摘要
针对自主机器人的动态避障问题,借鉴异联想记忆Hopfield神经网络对样本模式的记忆能力和强化学习解决问题的突出能力提出了一种新的融合学习方法即异联想记忆神经网络——强化学习方法。通过在强化学习中引入记忆来增强学习方法的能力,可以使自主机器人快速和适应的学习,从而实现机器人的动态避障。仿真结果表明了该避障方法的有效性。
Aiming at the problem of obstacle avoidance in the dynamic environments of autonomous robot .The paper presents a new mixed learning method, hetero associative memory artificial neural network- reinforcement learning by using the ability of memory of hetero associative memory Hopfield artificial neural network and the outstanding ability of solving problem of reinforcement learning. Strengthen the ability of learning method by adding memory into reinforcement learning.It can make the fast and accommodative study of autonomous robot. Thus carry out the dynamic environments of the robot to avoid the obstacles.The simulation results show that the obstacle avoidance method is effective.
出处
《微计算机信息》
北大核心
2008年第26期196-197,234,共3页
Control & Automation
