摘要
为了解决生物激励神经网络算法在全覆盖路径规划中陷入死区的问题,提出了脱困点搜索和脱困路径规划组合的脱困机制。建立了工作区域的栅格模型,分析了生物激励神经网络算法原理和缺陷。通过设计元胞演化规则,给出了基于元胞自动机的最佳脱困点搜索方法。对于脱困路径规划问题,传统RRT算法的采样和扩展具有随机性和盲目性,提出了神经元活性引导RRT算法,使RRT算法的随机树扩展具有较强的方向性。经仿真验证,与传统RRT算法相比,神经元活性引导RRT算法的耗时减少了一个数量级,扩展节点数减少了2倍,脱困路径减少了12.96%,是一种非常高效的脱困方法。另外,具有脱困机制的生物激励神经网络算法能够完成工作区域全覆盖,有效解决了死区问题。
In order to solve the dead zone problem of biologically inspired neutral network on complete coverage path planning,escape mechanism consisting of escaping point search and escaping path planning is proposed.Grid model of working area is built.Principle and shortcoming of biologically inspired neutral network are analyzed.Through designing cellular evolution rule,optimal escaping point searching method based on cellular automata is given.For the problem of escaping path planning,sampling and extension of traditional RRT are random and blind.To solve the problem,RRT guided by neuronal activity is put forward,and which make random tree extend directionally. It is clarified by simulation that,compared with traditional RRT,RRT guided by neuronal activity time-cost decreases by one order of magnitude,expanding point quantity decreases by 2 times,and escaping path length decreases by 12.96%,which means escape mechanism is effective. Besides,biologically inspired neutral network with escape mechanism can accomplish complete coverage path planning,and the problem of dead zone is solved availably.
作者
江静岚
JIANG Jing-lan(Liuzhou Railway Vocational Technical College,Guangxi Liuzhou 545616,China)
出处
《机械设计与制造》
北大核心
2022年第6期295-299,304,共6页
Machinery Design & Manufacture
基金
2017年度广西高校中青年教师基础能力提升项目(2017KY1236)。
