VISUAL ODOMETRY FOR LANE-CHANGE DETECTION IN VEHICLE LOCALIZATION

NURLIZA BINTI MAHLI, - VISUAL ODOMETRY FOR LANE-CHANGE DETECTION IN VEHICLE LOCALIZATION. [Final Year Project Report] (Unpublished)

PDF (Please get the password by email to repository@unimas.my , or call ext: 082-583914/3973/3933) Nurliza Binti Mahli.pdf Restricted to Registered users only Download (2MB)

Abstract

This project describes the visual odometry for lane-changing detection in-vehicle localization. The main objective of this research is to analyze and detect whether a vehicle changes lanes on the road based on the visual odometry trajectory in the KITTI dataset. Nowadays, the Global Positioning System (GPS) applications are easy to use and most used in vehicle localization. In the GPS application, there is no stating that drivers change lanes on the roads. This affects the smooth transition and path planning for those using the in-vehicle navigation system. Therefore, this research analyses and determines whether vehicles have made lane changes based on the visual odometry even if a minor change occurs. This analysis has several methods to detect lane change. The visual odometry method used was the ORB SLAM, then from the visual odometry trajectory, the method of curve fitting is used to analyze any change or discrepancy in the trajectory path. The lane changes can be detected by using graph results. The evidence of the graph result shows that the lane changes occurred. The results with different methods can detect a small change in straight or curved road lanes. The main advantage of this study is that it can detect the occurrence of lane changes on a straight road or a curve lane even if there is only a small change.

Actions (For repository members only: login required)

View Item