Research
Theme 2: Vision for Action and Robotics
Theme Leader: M. Srinivasan
Nature has equipped animals and humans with vision systems par excellence. This is readily apparent to anyone observing a fly deftly avoiding the swat of a hand, a bee returning unerringly to its hive after finding food several kilometres away, or a cricketer executing a brilliant running catch in the outfield. Man-made machines and robots continue to perform such tasks with far less finesse. We are thus faced with two major and intriguing challenges:
(a) what are the cues and strategies that natural vision systems exploit to solve complex visuomotor tasks? and
(b) can we use some of these biologically inspired strategies to design novel algorithms for machines that see, perceive, steer and navigate?
Work in this theme is being done in CI laboratories at the ANU and the Universities of Queensland and Sydney, with contributions from the CSIRO ICT Centre (Brisbane), Seeing Machines (Canberra), the Universities of London, Zurich and Bielefeld, the Swiss Federal Institute of Technology (Lausanne), and the Yerkes Primate Center (Atlanta). Work in this theme will benefit strongly from interaction with projects in Theme 1 in relation to understanding the neural basis of perception and behaviour. Equally, projects in Theme 1 will benefit from the insights provided by the behavioural analyses in Theme 2.
The specific projects in this theme are:
Integration of form based motion signals
Critically testing the role of the polarization compass in honeybee navigation
Wind compensation in honeybee flight
Visual Cognition in honeybee navigation
Neural basis of odometry in honeybees
Characterization of the spectral and polarization properties of fiddler crab photoreceptors
The structure and processing of natural optic flow in insects
