Abstract
In this work, eye-movement data, recorded while operating a driving simulator, was used for assessing visual behavior. The driving scene is based on modified video-clips containing various composite hazards, such as a pedestrian crossing just in front of a car. In addition, most of salient objects in the scene (i.e. pedestrian, vehicle, traffic signal, and so on) have been tagged manually with both object type and position at each frame. The gazing object was identified by matching the eye-tracking data with the tag information.
We measured and compared eye-movement data among three groups; older (N=10, mean 68.7 years), middle-aged (N=7, mean 41.4 years) and a driving instructor (N=1, 35 years). To assess visual behavior while driving, we defined two measures as follows: 1) number of fixations (NOF) on the specific objects to which drivers must allocate visual attention in the frontal view; 2) latency of visual awareness (LOA) of the suddenly appearing hazard to which drivers must respond in order to avoid a collision. Additionally, we investigated the changes of NOF after giving subjects simple advice on traffic rules in order to find possibilities to improve their visual behavior.
Experimental results revealed that an older group had significantly lower NOF values, approximately corresponding to two thirds of the value obtained from the middle-aged group, a half of that from the instructor, and six out of ten older drivers had an improvement of NOF after advice. Secondly, an older group had significantly larger LOA value, approximately twice of middle-aged group´s and instructor´s, especially in the case of hazards appearing in the peripheral area, although there was smaller difference of LOA in the central area. Thirdly, latency of stepping on the brake pedal after viewing the hazard was significantly slower, approximately twice of the value from the middle-aged group. These results suggest that the older drivers have poor visual behavior to distribute their attention and have difficulties in finding suddenly appearing hazards in the peripheral view. However, this could be improved with simple advice on traffic rules. Potential applications of this research include driver assessment and training systems for older drivers.
Keywords:active safety, visual behavior, older driver, driving simulator, attention