基于神经网络的移动机器人路径规划方法

Path planning methods of mobile robot based on neural network

针对动态环境下移动机器人路径规划,提出了一种基于递归神经网络的实时路径规划方法。利用神经网络表示机器人的工作空间,每个神经元都只有局部侧连接。目标点位置神经元具有全局最大的正活性值,该活性值通过神经元之间的局部侧连接逐渐衰减地传播到整个状态空间,障碍物及其周围区域神经元活性值则被抑制为零。目标点全局地吸引机器人,障碍物局部地将机器人推开实现避障,从而能够在动态环境下产生最优规划路径。仿真结果表明该方法具有较好的环境适应性和实时性。

To investigate the path planning methods of mobile robot in dynamic environment, a method is proposed based on recurrent neural networks in real-time environment. The arrangement of the neurons coincides with the discretized representation of configuration space. The target neuron has the maximal positive neural activity, which is damply promulgated to the whole state space via local lateral connections of neurons. The activities of the neurons, in obstacle fields and the local neighborhoods, are made to zero. The target globally attracts the robot, and the robot can avoid obstacles locally. The robot can generate the optimal trajectory in dynamic environment. Simulation results demonstrate that the method has high adaptability to dynamic environment and real-time ability.