Abstract
Most mammals efficiently overcome self-localization deviations by coordinating grid and place cells in their brain's navigation system. However, the coordination of grid cell modules during spatial navigation and its impact on position estimation are poorly understood. This study addresses this issue by introducing a system that decodes grid-cell module activity and integrates networks of multiple grid-cell modules for self-position estimation in a mobile robot. Our results show that even when individual grid module estimates deviated substantially from the robot's actual location, the modules remained tightly coordinated. Corrections of these deviations were studied based on anchoring the activity of grid cells to spatial landmarks. Detailed numerical investigations indicate that path integration is critically dependent on the intrinsic coordination between grid cell modules which enhances the accuracy and reliability of spatial navigation. Furthermore, we show that this coordination enables effective vector navigation, even when the overall position estimation is inaccurate. These insights advance our understanding of grid-cell module coordination in location estimation during path integration and offer potential applications in robotics.