摘要
类脑导航是一种受生物大脑空间导航神经机制启发的仿生智能导航技术。在经典的类脑导航系统RatSLAM中,认知地图以拓扑结构形式存在,物体之间的关系通过欧几里得坐标来组织。然而,RatSLAM系统在使用Dijkstra算法进行全局路径规划时存在需要遍历全部节点的问题。因此,文章提出了改进的A*算法,通过模仿哺乳动物大脑中的栅格细胞矢量导航,可以有效利用认知地图中的欧几里得坐标,提升了路径规划效率。改进的A*算法首先利用欧几里得启发函数有选择地搜索节点。其次,对实际代价和估计代价进行了归一化,以平衡两者对搜索结果的影响,旨在解决算法的局部最优问题。在仿真环境和公开数据集上进行的实验表明,在保证最优路径的前提下,改进的A*算法显著减少了路径规划所需的时间。
Brain like navigation is a biomimetic intelligent navigation technology inspired by the neural mechanisms of biological brain spatial navigation.In the classic brain like navigation system RatSLAM,cognitive maps exist in the form of topological structures,and the relationships between objects are organized through Euclidean coordinates.However,the RatSLAM system has the problem of needing to traverse all nodes when using Dijkstra’s algorithm for global path planning.Therefore,the article proposes an improved A*algorithm that mimics the grid cell vector navigation in mammalian brains,effectively utilizing Euclidean coordinates in cognitive maps and improving path planning efficiency.The improved A*algorithm first uses the Euclidean heuristic function to selectively search for nodes.Secondly,to solve the local optimum problem of the algorithm,the actual cost and estimated cost are normalized to balance their impact on the search results.Experimental verification conducted in simulation environments and public datasets shows that the improved A*algorithm significantly reduces the time required for path planning while ensuring the optimal path.
作者
董骁
李翔宇
张志慧
DONG Xiao;LI Xiangyu;ZHANG Zhihui(Shenyang Ligong University,Shenyang 110158,China;Shenyang Institute of Automation,Chinese Academy of Sciences,Shenyang 110017,China)
出处
《计算机应用文摘》
2024年第15期121-125,共5页
Chinese Journal of Computer Application
基金
辽宁省本科教改优质教学资源建设与共享项目(SBKJGYZ-2021-06)。
