Hiroshi Inou

Design of a Preview Driver Model Based on Optical Flow

Automated driving systems have been released in recent years to improve the active safety. In this research field, a preview/predictive driver model has been widely used as a steering controller and for driver behavior analysis. This driver model has been implemented by many researchers and, through the vehicle simulations and experiments, has been shown to reproduce driver behavior, such as the drivers steering and handle torque. On the other hand, optical flow is a part of visual information, and in our previous works we focused on it as the index used in human locomotor behavior. Humans are known to perceive the direction of their self-motion from optical flow and utilize it during driving. In experimental applications of the optical flow model to an in-vehicle system, the system was shown to be effective in practical use. In particular, the system can express the same driver behavior as the preview driver model. In this study, the preview driver model was analyzed from the viewpoint of optical flow to understand the preview driver models interpretation of human behavior. In addition, in this paper a new preview driver model that utilizes our knowledge of optical flow is proposed. This model shows a better control performance and ride quality than the conventional preview driver model. As a result, this approach serves as a foundation of the preview driver model.

Optical flow-based control for automatic steering systems

Recently, various driving support systems have been developed to improve safety. However, because drivers occasionally feel that something is wrong, systems need to be designed on the basis of the information that drivers perceive. Therefore, we focus on optical flow, which is one of the types of visual information used by humans to improve their driving feel. Humans are said to perceive the direction of self-motion from optical flow and utilize it during driving. In our previous works, applying the optical flow model to automatic steering systems, a human-oriented system that simulates human behavior might be able to be developed. In this paper, a nonlinear controller considering camera motion is proposed to apply the optical flow model to an in-vehicle control system. The nonlinear controller is implemented to an in-vehicle control system, and in-vehicle studies are conducted. Then, experimental results are presented.