A Biological Model for Robotic Cognitive Mapping and Global Planning

Hippocampus place cells and entorhinal grid cells have been supposed to be able to integrate path, and perform spatial representation. In early studies, neural network model and only hippocampal models are used for cognitive map building. And episodic memory, storing information as state neuron and event nodes, expresses the spatial cognition through tracing back and recalling. This paper proposes a novel cognitive computation model, which integrates grid cell, place cell and episodic memory. The model of grid cells and place cells provides coordinate information computed from raw odom data to the events in the episodic memory model. And we use visual cues of environments to detect loop closure. And a navigation approach based on episodic memory was proposed. A shortest path can be chosen using events reorganization. Experiments performed on a mobile robot show that cognitive maps of the real environment can be efficiently built. And the experiment results indicate that the robot can efficiently navigate to target destinations.