基于FNN的机器人局部路径规划中实际隶属函数的推导

Establishment of Actual Membership Functions Based on FNN Local Path Planning of Robot

机器人路径规划中建立符合客观实际的隶属函数,是动态环境下应用模糊神经网络解决避碰问题的一个难题。结合动态环境下机器人路径规划的实际,在着重比较研究两种具有实际意义的隶属函数的基础上,综合考虑二维直角坐标体系下机器人、障碍物的位置、速度及运动方向等实时信息,推导出一种新的具有实际含义的隶属函数,有助于判断表征机器人和各移动障碍物之间碰撞的可能性。

In the dynamic environment, the establishment of actual membership functions based on FNN local path planning of robot is difficult to solve. Taking the actual robot path planning into dynamic environments, based on comparative study of two kinds of meaningful membership functions, and based on comprehensive real - time information of the robot and the obstacle among two -dimensional Cartesian coordinate system, such as the position, velocity and direc- tion of movement, this paper deduced the practical implications of a new membership function. These conclusions help to show the possibility of collisions between the robot and moving obstacles.