CASIA OpenIR  > 毕业生  > 博士学位论文
仿生机器鱼游动与滑翔运动的建模与控制研究
袁俊
学位类型工学博士
导师谭民
2017-05-23
学位授予单位中国科学院研究生院
学位授予地点北京
关键词仿生机器鱼 仿生机器海豚 滑翔运动 动力学建模 运动控制
其他摘要       仿生机器鱼和机器海豚的研究旨在为高性能水下航向器的设计与控制提供重要的理论基础和技术支撑。与水下生物相比,机器鱼及机器海豚在运动性能和续航能力等方面还存在较大差距,还需更为深入的探究。本文围绕机器鱼及滑翔机器海豚的系统设计、动力学建模及运动控制开展研究,主要内容如下: 
       一、针对机器鱼平面游动数据获取问题,从位姿测量、时间基准和速度估计三个方面考虑,提出了一种机器鱼运动测量方法,为后续游动建模提供数据支撑。给出了一种运动跟踪算法,实现了机器鱼位姿的准确测量。提出了一种时间同步算法,利用机器鱼的多连杆结构和关节运动规律,将板载控制系统和运动测量系统之间的时间同步问题形式化为一个优化问题,并求解了关键时间参数。在此基础上,采用平滑样条曲线实现了位姿平滑和速度估计。 
       二、针对具有复杂流线型的机器鱼,提出了一种数据驱动的动力学建模方法。采用Morrison方程和Strip方法分析了流体作用力,通过反向递归推导出了动力学模型的显示表达。在建模方法中融入了水动力参数辨识,从机器鱼的运动数据中直接辨识出水动力参数,避免了复杂流线型带来的水动力参数难以确定的问题。通过水下实验验证了所提方法的有效性。进一步,给出了一种基于中枢模式发生器(Central Pattern Generator,CPG)振荡器相位差的游动性能优化方法,分析和对比了CPG相位差一致和非一致两种情况下的游动性能。 
       三、为了解决机器海豚续航能力不足的问题,将浮力驱动机制引入到机器海豚上,给出了一种潜深300 m、体长1.5 m的滑翔机器海豚设计方案。针对性能要求和实际环境中水压等因素,从整体外形、关节配置、密封方式和材料选型等方面进行了总体设计。进一步,给出了各模块的机械结构设计,开发了控制系统的软、硬件,完成了滑翔机器海豚样机的研制,实现了多模态运动控制,并通过实验验证了其多模态运动能力。 
       四、针对三维滑翔运动,构建了全状态动力学模型,并提出了一种基于可动鳍面的姿态控制方法。首先,结合滑翔机器海豚鳍面可动的特点,构建了三维滑翔动力学模型,并验证了模型的有效性。然后,结合实验数据探讨了胸鳍和尾鳍对于滑翔运动的调节作用。进一步,提出了一种基于胸鳍差动的偏航控制方法,主要包括三个部分:(1) 针对滑翔速度难以测量的问题,设计了滑模观测器来估计运动速度;(2) 基于backstepping方法推导了控制律;(3) 针对水动力的非线性,设计了胸鳍偏置角解算器,可在偏航控制的同时消除耦合的运动。采用相同的系统框架,给出了基于尾鳍的俯仰控制方法。最后,通过仿真和实验验证了所提姿态控制方法的有效性。;     The research on bio-inspired robotic fish and robotic dolphins aims to lay theoretical and technical foundations for design and control of high-performance underwater vehicles. There is still a large gap between robotic fish/dolphins and their biological counterparts in the aspects of motion performance and endurance, which deserve further investigation. This dissertation mainly concerns system design, dynamic modeling and motion control of a robotic fish as well as a gliding robotic dolphin. The technical contributions are summarized below. 
     Firstly, the motion measurement problem of robotic fish is investigated in terms of pose acquisition, timebase unification, and velocity estimation, in order to provide data support for subsequent swimming modeling. A vision-based tracking algorithm is designed for precise motion measurement. Thereafter, a time synchronization method is proposed to unify the timebase between the onboard motion control system and the motion measurement system. Based on the multi-link structure and motion law of the robotic fish, the time systems are synchronized by solving a formalized nonlinear optimization problem. Moreover, smoothing splines are utilized to smooth the obtained pose sequence and estimate continuous motion states. Experiments are conducted to validate the proposed method. 
    Secondly, a data-driven dynamic modeling method is proposed for robotic fish with irregular geometric profiles. Fluid forces are analyzed via the Morrison equation and the strip method. Backward recursion is implemented in dynamic analysis, and a dynamic model with an explicit formulation is derived consequently. Further, the parameter identification technique is integrated into dynamic modeling, which reshapes it with data-driven feature and overcomes the difficulty in acquisition of hydrodynamic parameters. Aquatic experiments demonstrate the effectiveness of the method. In addition, we present an approach of enhancing swimming performance via optimizing oscillator phase differences of a central pattern generator (CPG) model. Particularly, two conditions are studied and compared: consistent and inconsistent phase differences. 
    Thirdly, an improved gliding robotic dolphin with a body length of 1.5 m and a diving depth of 300 m is designed, which integrates the buoyancy-driven mechanism to achieve strong endurance. Considering the performance requirements, we implement the overall design in aspects of streamlined profiles, joint configurations, mechanical seal, and materials. Further, the mechanisms are constructed, and the hardware as well as the software of the control system are also developed. Experiments are carried out to validate the multi-modal motion capabilities of the built robot. 
    Fourthly, a full-state dynamic model for three-dimensional (3D) gliding motion is established, and an attitude control method is proposed for the gliding robotic dolphin. In comparison with traditional underwater gliders, the robot's flippers and fluke are controllable and thereby enrich the state-regulating modes of gliding motion, which is considered specially in the dynamic model. The built model is validated by data collected from gliding tests. Thereafter, we investigate the regulating effect of the controllable fins on gliding motion, by virtue of experimental data. On this basis, a heading control method is proposed, in which the heading direction is regulated by differential actions of the flippers. The framework of the control algorithm is composed of three main components: (1) a sliding mode observer is designed to deal with the difficulty in velocity sensing; (2) a heading control law is derived via the backstepping methodology; (3) a solver is designed to convert the controller's instruction to deflection angles of the flippers, which can simultaneously eliminate some coupled but undesired movements. Based on a similar framework, a fluke-based control algorithm is also implemented for pitching regulation. Both simulations and experiments are performed to verify the effectiveness of the proposed attitude control method.
文献类型学位论文
条目标识符http://ir.ia.ac.cn/handle/173211/14664
专题毕业生_博士学位论文
作者单位中国科学院自动化研究所
推荐引用方式
GB/T 7714
袁俊. 仿生机器鱼游动与滑翔运动的建模与控制研究[D]. 北京. 中国科学院研究生院,2017.
条目包含的文件
文件名称/大小 文献类型 版本类型 开放类型 使用许可
thesis_final.pdf(35899KB)学位论文 暂不开放CC BY-NC-SA请求全文
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[袁俊]的文章
百度学术
百度学术中相似的文章
[袁俊]的文章
必应学术
必应学术中相似的文章
[袁俊]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

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。