Brain-like path planning algorithm based on spiking neural network

Le résumé fourni par la source

This paper proposes a brain-like path planning algorithm based on spiking neural networks (SNNs), which draws on the neural phenomena of hippocampal place cells and related navigation behaviors in neuroscience. By simulating the pulse sequence propagation characteristics of spiking neurons, the algorithm realizes the inference and search of achievable paths in the spatial navigation of intelligent agents. The propagation characteristics of the pulse sequence enable the activation response of each spiking neuron to accurately simulate the path reasoning and selection process in space. In addition, this paper designs a local spatial neural stimulation queue to control the subtask distribution of the path planning task. The queue effectively improves the computational efficiency and reduces the redundant search space by updating the achievable space and local path planning tasks related to the agent task in real time, further improving the accuracy of path planning. By dynamically decomposing the global path planning task into multiple local subtasks, the algorithm realizes efficient task allocation and path integration, significantly improving the path search efficiency. Experimental results show that the algorithm can effectively complete the path planning task in environments of different complexity, especially in dynamic environments and complex obstacle scenes, showing high flexibility and adaptability. In addition, the proposed algorithm shows good biological interpretability through the path integration behavior similar to biological intelligence. This algorithm has high practicality and provides a novel and potential path planning solution for fields such as robot control and intelligent agent navigation, and can show significant performance improvements when handling large-scale space tasks.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.