Variation of Drivers' Visual Features in Long-Hunnel Entrance Section on Expressway

摘要

To avoid traffic accidents in long tunnel entrance sections, the authors studied the variation of driver's visual features based on real road experiments on the expressway. Drivers' visual feature parameters were recorded in real-time using Eye link (eye tracking system) during the driving test. Mathematic models of drivers' fixation duration, the number of fixations, and saccade amplitude in tunnel entrance were established based on BP Neural Network (Error Back Propagation Network) simulation. Results showed that fixation duration increased gradually as the vehicle moving closer to the tunnel entrance, whereas the number of fixations and saccade amplitude decreased. Meanwhile, drivers' fixations shifted from straight ahead to the right side, which resulted in the number of fixations on the right side increased. After drivers entering the tunnel, fixation duration firstly decreased and then increased afterward, while the number of fixations and saccade amplitude kept increasing.