机器人辅助柔性针穿刺系统研究
文献类型：学位论文
| 作者 | 霍本岩 |
| 学位类别 | 博士 |
| 答辩日期 | 2017-05-27 |
| 授予单位 | 中国科学院沈阳自动化研究所 |
| 授予地点 | 沈阳 |
| 导师 | 徐卫良 ; 韩建达 |
| 关键词 | 柔性针 机器人辅助系统 路径规划 粒子群优化方法 模型预测控制 |
| 其他题名 | Research on Robot-assisted Steering System for Bevel-tip Flexible Needle |
| 学位专业 | 机械电子工程 |
| 中文摘要 | 穿刺术是一种微创手术，其具有手术创伤小、病人恢复快的优势，在临床上有着广泛的应用。穿刺术大都使用刚性针完成穿刺。但是刚性针穿刺有以下缺点：无法避障、针尖可控性差、修正针尖偏差伤害大。为了克服这些缺点，研究人员提出了斜尖柔性针(Bevel-tip flexible needle)的概念。柔性针直径小，对组织造成的损伤小；穿刺轨迹可控，能够实现曲线穿刺轨迹，绕开神经、血管等敏感区域，到达直线轨迹无法到达的病灶点。柔性针为穿刺术的发展提供了新思路和方向，对穿刺术的发展和推广具有重要意义。但是，柔性针为连续柔性体，建模困难；与黏弹生物组织之间交互复杂，且针尖定位方式有限。这些问题对柔性针精确穿刺提出了挑战。为解决上述问题，本文在国家自然科学基金的资助下，研制了机器人辅助柔性针穿刺系统，并深入研究了柔性针模型、穿刺路径规划、穿刺控制方法三项关键方法，具体内容如下： 首先，介绍柔性针运动学模型的两种建模方式，并在独轮车模型的基础上提出了穿刺长度域模型。根据柔性针穿刺过程的时间无关性，将时间从柔性针模型中剥离，建立穿刺长度域内柔性针运动学模型，解耦了状态和输入，简化了模型方程。然后介绍柔性针模型参数辨识方法，指出辨识过程中存在的误差，并对辨识精度进行分析。最后提出两种柔性针姿态的估计方法，并分析其理论依据及适用范围。 其次，对柔性针穿刺路径规划方法进行研究。柔性针穿刺路径规划方法的研究在二维平面和三维空间内开展。针对二维平面内多层组织穿刺路径规划问题，使用动态规划搜索最优穿刺路径。为了提高算法效率，提出了基于针尖偏离度的启发函数，在保证规划效果的前提下有效地减少了激活状态个数。而对三维复杂环境中穿刺路径规划问题，则提出使用起始点向量和穿刺长度描述柔性针轨迹的方法，并基于空间几何知识建立控制序列与穿刺路径之间的关系。在此基础之上建立穿刺路径的优化函数，并使用粒子群算法搜索最优或者次优的规划路径。该方法能够约束路径蕴含的旋转次数，降低路径对组织的伤害。 再次，对不同情况下柔性针穿刺控制问题进行研究，提出了四种控制方法。针对无障碍环境下柔性针穿刺控制问题进行研究，定义了目标可达度的概念，据此提出了可达度控制方法，通过调整可达度阈值能够有效的减少穿刺需要的旋转控制，减少控制对组织造成的伤害。之后构建Two-Step控制方法，并对算法进行证明，分析算法的适应范围。然后，分析了基于路径重规划控制方法的优缺点，并通过仿真实验进行验证。针对预规划路径跟踪问题，提出了基于模型预测控制的控制方法，构建了MPC控制器的三要素：预测模型、优化函数和优化算法。通过仿真实验对控制器的路径跟踪能力和模型参数存在偏差时的跟踪精度进行分析，得出如下结论：当期望路径的模型参数小于控制器的模型参数并且控制器的模型参数小于系统的模型参数时，能够实现期望路径的精确跟踪。 最后，针对透明仿真组织和离体生物组织分别搭建了实验平台，构建柔性针穿刺控制系统，将本文所提出的方法融合为一个整体。以柔性针穿刺精度为评判标准，对柔性针模型参数辨识与针尖姿态估计、针尖定位与路径规划、多种穿刺控制方法进行验证和分析。实验结果证实在透明仿真组织内本文所提出的方法能够实现高精度的穿刺，并且对模型参数不确定性、噪声干扰等有鲁棒性。但是对离体生物组织，为了实现精确的穿刺需要进行更多的研究工作。 综上所述，本文对柔性针模型、定位及规划方法和穿刺控制方法进行了深入研究，为柔性针发展提供了理论基础。同时，在研究中遇到的问题也为柔性针下一步的工作指明了方向。 |
| 英文摘要 | In the view of less trauma and rapid recovery, paracentesis is widely used in clinical. However, there are some disadvantages of rigid puncture needle used in paracentesis such as unable to avoid obstacles, poor controllability and huge correction damage of needle tip error. Bevel-tip flexible needle is an improvement of rigid puncture needle to overcome the disadvantages. Flexible needle causes less trauma due to the thin needle shaft. The needle tip can be controlled to avoid sensitive area, such as nerve and blood vessel, and reach nidus which is unreachable for rigid needle. Flexible needle provides new thoughts and directions of the development of paracentesis and is significant to the development and generalization of paracentesis. However, there are several challenges to steer flexible needle precisely, for example, difficulty to model flexible needle which is flexible continuum, the complex interaction between flexible needle and viscoelastic biological tissue, restrictions of needle tip location methods. To overcome these challenges, this paper designs a robot assisted system for steering flexible needle and focuses on the key methods of the system including the flexible needle model, needle tip location method and path planning method, needle steering methods under the support of State Natural Sciences Foundation. The concrete contents are as following: Firstly, two modelling methods of flexible needle kinematics model are introduced. And a kinematics model in inserted length domain is proposed based on the unicycle model. Time factor is stripped from the flexible needle model due to the time independence. The model is simplified via decoupling states and inputs. Then, the model parameter identification method is presented. The identified error is indicated and