CASIA OpenIR  > 毕业生  > 博士学位论文
面向水下搜索的仿生机器金枪鱼系统优化与运动控制
仝茹
2024-05
Pages134
Subtype博士
Abstract

近年来,水下仿生机器人以其高机动和友好的环境亲和特性成为了海洋环境探索的新型载体。然而,现有的仿生机器鱼与生物鱼类的游动性能仍存在差距,难以完成水下搜索、海洋科考等对运动能力要求较高的任务。自然界中,金枪鱼以其超高的游速和续航能力,启发着仿生机器鱼领域研究。本文围绕高性能仿生机器金枪鱼系统设计和运动控制开展工作,重点研究高速仿生推进的鱼尾结构优化、高性能仿生推进系统设计、高机动运动控制和高时效水下搜索策略等内容,为高游速水下仿生机器人系统研制与运动控制提供重要的理论基础与技术支撑。主要内容如下:
一、针对仿生推进游速优化问题,从尾鳍面积优化和柔性鱼尾优化两个方面,探究了仿鱼摆尾式的高速推进机制,并提出了鱼尾结构优化方法。首先,通过分析尾鳍面积与仿生推进力之间的关系,建立驱动力、尾鳍面积和巡航速度之间的动力学关系,提出了基于驱动力和平衡约束的尾鳍面积优化方法,并通过优化求解和实验测试获得了最优尾鳍面积区间。其次,针对刚柔耦合的鱼尾结构,通过实验探究了不同频率下的鱼尾刚性优化规则,并基于悬臂梁结构分析了柔性鱼尾优化机理。最后,设计了一种具有连续柔顺性、中性浮力和有限收缩性的一体成型被动柔性鱼尾,并针对柔性鱼尾结构,提出基于预测网络的数据驱动建模方法,实验验证了所提模型的实时预测能力和所提鱼尾在游速、能效、运动稳定性方面的性能提升。
二、以金枪鱼为仿生对象,基于高速推进机理,提出了兼具游速和机动性能的仿生机器鱼系统设计方案。首先,面向仿生高速高机动推进需求,设计了仿生流线外形、高频高效仿生推进机构、俯仰机动胸鳍和一体化柔性鱼尾等关键机构,构建了多模态感知系统和控制系统框架,搭建了机械和软硬件系统,完成了样机研制,实现了3.28倍体长每秒(BL/s)的高游速和0.48~BL转向半径的高偏航机动性。其次,面向仿生拍动模式优化,提出了一种归一化多样性中枢模式发生器(ND-CPG),获得了稳定、精准、多样的拍动节律,并以游速为目标开展仿生拍动节律的参数优化。进一步,通过搭建动力学模型,完成了机器鱼运动性能仿真与分析。最后,实验验证了所提新型模式发生器的性能优势和机器鱼平台的综合运动性能。
三、针对俯仰机动运动控制问题,基于构型要素分析,提出了多运动目标模型预测控制和竖直面俯仰机动运动规划方法,实现了俯仰角和深度的耦合控制。首先,从浮力特性、外形水动力以及胸鳍的位置和大小等角度探究了机器鱼在不同游速和姿态下的系统稳定性和可控性,得到高游速机器鱼系统的俯仰机动运动控制所需满足的构型要素。其次,基于数据驱动的动力学模型,构建了多目标模型预测控制器,实现了对俯仰角和深度值的耦合控制,在各游速下均表现出较好的控制性能。最后,基于俯仰机动性约束,提出了一种竖直面运动轨迹生成方法,基于俯仰角--深度值的关键点生成了竖直面运动动作轨迹。实验结果表明,机器鱼能够成功实现俯仰姿态和深度控制,并能够基于生成轨迹完成特定的竖直面运动动作。
四、针对水下搜索任务场景,提出了一种广域仿生鱼眼视觉感知系统,并提出了基于视觉的水下搜索策略和搜索路径规划方法。首先,受生物鱼眼视觉启发,提出了一套仿生鱼眼双目视觉系统,通过视场设计和视野拼接方法,获得了超300°的广域视觉感知。其次,结合水流速、感知信息提出了搜前规划方法,建立了搜索区域的概率模型,并基于有限状态机设计了水下搜索策略。再次,根据视觉能见度、洋流、障碍物等信息提出了基于概率先验的搜索路径规划。最后,搭建水下搜索任务场景,开展了系统集成和实验验证,仿真结果验证了所提水下搜索方法的有效性。

