Visual route navigation in ants: a situated and embodied approach Event

Abstract

The use of visual information for navigation appears to be a universal strategy for sighted animals. One particular group of expert navigators are the ants. The broad cross-disciplinary interest in studies of ant navigation is in part due to their small brains; biomimetic engineers take inspiration from elegant and parsimonious control solutions, while biologists look for a description of the minimal cognitive requirements for complex spatial behaviours. We also take an interdisciplinary approach to studying visual guided navigation of ants where we emphasise tenets familiar to Adaptive Systems practitioners; to get a full understanding of complex behaviour and how it emerges from the interaction of sensory system, body and environment, you must study natural behaviour in the natural habitat.

Behavioural experiments and natural image statistics show that visual navigation need not depend on the recognition of objects. Modelling suggests how simple behavioural routines enable navigation using familiarity detection rather than explicit recall. This leads to a new navigation algorithm which successfully navigates visually complex worlds, with routes that show many characteristics of desert ants. In particular, we show that robust navigation can be achieved with simple computation, after a single training run without specifying when or what to learn. Further, the model replicates data from classic ant navigation experiments. As such, our model represents the only detailed and complete model of insect route guidance to date. I will end by discussing applications to robotic navigation.

 

The talk will run from 4-5pm in B53/4025, Highfield Campus.

All CS4 talks are free and refreshments will be provided from 5pm. For videos of previous talks and interviews and details of future talks please visit:

http://cs4southampton.wordpress.com

To help us manage numbers, if you would like to attend one of our seminar series please can you inform Jo Corsi josephine.corsi@soton.ac.uk