Search

To improve the comfort of information cognition and safety in expressway long-tunnel exit, and to study the variation of drivers' visual features in expressway long-tunnel exit, medium passenger vehicle was used to perform driving test in real tunnel environment. Parameters of visual features were recorded using eye tracking system when vehicles traversed the section from 200 meters inside the tunnel to another spot 200 meters outside the tunnel. Mathematic models of drivers' fixation duration, the number of fixations, and saccade amplitude in the tunnel exit were established based on BP Neural Network (Error Back Propagation Network) simulation. Results showed that as the vehicle moving closer to the tunnel exit, fixation duration and the number of fixations gradually decreased, while saccade amplitude increased. After drivers left the tunnel exit, fixation duration slightly decreased and the number of fixations increased, while saccade amplitude was slightly bigger than that inside the tunnel, only with litter amplitude variation.