基于人工神经网络实现智能机器人的避障轨迹控制

利用人工神经网络中的二级BP网，模拟智能机器人的两控制参数(左、右轮速）间的函数关系，实现避障轨迹为圆弧或椭圆弧的轨迹控制，并且通过调整椭圆长、短轴大小，能实现多个及多层障碍物的避障控制．该方法的突出特点是方法简单、算法容易实现，使机器人完成多个及多层避障动作时，不滞后于动态环境里其它机器人（障碍物）位置的变化．在仿真实验中，取得了理想的效果．

TRAJECTORY CONTROL WITH OBSTACLE AVOIDANCE OF MOBILE ROBOTS BASED ON NEURAL NETWORK

The proposed method in the paper simulates the function connection of two control parameters (left and right wheel velocity ) of the mobile robots by two layer BP network, and implements the trajectory control for trajectory equation being an arc of circle or ellipse. It can implement trajectory control with obstacle avoidance for multi number or multi layer obstacles by adjusting the long axes and short axes of the ellipse in length. The evident features of the proposed method are simple in means and are implemented easily in algorithm, so that the completion of obstacle avoidance don't lag the position change of other robots (obstacles) in the dynamic environment. The method is effectual in simulation.