2020 年 86 巻 889 号 p. 20-00190
As the authority of the self-driving system becomes large, the driver’s contribution to the safety decreases involving drowsiness or inattention to the driving. A sudden disengagement of self-driving system may result in a critical incident. Therefore, a monitoring system of the driver’s condition is required for self-driving car of level 2 or 3. We analyzed the eye movement model of driver watching preceding vehicle aiming at the application to the estimation of attentional state. Input and output of the eye movement model are the position of the preceding vehicle and the eye movement, respectively. The eye movement models were identified through driving simulator experiments. The relations between the model parameters and the attention degree to the preceding vehicle, which was varied by subtask, were studied. Statistical analyses revealed that the steady gain of eye movement model correlates with the attention degree of driver. This correlation can be applied to the future estimation system of driver’s condition.
