机器人姿态路径目标航迹优化仿真  

Optimization Simulation of Robot Attitude Path Target Track

作　　者：宫海洋 任红格 史涛 李福进 GONG Hai-yang;REN Hong-ge;SHI Tao;LI Fu-jin(College of Electrical Engineering,North China University of Science and Technology,Tangshan Hebei 063210,China)

机构地区：[1]华北理工大学电气工程学院,河北唐山063210

出　　处：《计算机仿真》2018年第7期285-291,共7页Computer Simulation

基　　金：国家自然科学基金(61203343);河北省自然科学基金(E2014209106);河北省高等学校科学技术研究青年基金项目(QN2016102;QN2016105)基金资助

摘　　要：针对机器人姿态路径目标航迹仿真中一些传统机器学习方法不能进行连续行为学习的问题,通过模拟人类心理认知机理与脑部神经运动现象,提出了一种基于小脑-基底核-大脑皮层回路的认知发育算法,解决机器学习中不能连续学习的问题。上述算法以学习自动机为框架,以生物感觉运动系统为理论基础,使小脑通过监督学习方式实现机器人状态到行为的映射;然后通过基于内在动机原理的概率选择评价模块对基底神经节动作进行选择;最后将大脑皮层接受的神经信号传递给基底核与小脑,形成完整的感觉运动反馈回路。将该算法应用到机器人姿态路径目标航迹系统进行优化仿真。仿真结果表明,机器人在未知环境下,通过自主学习发育,逐渐掌握运动平衡控制技能,并实现目标的实时跟踪,与经典认知发育算法进行了对比分析,体现了该算法的有效性和鲁棒性。In order to study the continuous behavior of two-wheel robots and simulate human psychological cogni- tive mechanism and brain neurological phenomenon, a robot path optimization algorithm was proposed. The algorithm takes the learning automaton as the framework and the bio-sensory motor system as the theoretical basis to enable the cerebellum to realize the mapping of the robot state to the behavior through supervised learning. Then it selects the basal ganglia actions through the probability selection module based on the intrinsic motivation principle. Finally, the neural signals received by the cerebral cortex are transmitted to the basal nucleus and cerebellum, forming a complete sensory feedback loop. The algorithm was applied to the two-wheeled robot system to carry out experiments. The sim- ulation results show that under the unknown environment, the robot gradually acquires the skills of motion balance control and achieves real-time tracking of the target through classical learning development. The comparative analysis shows the effectiveness and robustness of the algorithm.

关 键 词：内在动机 感觉运动系统 平衡控制 认知发育 两轮机器人 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]