Research

My research is in computer vision, especially robotic vision and in machine learning for robotic and computer vision.

Lifelong Learning for SLAM

I am interested in building visual SLAM systems that are long-lived, accurate, and semantically meaningful. This research aims to take SLAM maps beyond the purely geometrical to semantic ones that are long-lived, dyanmic and queriable. I envisage a system where multiple mobile visual sensors continuously and in real-time update a geometric and semantic representation of an environment such as in a construction site, a mine, or an urban environment.

Centre of Excellence in Robotic Vision

My group is part of an Australian Research Council funded Centre of Excellence ($25M over 7 years 2014-2021) condusting basic ansd applied research in visual perception for robots. The inability to "see" their enviroments is a fundamental roadblock to further deployment of rorobits in areas where the environment is uncertain or changing, and which have the potential to impact areas such as healthcare, environmental and infrastructure monitoring, safety in construction, smart manufacturing and driverless cars.

Within the Centre I and my group focus mainly on Semantic Representation and Scene understanding, as well as on fundamental learning, especially deep learning.

For more information see the Centre's web pages

Multi-target Tracking

Under construction

Deep Learning

Under construction