Development of Human-Like Driving Decision Making Model based on Human Brain Mechanism

Tsuyoshi Sakuma, Satoshi Miura, Tomoyuki Miyashita, Masakatsu G. Fujie, Shigeki Sugano

研究成果: Conference contribution

抄録

Recent driving assistance technologies such as Electronic Stability Control (ESC) and auto brake system release drivers from complicated driving tasks. On the other hand, there is concern that it reduces pleasure feelings of a driver if these system's behaviors are different from the driver's intention. To avoid such problem, it is important to evaluate the driver's intention and decision-making process, and design the assistance system to fit it. In this research, we propose an unsupervised reinforcement learning driver model based on human cognitive mechanism and human brain architecture. Because this study's objective is to analyze the process of driving decision making, we hire a simple actor-critic model as a driver model. We set learning parameters from the driver's decision making characteristics which are derived from the task execution process of the human brain, and set state space from driver's sensory characteristics. This driver model can predict lane change decision making adequately and shows high accuracy (ACC=94%) on verification tests with real driving data. This result is similar to unpublished results of a deep neural network driver model which use the same data as teaching data. From these results, we consider that the proposed reward function and learned state space represent the driver's decision making characteristics.

本文言語English
ホスト出版物のタイトルProceedings of the 2019 IEEE/SICE International Symposium on System Integration, SII 2019
出版社Institute of Electrical and Electronics Engineers Inc.
ページ770-775
ページ数6
ISBN(電子版)9781538636152
DOI
出版ステータスPublished - 2019 4 25
イベント2019 IEEE/SICE International Symposium on System Integration, SII 2019 - Paris, France
継続期間: 2019 1 142019 1 16

出版物シリーズ

名前Proceedings of the 2019 IEEE/SICE International Symposium on System Integration, SII 2019

Conference

Conference2019 IEEE/SICE International Symposium on System Integration, SII 2019
国/地域France
CityParis
Period19/1/1419/1/16

ASJC Scopus subject areas

  • 原子分子物理学および光学
  • 電子工学および電気工学

フィンガープリント

「Development of Human-Like Driving Decision Making Model based on Human Brain Mechanism」の研究トピックを掘り下げます。これらがまとまってユニークなフィンガープリントを構成します。

引用スタイル