Toward long range robot navigation

Following the concepts of bio-inspired robotic and a constructivist approach we present here integrated robotic control architectures resulting from a close feedback loop between experiments on animals and robots. Robust control architectures for mobile robot navigation in both indoor and outdoor a priori unknown environment are developped. This also leads to a better understanding of the mechanisms by which the brain processes spatial information. From a computational neuroscience view-point, our control architectures are based on a functional model of hippocampo-cortical interactions implicated when rats solve complex navigation tasks. After a short review of previous models, we highlight the difficulty to scale these control architectures to large environment. We propose to overcome these limitations with a new bio-robotic architecture modeling grid cells.