On-Road Vehicle Trajectory Collection and Scene-Based Lane Change Analysis: Part II

摘要: This two-part paper aims to study lane change behaviors at the tactical level from an on-road perspective, with a special focus on analyzing interactions between ego and surrounding vehicles during procedure. Part I addresses vehicle trajectory collection, whereas II extraction scene-based behavioral analysis. Different general technique of moving object detection tracking, collection for driving behavior is required have properties consistency, completeness, continuity, accuracy. proposes system where instrumented developed multiple horizontal 2-D lidars that 360° coverage. The software by fitting laser points all model using coupled estimation features reliability along frames achieve accurate state estimations occluded data robust association in multiviewpoint sensing. performance investigated extensively, large set through Fourth Ring Road Beijing total distance 64 km, more than 5700 environmental trajectories length over 19 h. demonstrated be high quality terms properties. To authors' knowledge, this first able automatically collect all-around demonstrate good providing high-quality database studies perspective real-world traffic level.