基于神经网络的未知环境路径规划算法研究  

A Neural Networks Algorithm for Robot Path Planning in Unknown Environment

作　　者：倪斌[1] 陈雄[1] 鲁公羽[1] 

机构地区：[1]复旦大学电子工程系,上海200433

出　　处：《计算机工程与应用》2006年第11期73-76,109,共5页Computer Engineering and Applications

摘　　要：文章提出了一种新的神经网络算法——“距离定位法”来实现机器人在未知环境情况下的路径规划。地图采用神经元阵列表示,机器人通过自带传感器收集周围局部环境信息以及与目标点的距离信息,经过对局部神经网络的实时训练,可以快速地产生一条光滑无碰撞且简捷有效的运动轨迹。算法在静态和动态环境下均能有高效率的路径搜索表现。模拟仿真结果也证明了算法的有效性。An improved model of organized neural network is shown for path planning and obstacle avoidance in an unknown map which is represented by topologically ordered neurons.After real-time neural networks training with information of neighbors and distance of target which is collected by inner sensors,robot will autonomously provide a proper path with free-collision and no redundant exploring in the process of exploring in both static and dynamic environment.The computer simulation will illustrate the performance.

关 键 词：神经网络 机器人 路经规划 未知地图 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]