轮式机器人鲁棒仿生自主学习算法的研究  

Research of Robust Biomimetic Autonomous Learning Algorithm on Wheeled Robot

作　　者：史涛[1,2] 杨卫东[1] 任红格[2] 

机构地区：[1]北京科技大学自动化学院,北京100083 [2]河北联合大学电气工程学院,河北唐山063009

出　　处：《计算机测量与控制》2014年第4期1209-1211,共3页Computer Measurement &Control

基　　金：国家自然科学基金项目(61203343)

摘　　要：针对本质不稳定的轮式机器人运动平衡问题,基于操作条件反射原理,结合鲁棒控制,提出了一种鲁棒仿生自主学习算法作为轮式机器人的学习机制;该算法利用鲁棒控制可以提高系统抑制干扰的能力,产生最优控制行为的特点,将其与操作条件反射原理相结合,使机器人通过与未知环境的交互、学习和训练,模拟生物操作条件反射机制以及自学习和自适应技能,实现对轮式机器人的运动平衡控制,并从理论上证明了算法的稳定性;最后,分别用该算法在无干扰和有干扰的两种情况下对机器人做了仿真实验并进行了比较,结果表明,鲁棒仿生自主学习算法能够使机器人获得自主学习和平衡控制的技能,并体现出了较好的学习性能抗干扰能力。Aiming at the nature unstable movement balance problems of the wheeled robot, the robust bionic autonomous learning algo- rithm is proposed as the wheeled robot learning mechanism based on the operant conditioning principle and the robust control. The algorithm uses the characteristics of the robust control, which can improve the system ability at suppressing interference and produce the optimum con- trol behavior. Combining with the operant conditioning principle, the robot can simulate the biological operant conditioning mechanism, as well as the self--learning and self--adapting abilities through the interaction and learning and training with the unknown environment, and a- chieve the movement balance control of the wheeled robot. And the stability of the algorithm has been proved theoretically. Finally, this pa- per uses this algorithm to make the simulation experiments in both cases of the absence of interference and interference respectively, and the compared results show that the robust bionie autonomous learning algorithm can make the robot obtain the skills about autonomous learning and balance control, and reflects the better studying and anti--interference ability.

关 键 词：鲁棒控制 操作条件反射 仿生自主学习 运动平衡控制 轮式机器人 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置]