Abstract: An improved model of neural network named “tactile sensation perception” is proposed for mobile robot to find a path in unknown environment with static and dynamic obstacles.In the proposed model，robots only need distance and orient of the target and information of neighbors，which is collected by inner sensors without the information of map.BP neural networks is the decision center of robot.Robot can provide a proper trail with free-collision in the process of exploring in both static and dynamic environments after training.With simulations，the experimental results demonstrate the efficiency of the method.

Key words: neural networks, robot, path-finding, dynamic environment

摘要： 提出了一种新的路径搜索算法——“触觉感知法”来实现机器人在未知静态与动态环境情况下的路径搜索。该方法不需要提供地图信息，机器人仅收集目标点的距离和方位信息以及通过自带传感器作为触觉器收集周围局部环境信息。机器人以BP神经网络作为决策器，经过训练，可以在静态和动态环境中搜索出一条光滑无碰撞且便捷并能有效避开动态障碍物的运动轨迹。对所提出的方法进行了仿真实验，仿真结果表明算法在静态和动态环境下均能有高效率的路径搜索表现。

关键词: 神经网络, 机器人, 路径搜索, 动态环境