analyzed. Two ways to get the needle tip attitude angles are proposed. We analyze their theoretical foundation and scope of application. Secondly, we focus on path planning methods. Path planning methods are researched in 2D plane and 3D space. To optimize the inserted path in 2D multi-layer tissue, the workspace and needle tip states are discretized and the dynamic programming method is employed to search the best path in states space. An active states screening method based on tip deviation degree is proposed to reduce the active states and improve algorithm efficiency. While in 3D space, the inserted path is represented by tangent vectors of the start points and inserted lengths. The relation between control sequence and inserted path is established based on space geometry knowledge. Then, an optimization function is set up and optimized by intelligent optimization algorithm method, partial swarm optimization. This method reduces the damage of a planned path via restraining the rotary control times. Thirdly, four control methods are proposed for needle steering in different situations. Dealing with the control problem in non-obstacle environment, we propose a control method based on the reachable degree which is established on the foundation of reachable degree of a target. The reachable degree threshold is set up and adjusted to reduce the rotary control times. Then, a Two-Step control method is set up and proved. The application scope of the control method is analyzed. A control method based on path re-planning is introduced and verified via simulations. Aiming at the path tracking problem, a path tracking method is proposed based on Model Predictive Control (MPC). The three essential elements of MPC are established, i.e. predictive model, optimization function and optimization algorithm. The path tracking ability and path tracking error due to the model bias are analyzed and verified via simulations. A conclusion comes up that to realize precisely path tracking control, the model parameter of desired path should be less than the controller's and the controller's should be less than the actual system's. Lastly, two experimental platforms are established respectively for phantom tissue and ex-vivo biological tissue. The robot assisted needle steering system is set up and proposed methods work together to steer flexible needle. Proposed methods, i.e. identification method of model parameter and estimation method of needle tip attitudes, needle tip location methods and inserted path planning methods, several needle steering methods, are verified and analyzed via experiments with the evaluation criterion which is the inserted precision. Experimental results in phantom tissue indicate that proposed methods can steer flexible needle to a target precisely and are robust to handle noise effect and model parameter bias. However, more works should be done to steer a flexible needle precisely in ex-vivo biological tissue. In conclusion, this paper researches on flexible needle model, location methods and path planning methods, steering methods. These researches lays the theoretical foundation for the development of flexible needle. At the same time, problems arising in this paper indicate the direction of future work about bevel-tip flexible needle. |
| 语种 | 中文 |
| 产权排序 | 1 |
| 源URL | [http://ir.sia.cn/handle/173321/20555]  |
| 专题 | 沈阳自动化研究所_机器人学研究室 |
| 推荐引用方式 GB/T 7714 | 霍本岩. 机器人辅助柔性针穿刺系统研究[D]. 沈阳. 中国科学院沈阳自动化研究所. 2017. |
入库方式： OAI收割
来源：沈阳自动化研究所
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。