摘要: |
This project will mine an existing set of radar data surrounding real-world lane change events executed by drivers relying on both conventional mirror and camera-based systems. The data set provides valuable opportunities to develop computer-based algorithms for dealing with and managing radar traces to identify normative lane change signatures as well as conflict-based events (inappropriate lane changes, or lane changes executed with small-time gaps). This research is expected to greatly contribute to the development of automated and partially automated driving systems by 1) Developing and validating algorithms using radar trace data to classify “safe” and “unsafe” lane change situations which may be used to guide the implementation and management of automated lane change systems, 2) Helping to develop automated lane change systems that naturally mimic a good driver’s performance thereby increasing driver acceptance and comfort, and 3) Development of warnings to drivers operating with partially automated systems under situations where drivers need to assume control and guarding against inadvisable lane changes. Understanding how drivers manage lane changes under manual driving situations (e.g., time-to-collision judgments, conflicts, etc.) can therefore greatly enhance and aid in the development and implementation of automated lane change and driver warning systems. |