The severity of a road departure event strongly depends on the features around the roadway: trees and other "hard" fixed objects represent a severe collision hazard, steep slopes nearby may be representing rollover threats or frontal impact hazards, and sharp road-edge drop-offs may exist that prevent high-speed road recovery. But the near-road area may also contain traversable shoulders and medians, such that a road departure can be a fully safe and recoverable event. This paper presents a simulated road departure warning system, sensitive to severity of hazards based on near-road terrain geometry analysis and subsequent threat assessment. To serve as a demonstration, many random 3D models of a highway road and near-road features following AASHTO guidelines were generated. Near-road features were subdivided into three categories corresponding to high, medium or low severity based on their geometries. We assume that geometric parameters of features used in this study are available from pre-processed LIDAR or other map data. Due to unavailability of threat correlation between different types of features, a relative threat index defined as Normalized Average Severity index is used to determine threats associated with a feature. To simulate a driver-warning system, geometries and hazards were tagged with different colors on the generated 3D model. The 3D model is designed to serve as an additional visual warning system for the driver providing information about risk zones nearby the present vehicle position.
展开▼
