A Multi-Scale Map Method Based on Bioinspired Neural Network Algorithm for Robot Path Planning

• Published in: IEEE access Vol. 7; pp. 142682 - 142691
• Main Authors: Luo, Min, Hou, Xiaorong, Yang, Simon X.
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 2019; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; bioinspired neural network; Biological neural networks; Biological system modeling; Dijkstra algorithm; Dijkstra's algorithm; Heuristic algorithms; Mathematical analysis; Multi-scale map method; Neural networks; Path planning; Real-time systems; Robots; Terrain; Three-dimensional displays; Virtual environments
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2019.2943009

With the wide application of Bioinspired Neural Network in the field of robot path planning, the environmental scale of robot path planning is getting larger, and the environmental resolution requirements are getting higher. However, with the increase of the environment size and resolution requirement, the neuronal activity value calculation cost and the time cost of the Bioinspired Neural Network will increase sharply. Aiming at this problem, this paper proposes an improved Bioinspired Neural Network path planning method based on Scaling Terrain. Using a Multi-Scale Map method and Dijkstra algorithm, the optimal path of a Coarse Scale Map is calculated. The optimal path obtained from the Coarse Scale Map is used to guide the neural network planning weights of the Fine Scale Map from the same terrain. Thus, the optimal path of the Fine Scale Map can be calculated by the improved BNN algorithm. Introducing this Multi-Scale Map Method into the Bioinspired Neural Network can greatly reduce the time cost of the Bioinspired Neural Network path planning algorithm and reduce the mathematical complexity. Simulation results in some computer integrated virtual environments further demonstrate the superiority of this method and the experimental results are encouraging.