Vehicle positioning using lidar

Project Leader: Kourosh Khoshelham
Primary Contact: Kourosh Khoshelham (k.khoshelham@unimelb.edu.au)
Keywords: computer vision; deep learning; geomatics; spatial computing; spatial information
Disciplines: Infrastructure Engineering
Domains:

Accurate positioning is important for autonomous vehicles and mobile mapping systems. Current positioning methods based on Global Satellite Navigation System (GNSS) are challenged in densely built-up areas, urban canyons and tunnels, where GNSS signals are not available. In the absence of GNSS signals, simultaneous localization and mapping methods using images and lidar data have been used for estimating the position of the vehicle. However, these methods are susceptible to the drift of the estimated trajectory, and are therefore suitable only for short periods of GNSS signal outage.

This PhD project aims to develop a method for vehicle positioning by integrating imagery/lidar data captured by the vehicle with information extracted from an existing 3D map or model of the environment.