CASIA OpenIR  > 复杂系统认知与决策实验室  > 先进机器人

一、针对水下自主作业任务,提出了一种仿生蹼推进水下作业机器人系统设计方案,并研制样机平台。首先,设计了仿生蹼推进器,并将其引入到水下作业机器人中。其次,根据作业任务的需求,搭载了轻量化四自由度机械臂,并安装了水下传感系统。此外,构建了水下作业机器人的底层控制系统和基于ROS(Robot Operating System)的软件系统。最后,通过系统软硬件集成及实验,验证了仿生蹼推进水下作业机器人样机的有效性及可靠性。
二、针对水下作业机器人在未知环境下的地形导航问题展开研究,提出了一种水下地形预测与跟随控制的方法。首先,针对未知的水下地形预测问题,给出了一种基于长短时记忆网络和非线性预测模型的地形预测方法。其次,根据所得到的地形预测信息,给出了一种基于非线性模型预测控制的地形跟随控制优化方法。然后,利用基于连续法的广义最小残量算法来提高优化方法的计算效率。最后,开展了基于 ROS 的水下地形预测与跟随控制的仿真实验和海洋环境下的水下地形预测与跟随实验,实验结果验证了所提方法的有效性。
四、针对水下物体搜索和扰流环境下的物体自主抓取任务,提出了一种局部水域内的水下作业机器人目标搜索策略和扰流环境下的自主抓取控制方法。首先,给出了一种结合避障声呐检测障碍物和水下地形跟随的目标物体搜索策略。然后,给出了一种基于径向基函数神经网络的扰流观测器,用于实时估计扰流值。其次,给出了一种基于长短时记忆网络的模型预测网络,它包括扰流预测网络和机器人状态预测网络,用于扰流下的机器人状态序列预测。将扰流下机器人状态序列作为优化算法的预测值,通过滚动优化,获得最优控制输入。最后,在基于 ROS 的水下仿真环境开展了水下目标物体搜索和自主抓取仿真实验,仿真结果验证了所提方法的有效性。


Other Abstract

Underwater vehicle-manipulator systems (UVMSs) are valuable tools for the exploration and development of the ocean environment, and has been widely used for applications such as the salvage of submerged objects, underwater archaeology, rescue and salvage operations, and fishing. The present thesis focuses on the system design of a fippers-propelled UVMS (F-UVMS), underwater terrain prediction and following control, coordinated control of underwater biomimetic vehicle and manipulator, and autonomous control for grasping objects underwater current disturbance. The main contents of the present thesis are as follows:
     First, to perform underwater autonomous manipulation tasks, the design method of the F-UVMS and the development of a first prototype are presented. A fippers propulsor was designed and integrated to the UVMS. According to the requirements of underwater manipulation tasks, a four-degrees of freedom (DOF) underwater manipulator and an underwater sensor system are mounted on the underwater vehicle. Furthermore, an onboard control system and software system based on Robot Operating System (ROS) are also developed. The reliability and validity of the F-UVMS is analyzed by system integration and experiments.
    Second, an underwater terrain predicting and following control method is proposed to perform underwater terrain navigation with UVMS in unknown environments. As a first step, an underwater terrain predicting method based on long short-term memory network (LSTM) and nonlinear prediction model is put forward. Then, by virtue of the predicted underwater terrain information, an optimal method based on nonlinear model predictive control (NMPC) for underwater terrain following is presented. In addition, the continuation/general minimum residual (C/GMRES) algorithm is used to improve the computational efficiency of the NMPC. ROS-based simulation experiments and sea experiments for underwater terrain predicting and following control were conducted, and results are used to verify the effectiveness of the proposed underwater terrain predicting and following control method.
    Third, a vehicle-manipulator coordinated control method based on non-singular terminal sliding mode control (NTSMC) is proposed to address the UVMS dynamic coupling problem. An improved NTSMC (I-NTSMC) was first formulated as the main controller. A sliding mode reaching motion control law, an adaptive tracking differentiator, and a state observer are integrated into the I-NTSMC to relieve the chattering
phenomenon and speed up convergence. Then, the Newton-Enler model is formulated to estimate the disturbance imposed on an underwater biomimetic vehicle generated by instantaneous manipulator motion. The estimated disturbance is compensated for the main controller. The effectiveness of the proposed method was verified through experiments on an underwater autonomous opening door and grasping objects using UVMS under free-floating status.
   Fourth, focusing on searching for underwater objects and grasping them in a water current disturbance environment, an objects searching strategy and autonomous grasping control method with water current disturbance compensation are studied. A search strategy for underwater objects is proposed, combining underwater terrain navigation and obstacle detection information using collision obstacle sonar. Then, an autonomous grasping control framework is presented, which consists of a radial basis function neural network based disturbance observer (RBF-DOB), LSTM-based predictive model network, NMPC, and Newton-Enler model. The RBF-DOB is formulated to estimate the water current disturbance. The LSTM-based predictive model network is composed of a state prediction network and a water current disturbance prediction network, which can be used to predict UVMS state sequences underwater current disturbance. The UVMS state sequences were used to generate optimal control with the help of NMPC. Meanwhile, the manipulator motion disturbance is considered as feedforward compensation. ROS-based underwater simulations for objects searching and autonomous grasping were conducted to demonstrate the effectiveness of the proposed method.

Keyword仿生推进 水下作业机器人 水下地形导航 非奇异终端滑模控制 水下自主抓取控制 非线性模型预测控制
Subject Area控制理论
MOST Discipline Catalogue工学::控制科学与工程
Sub direction classification智能控制
Document Type学位论文
Recommended Citation
GB/T 7714
蔡明学. 仿生蹼推进水下作业机器人自主作业控制研究[D]. 在线. 中国科学院大学,2020.
Files in This Item:
File Name/Size DocType Version Access License
博士论文-蔡明学.pdf(32012KB)学位论文 开放获取CC BY-NC-SA
Related Services
Recommend this item
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[蔡明学]'s Articles
Baidu academic
Similar articles in Baidu academic
[蔡明学]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[蔡明学]'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.