Computer Vision PhD Scholarship, Australia

Expressions of interest are sort for a PhD candidate to work on a project starting in late 2011 or early 2012 to develop a high-speed light-weight embedded vision system for robotics and computer vision applications.

Autonomous and semi-autonomous robotic vehicles, including aerial, terrestrial and marine vehicles,

are a key technology for a wide range of emerging applications in society. Robotic vehicles that function in changing and dynamic environments require sensor systems that provide the vehicle with robust and rich three dimensional information on the environment, enabling the vehicle to avoid collisions and move in a natural manner. Vision is arguably the most robust, light-weight, low-cost and information-rich sensor system available. However, vision sensors generate prohibitive amounts of data, especially at the high frame rates that are required to control dynamic systems such as aerial robotic vehicles, and vision data requires significant processing before it can be used in any given application. Dealing with vision data transmission and managing fast vision processing is arguably the limiting factor in the implementation of vision based control of robotic systems.

This project will address the fundamental science and technology for embedded vision systems for robotic applications. The key philosophy of the project is that a vision sensor for robotic applications should be embedded, that is, designed to to be fitted to a mobile vehicle and provide considerable processing of the data locally (in the sensor unit) with only a fraction of the data, but a majority of the important information, passed back to the central control system. The student will be involved in developing vision processing algorithms for optic flow computation and estimation of the local three dimensional environment that are tailored to high speed vision and implementable on light-weight low power processors. Integration of inertial data into the embedded vision system is a crucial aspect of the project. Low dimensional representation of the visual information is also a crucial aspect of the project that will make exploiting the data practical. This work will be grounded in the development of a prototype small-scale lightweight low-power embedded vision sensor for use on mobile robotic vehicles.

A full APA equivalent scholarship is available for the successful students and international students will be considered. The successful candidate will have a strong interest and preferably prior background in embedded systems, FPGAs, and vision systems. The project will have a significant practical aspect with opportunity for theoretical contribution.

The funding for this project is forecast for later in 2011. At the moment the chief investigators are seeking expressions of interest with the intention of recruiting quickly when the funding is finalised.

Interested students should contact Dr David Austin or Prof. Robert Mahony for further information.

Further Scholarship Information and Application