Bionic hippocampus cognitive map construction method based on convolutional neural network

Abstract

The invention provides a bionic hippocampus cognitive map construction method based on a convolutional neural network, and the method can complete the construction of a cognitive map and guide the navigation of a robot. The method belongs to the technical field of robot environment cognition and motion navigation, and aims to solve the problem that a closed-loop detection module based on mouse brain hippocampus navigation is poor in robustness in a complex, changeable and high-repeatability environment at present. The specific process comprises the steps of firstly, in the environment exploration process of the robot, collecting a speed image, a direction image and an RGB image; inputting the speed and direction into a position sensing module to obtain real-time position information of therobot, and inputting the RGB image into a visual processing module to obtain image characteristics of the surrounding environment of the robot; and finally, associating the position information and the environment information of the robot and storing the position information and the environment information to cognitive map network nodes to complete construction of a cognitive map. And in the environment exploration process of the robot, when a closed loop is detected, position correction is carried out by using position information associated with the current image.

Description

technical field [0001] The invention belongs to the field of brain-like computing and intelligent robot navigation. Specifically, it involves a method of combining the convolutional neural network and the hippocampal spatial cell computing model to jointly construct a bionic hippocampal cognitive map, and uses the method of image similarity measurement to realize the positioning and position correction of the robot in a familiar environment. Background technique [0002] With the development of robots, more and more intelligent robots are needed in more and more places, such as autonomous environment modeling, intelligent recognition, scene memory, and object-oriented navigation in dynamic and complex environments. [0003] Organisms in nature have a strong cognitive ability to complex environments. They can understand the external environment and form specific mental representations through observation and learning. When facing familiar scenes again, they can recognize the ...

AI Technical Summary

Problems solved by technology

[0006] The present invention aims to solve the problem that in the spatial cognition navigation model, the traditional manual feature extraction method is used for template matching, and the robustness is poor in complex, changeable and repetitive environments, and it is easy to cause wrong matching.

Images