Hybrid Steering Model Estimated by Particle Swarm optimization Based on Driver’s Eye Tracking Information

The driving model is important for understanding driver’s condition and designing a comfortable automatic driving. It is well known that the PID gain of the model is assumed to be varied depending on driver’s gazing distance. The aim of this study is to estimate the precise steering model depending on gazing distance using Particle Swarm optimization (PSO). In previous researches, the steering model in a single curve was estimated assuming an ideal condition, however, the estimation of the model on the curvy road or with an inattentive driving have not established yet. In this paper, the estimation method of the steering model by PSO depending on the gazing point is applied to the curvy road with an inattentive driving. The eye tracking information is important to understand the driver’s will. To acquire the accurate driving model, the steering model with real-time measurement of gazing point and a hybrid model switching according to the effective visual field are estimated simultaneously by PSO considering driver’s eye tracking information. The effectiveness of proposed method is evaluated by using HONDA driving simulator.