This is part of the RobotsprogrammeBack
Bumblebees have 1 million brain cells, which is 250-times fewer than we have in one retina. And yet, they solve the visual problems that our most sophisticated computers cannot solve: for example, recognising the same flower from different perspectives under different conditions of illumination. How is this possible?
Research on simple systems such as bumblebees has the potential to explain the biological principles that are common to all visual animals, including humans. By creating model synthetic systems, we can elevate that understanding to another level, as there are certain questions that cannot be answered by studying humans, or indeed any biological system.
Evolving virtual bees – called ‘apiants’ – is one of the major scientific research themes of lottolab, since it has the potential to not only explain vision, but also to create highly robust Machine Vision and Robotic Vision systems.
Our approach combines research on the colour vision of real bumblebees raised in a highly controlled environment (see Bee Matrix,)and physiological research on the functional structure of their brains with mathematical techniques (computational evolution, non-linear dynamics, complexity and information theory, Bayesian statistics, etc.) in order to evolve and analyse virtual bees that solve the same challenges in the same way the natural bees do.