Estimatoin of risky driving situation based on driver gaze behavior modeling

2017 Fiscal Year Final Research Report

• Project/Area Number: 15K00231
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Perceptual information processing
• Research Institution: Nagoya University
• Principal Investigator: Chiyomi Miyajima 名古屋大学, 未来社会創造機構, 特任准教授 (90335092)
• Co-Investigator(Kenkyū-buntansha): 武田 一哉 名古屋大学, 情報学研究科, 教授 (20273295)
• Project Period (FY): 2015-04-01 – 2018-03-31
• Keywords: ドライバ / 視行動 / 運転操作行動 / 危険度推定 / 隠れマルコフモデル / 車線変更

Outline of Final Research Achievements

We proposed a method to quantify the safety of driver gaze behavior based on the relationship between the gaze directions and the positions and the risk levels of surrounding vehicles. Experimental results showed that driving instructors exhibit safer gaze behavior than other drivers. We also proposed a method of modeling the sequences and co-occurrence of driver gaze behavior and vehicle operation behavior. We modeled safe and risky lane change maneuvers based on gaze directions and discrete acts of vehicle operation behavior using hidden Markov models. We found that there were significant differences between the model parameters of safe and risky behavior. We then detected risky lane changes using the models. The integrated model of gaze and vehicle operation behavior achieved better risky lane change detection performance than the gaze-only and vehicle operation-only models.

Free Research Field

運転行動情報処理