Robust Lane Change Detection and Tracking in Urban Environment
Saumya Srivastava, Rina Maiti · 2024
Lane detection in autonomous vehicles is crucial for safe navigation, requiring robust methods capable of handling diverse road scenarios, including lane changes. This paper introduces a novel approach integrating adaptive angle constraints and a mark- ROI technique for effective lane change detection and tracking in urban environment, which is capable of providing departure warning signals to the driver and recognizing lane changes. The method involves a series of preprocessing steps to ensure accurate lane localization. These steps include generating an edge map, reducing noise through orientation constraints at the sub-image level, applying the Hough transform for line detection, followed by a mask-based feature extraction to isolate lane markers. A tracking mechanism utilizing a Kalman filter is then employed to maintain detection continuity despite occlusions or missing lane markers. The results demonstrate that the proposed method can detect lane markings in real time across various complex environment.