Brain-like robot navigation method based on Bayesian multimodal perception fusion

Abstract

The invention relates to a brain-inspired robot navigation method based on Bayesian multi-modal perception fusion, comprising the following steps: integrating cells for vestibular information fusion: changing the firing rate of integrating cells according to the acquired vestibular information; calibrating cells for visual information fusion; Inject energy into the grid cell network and the head-to-head cell network, change the firing rate of the integral cell and the calibration cell; perform global suppression; obtain the actual position of the current robot by estimating the phase of the firing rate of the grid cell network and the head-to-head cell network and head orientation; building a topological map. The model designed in the present invention can carry out the fusion of multi-modal perception, and realize the stable encoding of the space environment where the robot is located and the state of the robot itself. The model is consistent with the neural mechanism of head orientation in mammals, similar to the results of single neuron recording experiments collected in neurobiology, and has high biological fidelity.

Description

technical field [0001] The invention belongs to the field of robot navigation, and in particular relates to a robot navigation method inspired by mammalian neurobiology and utilizing Bayesian for multi-mode perception fusion. Background technique [0002] Multimodal information fusion in robotic navigation systems has been extremely challenging. Since the available perception information is usually unreliable and affected by noise, multimodal perception fusion can more accurately encode the pose and environment of the robot. The two main mechanisms by which animals explore their environments are path integration and landmark calibration. To improve the robot's navigation performance, it is necessary to address the uncertainty caused by the accumulation of path integration errors and blurred landmark perception. [0003] The fusion of multimodal perception is the key to the precise perception and behavior of animals. Animals are able to explore long distances, navigate com...

AI Technical Summary

Problems solved by technology

[0005] Although some biologically similar attractor network models have been proposed to demonstrate the mechanism of probabilistic computation in the brain, no neurobiologically inspired Bayesian model of multimodal perception fusion has been used in robot navigation systems.

Images