中国科学院机构知识库网格
Chinese Academy of Sciences Institutional Repositories Grid
多模态仿生两栖机器人运动实现与行为控制

文献类型:学位论文

作者丁锐
学位类别工学博士
答辩日期2011-05-27
授予单位中国科学院研究生院
授予地点中国科学院自动化研究所
导师谭民
关键词两栖机器人 CPG 控制 多模态运动 动力学建模 传感反馈 Amphibious robot CPG control multimodal locomotion hydrodynamics modeling sensory feedback
其他题名Locomotion Implementation and Behavior Control of a Biomimetic Amphibious Robot with Multi-Mode Motion
学位专业控制理论与控制工程
中文摘要两栖机器人作为一种新型的仿生机器人,由于兼具陆地、水下以及近海滩涂等环境下的运动功能,近年来受到了越来越多研究者们的关注。本文针对两栖机器人的运动实现和行为控制进行研究,针对地面爬行和水下游动建立了一种基于传感信息的复合CPG(Central Pattern Genrator–中枢模式发生器)控制模型,通过调整输入激励、方向因子、俯仰因子等控制参数,可实现地面和水下的多模态运动控制;此外,考虑到机器人的水下自由游动与流体作用力密不可分,将其简化为多连杆串联机构,对其进行了流体动力学分析,并将CPG模型控制律作为动力学模型的关节角信号,建立了一种联合的运动学-动力学模型,探讨了CPG参数与游动性能之间的关系。为了增强两栖机器人的环境适应性,构建了基于红外传感信息的状态聚类和避障策略,为两栖机器人在动态环境下的多模态行为自主运动奠定了基础。 本文在CPG建模、两栖多模态运动控制、流体动力学建模、传感信息耦合等方面开展研究,主要研究内容如下: 首先,根据两栖机器人的运动步态和功能需求,提出了一种集地面仿轮式爬行、水中仿鱼/仿海豚式复合推进于一体的多模态运动推进方案。基于模块化机器人的设计理念,推导了一种具有N个模块化单元数的地面最小半径转弯控制方法。同时,利用虚拟样机技术对机器人的地面直行、转弯、爬坡等进行了仿真模拟,为机构设计和优化提供了参考,在此基础上设计实现了两栖机器人样机并进行了地面实验验证。 其次,对于水中运动,利用刚体动力学理论,建立了两栖机器人身体多个关节摆动和胸鳍拍动的拉格朗日动力学方程;考虑到机器人的细长体结构,其在水下游动的主要作用力简化为:连杆和胸鳍正压力、头部截面阻力和形体摩擦阻力,结合拉格朗日第二方程推导出了水下游动的动力学模型。将基于CPG模型的关节控制角作为动力学模型的输入信号,建立了两栖机器人水下游动的联合运动学-动力学模型,并进行了Mathematica环境下的动力学仿真和实验研究,包括直游、转弯、仿鱼和仿海豚推进比较等。仿真和实验结果表明了该动力学模型的有效性。 第三,对于两栖多模态运动控制,针对水陆不同的驱动机构和运动形式,提出了耦合液位传感反馈的CPG控制方法。根据液位传感器采集到的不同环境信息,调整身体左右两侧振荡器之间的相位关系,完成地面关节锁定和水下往复摆动两种不同的振荡方式,实现了地面和水下的不同运动形式。在此基础上,借助于一对灵活的机械胸鳍机构,实现了俯仰、倒游、原地转等多模态运动,并建立了两栖机器人多模态运动的特征行为参数集。 第四,基于两栖机器人的自主避障控制,根据三个红外传感器采集到的障碍物距离信息,进行了细化分析,建立了避障模糊控制规则表,并对模糊控制输出进行聚类分析,将其结果耦合到振荡器幅度方程,提出了一种耦合红外传感反馈CPG的自主避障控制策略,仿真和实验结果表明了该避障策略的有效性。 最后,对所开展的工作进行了总结,并指出了进一步的研究工作。
英文摘要As a novel biomimetic robot, the amphibious robot has advantages in transcendent locomotor ability and excellent terrain adaptability both on ground and in water as well as in ocean surf zones, which has drawn great interest from researchers. This paper focuses on the locomotion implementation and behavior control of a biomimetic amphibious robot and has established a Central Pattern Generator (CPG) based control architecture coupled with sensory feedback. By modulating the control parameters involving input drive, direction factor, attack angle factors, the robot can perform multimodal locomotion. In terms of rigid-body dynamics, the robot swimming can be simplified as a locomotor organism of multi-link connected rigid body immersed in an incompressible fluid and the experienced hydrodynamics has been detailedly investigated. Consequently, the CPG-based control laws are then incorporated into the model serving as explicit joint angle control, where a simultaneous kinematics and dynamics system has been established to elicit the effect of CPG parameters on swimming performance. To enhance adaptive behaviors to environmental conditions, an infrared-based intelligent obstacle avoidance approach has been formulated by clustering state information, which lays the foundation for designing an amphibious robot capable of multimodal autonomous locomotion under dynamic environment. This paper mainly deals with the CPG modeling, amphibious multimodal locomotion control, hydrodynamics modeling, sensory integration, etc. The technical contributions of this paper are summarized below. Firstly, a propulsion mechanism integrating wheel-based crawling on ground, fish- or dolphin-like swimming underwater, has been proposed in terms of locomotor gaits and functional requirements. Based on the modular design philosophy, a steering approach for turning control on ground has been derived for a robot with N joints. Meanwhile, virtual prototype technology has been utilized to simulate crawling, turning and slope climbing on ground which will provide some insight into mechanism design and optimization. As a consequence, physical prototypes have been developed to perform preliminary testing on ground. Secondly, for the swimming locomotion, a Lagrangian reduction has been formulated in terms of rigid-body dynamics, which simplifies the robot as an oscillating multi-link serial mechanism coupled with a pair of pectoral fins of flapping motion. The Lagrange’s...
语种中文
其他标识符200818014628003
源URL[http://ir.ia.ac.cn/handle/173211/6349]  
专题毕业生_博士学位论文
推荐引用方式
GB/T 7714
丁锐. 多模态仿生两栖机器人运动实现与行为控制[D]. 中国科学院自动化研究所. 中国科学院研究生院. 2011.

入库方式: OAI收割

来源:自动化研究所

浏览0
下载0
收藏0
其他版本

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。