A bio-inspired model reliably predicts the collision of approaching objects under different light conditions

Abstract

In this paper, we present a model of the Lobula Giant Movement Detector, which is a part of a visual pathway responsible for triggering collision avoidance manouvres in the locust Locusta Migratoria. Also based on locust neural adaptation to transitions in light intensities, the model proposed here integrates a mechanism for light adaptation. The tests performed with the model demonstrate its ability to reproduce several characteristic properties of the LGMD response, including the firing rate profile for different visual stimuli. Additionally, results obtained for different light conditions show that the increase in the LGMD model efficiency is provided by the new mechanism of light adaptation. In here, the LGMD is presented as an ideal model to develop sensors for automatic collision detection. © 2012 Springer-Verlag.