Queensland Brain Institute 2010 Annual Report - University of ...

Laboratory Head Professor Mandyam SrinivasanFlying insects display remarkable visual agility, despite their relatively smallbrains and simple nervous systems. Professor Mandyam Srinivasan’s groupuses honeybees and budgerigars as models to understand how visualinformation is used to guide flight and facilitate navigation. Another aimis to explore whether some of these insights can be used to devise novel,biologically inspired strategies for the guidance of autonomous aerialvehicles.Above: Srinivasan group photo. 2010 Laboratory Members: Samuel Baker, Parthasarathy Bhagavatula, Daniel Bland, Natalie Bland, Brenda Campbell,Nikolai Liebsch, Tien Luu, Eliza Middleton, Richard Moore, Navid Nourani, Dean Soccol, Saul Thurrowgood, Gavin Taylor, William Warhurst. Above right:A tethered bee flying in a virtual-reality apparatus for investigating the visual mechanisms of flight control.PG | 48Translating flight findings into aeronautical applicationsAlthough research has revealed how insectsuse vision to guide their flight, little is knownabout whether flying insects actively avoidmid-air collisions and, if so, how they achievethis. The Srinivasan group’s recent study ofbees flying through narrow passages showsthat collisions occur at a much lower rate thanpredicted by chance, suggesting that beesactively avoid them. Three-dimensional reconstructionof the trajectories shows us howcollisions are avoided.Orchestrating a smooth landing is a challengingtask for any animal or machine, particularlyin the presence of wind. After establishingthat bees tend to land facing into the wind,we discovered in 2010 that bees do this evenwhen the substrate is devoid of visual texture,suggesting the direction of landing is guidedby cues additional to optic flow. We also foundthat bees are able to orient their landings atwind speeds as low as 0.8 m/sec, indicating ahigh sensitivity to this external disturbance anda remarkable ability to compensate for it.We filmed and analysed the flights ofbudgerigars to uncover strategies birds useto fly safely through dense and clutteredenvironments. Some findings parallel thosepreviously observed for flying insects,suggesting that certain principles of visualguidance may be shared by all diurnal, flyinganimals.In the area of biologically inspired robotics, wehave designed, developed and successfullytested vision systems that implementautomatic and efficient changes of aircraftaltitude and attitude, use horizon-basedinformation to perform a range of extremeaerobatic manoeuvres (such as loops) andenable a vehicle to detect the presence ofsalient landmarks in its environment fromthe patterns of image motion generated in acamera image.PhD student Samuel Baker with amulti-rotor aircraft that will embodybiologically inspired algorithms fornavigation.