Other Abstract

In recent years, underwater bionic robots have become a new carrier for marine environment exploration due to their high maneuverability and environmentally friendly characteristics. However, there remains a significant disparity between the swimming performance of existing robotic fish and that of biological fish, making it difficult to complete tasks that require high mobility, such as underwater search and marine scientific investigation. In nature, tuna, with its extremely high swimming speed and endurance, inspires research in the field of bionic robotic fish. This dissertation focuses on the design and motion control of a high-performance robotic tuna system, emphasizing the optimization of the tail structure, the design of high-performance bionic robot systems, the control of highly maneuverable motion, and the development of efficient underwater search strategies. The technical contributions are summarized as follows:

Firstly, aiming at optimizing the propulsion performance of fish tail, optimization methods for fish tail structures, focusing on tail fin size and flexibility of fish tail are proposed. By analyzing the relationship between tail fin size and propulsion force, a dynamic model related to driving force, tail fin size, and swimming speed is established. An optimization method based on driving force and equilibrium constraints for tail fin size is proposed, and the reasonable tail fin size range is obtained through optimization solving and experimental testing. For the rigid-flexible coupled fish tail structure, experiments were conducted to explore the optimization rules of tail rigidity at different frequencies. The flexible tail optimization mechanism was analyzed based on the cantilever beam structure. Subsequently, an integral molding flexible fish tail with continuous flexibility, neutral buoyancy, and limited compressibility is designed. For the flexible fish tail structure, a data-driven modeling method based on a predictive network is proposed. Experiments validate the real-time prediction capability of the proposed model and demonstrate performance improvements in swimming speed, energy efficiency, and motion stability of the fish tail.

Secondly, inspired by the physiological structure and swimming patterns of biological tuna, a robotic tuna system design that integrates swimming speed and maneuverability is proposed. Key modules including streamlined shape, high-frequency and high-fidelity actuation mechanism, pectoral fins, and integrated flexible fish tail are designed to meet the demands of high-speed and high-maneuverability propulsion. The perception system and control system framework are established, and the mechanical and software/hardware systems are constructed, culminating in the development of a prototype capable of achieving a high swimming speed of 3.28 body lengths per second (BL/s) and a high yaw maneuverability with a turning radius of 0.48~BL. Focusing on the optimization of flapping patterns, a normalized diversity central pattern generator (ND-CPG) is developed, which results in stable, accurate, and diverse flapping rhythms. The parameters of the flapping pattern generator are optimized with swimming speed as the target. Furthermore, by constructing a dynamic model, the swimming performance of the robotic fish is simulated and analyzed. Finally, experiments confirmed the performance advantages of the novel central pattern generator and the comprehensive swimming capabilities of the robotic tuna platform.

Thirdly, in order to address the pitch maneuver motion control problem, a multi-objective model predictive control and vertical plane pitch maneuver motion planning method are proposed based on configuration element analysis, achieving coupled control of pitch angle and depth. From the perspectives of buoyancy characteristics, hydrodynamics of the shape, and the position and size of pectoral fins, the system stability and controllability of the robotic fish under different swimming speeds and postures are explored. The configuration elements required for pitch maneuver motion control of high-speed robotic fish are identified. Based on the data-driven dynamic model, a multi-objective model predictive controller is constructed to achieve coupled control of pitch angle and depth values, demonstrating good control performance at various swimming speeds. Considering the pitch maneuverability constraints, a vertical plane motion trajectory generation method is proposed. Vertical plane motion trajectories can be generated based on key points of pitch angle-depth values. Experimental results showed that the robotic fish could successfully achieve pitch attitude and depth control, and perform specific vertical plane motions based on the generated trajectories.

Fourthly, for underwater search missions, a wide-angle fish-like visual perception system and a vision-based underwater search strategy are proposed. Inspired by the biological fisheye visual system, a set of fish-like binocular visual systems is proposed, which achieved a wide-angle visual perception of over 300° through field-of-view design and binocular image stitching methods. Based on the wide-angle perception ability, the underwater search strategy based on finite state machines is proposed, which combines water flow rate and perceptual information to achieve pre-search planning and in-search decision-making. A probability model of the search area is established, and search area planning is implemented. According to the search area, an adaptive search path planning method based on the probabilistic prior is proposed, generating highly efficient search paths. Furthermore, an underwater search scenario is constructed, and system integration and experimental verification are conducted. Simulation results validated the effectiveness of the underwater search method.

Keyword请输入关键词
Language中文
Document Type学位论文
Identifierhttp://ir.ia.ac.cn/handle/173211/57208
Collection毕业生_博士学位论文
Recommended Citation
GB/T 7714
仝茹. 面向水下搜索的仿生机器金枪鱼系统优化与运动控制[D],2024.
Files in This Item:
File Name/Size DocType Version Access License
博士学位论文-仝茹-提交版-202403(38243KB)学位论文 限制开放CC BY-NC-SA
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[仝茹]'s Articles
Baidu academic
Similar articles in Baidu academic
[仝茹]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[仝茹]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.