One of the most important perceptions of a living being is vision. As easy as it seems for us to recognize our environment, we still don't know much about the human vision system.
For a few decades, researchers have been able to investigate how visual information is processed in the retina and the visual cortex. Complex models have been built, drawing their foundation both from neurophysiology, as well as psychological empirical-obtained data.
However, as the analytical research grows, very few applicable models have shown to function in the real world. Most of the existing models only aimed at representing and evaluating the different theorems, disregarding their real-world applicability – being in terms of efficiency, robustness and generality.
We aimed to close this gap by investigating how effective biologically-inspired vision system can be realized on humanoid robots. The aim is to supply the humanoid with the ability to visually percept and understand its environment.
Our approach accounts for two objectives:
- to advance our scientific understanding of human (biological) vision; and
- to better enable humanoid robots with a human-like vision system.