Towards combining a neocortex model with entorhinal grid cells for mobile robot localization

Motion and navigation are fundamental abilities of all terrestrial animals. It is essential for foraging, reproduction, and more generally for survival. There are a couple of strategies to conduct navigation from simpler visual homing in ants to more complex and cognitive demanding techniques in mammals. Many species of mammals use several specialized cell types in the hippocampus and the entorhinal cortex to represent space in the brain like head direction cells to encode their orientation and grid cells to keep track of their position. In our recent work, we presented MCN - an algorithm that is inspired by working principles of the human neocortex for the navigational subtask visual place recognition. MCN makes decisions based merely on camera data without odometry about whether or not a currently visited place has been seen in the past. In this work, we intend to answer the question if we can combine our neocortex-inspired model with entorhinal cortex cells for space representation to exploit additional metric data like odometry in our system. We believe that the combination of bio-inspired techniques could help someday to create a biologically plausible and more robust navigation system like in animals. In this paper, we give an introduction to our neocortex-inspired algorithm MCN and to two cell types of the entorhinal cortex, answer how these concepts can be combined to perform visual place recognition, and provide proof-of-concept experiments with a mobile robot to show the performance of the proposed system.