
- Format
- Price
- Non-members (tax incl.):¥1,100 Members (tax incl.):¥880
- Publication code
- 20225207
- Paper/Info type
- Proceedings (Spring)
No.47-22
- Pages
- 1-5(Total 5 p)
- Date of publication
- May 2022
- Publisher
- JSAE
- Language
- Japanese
- Event
- 2022 JSAE Annual Congress (Spring)
Detailed Information
Author(J) | 1) 長妻 治志, 2) 及川 昌子, 3) 廣瀬 敏也 |
---|---|
Author(E) | 1) Harushi Nagatsuma, 2) Shoko Oikawa, 3) Toshiya Hirose |
Affiliation(J) | 1) 芝浦工業大学, 2) 芝浦工業大学, 3) 芝浦工業大学 |
Affiliation(E) | 1) Shibaura Institute of Technology, 2) Shibaura Institute of Technology, 3) Shibaura Institute of Technology |
Abstract(J) | 本研究は,先行車両の制動に対するドライバの減速,停止動作に着目し,NARX(Nonlinear Auto Regressive eXogenous)にてドライバモデルの構築を行った.実験は,先行車の速度,減速度を変更した複数のシナリオを用いた.そこで,最適パラメータの検討,学習に用いた実験シナリオと異なる実験シナリオをモデル化に使用した場合のモデル精度について検討した. Translation |
Abstract(E) | In this study, we focused on the driver's braking behavior in response to the braking of the forward vehicle, and constructed a driver model using NARX (Nonlinear Auto Regressive eXogenous). Since NARX uses current and past data as the input, it has an advantage to construct a driver model with time series data. We used several scenarios in the experiment, in which the speed and deceleration of the forward vehicle were set as experimental parameters. We investigated appropriate parameters for model construction, and the accuracy of the model when different experimental parameters were used as the training data. |