Visual odometry supported inertial navigation system

This final year project implements a high-accuracy navigation system for autonomous vehicles that assists GPS when it is not able to function properly. It integrates the visual data from the camera with data from the INS to predict the precise position of the vehicle. Real-time video from the camera is taken to compare the frames of the video, estimating the translation and rotation. INS will output the velocity and displacement, and fused together by Kalman Filter, the new location of the vehicle is estimated independent of GPS. A prototype was built on NVidia Jetson.