面向曲率及边界距离约束的弯曲管道内路径规划  

Path Planning for Curved Pipeline Interior with Curvature and Boundary Distance Constraints

作　　者：代煜[1] 赵煜[1] 尹晶晶 李睿[1] 张建勋[1] Dai Yu;Zhao Yu;Yin Jingjing;Li Rui;Zhang Jianxun(Institute of Robotics&Automatic Information System,Nankai University,Tianjin 300350,China)

机构地区：[1]南开大学机器人与信息自动化研究所,天津300350

出　　处：《天津大学学报（自然科学与工程技术版）》2024年第10期1010-1021,共12页Journal of Tianjin University：Science and Technology

基　　金：首都医学科学创新中心资助项目(CX23YZ12).

摘　　要：弯曲管道结构是工业探测或医用诊察中常见的一种场景,传统机器人的路径规划大多基于管道中心线的提取,而忽略了在狭窄空间中作业的真实参数约束,易使得作业工具末端受到弯折或挤压,影响作业效率,增加作业风险.针对以上问题,提出了一种面向真实曲率和边界距离参数约束的路径规划方法,在可视化的三维模型中生成管道的中心稠密点集,针对该点集提出两步走的空间特征点识别策略.在选定作业路径的初始点、终止点后,依据邻近点间的曲率均值、极值及差分关系识别曲率极值点、分段点,通过对粗扫描特征点的遍历进一步识别出拐点和宽域节点以及过密点.将识别出的特征点作为型值点,反算得到NURBS曲线及对应插值点的曲率值,通过模型切片搜索得到曲线插值点的边界距离不合格点集.将极值点添加至特征点集中,再通过搜索得到的曲率不合格点集,将极值点进行拟合圆调整至后更新特征点集、在完成自适应轮次的特征点迭代后,检测并标记为空或连续不变的全区间参数不合格点区间集.实验结果证明:弯曲圆柱仿真模型和CT重建肾腔模型的规划路径经由迭代分别满足曲率10 mm^(-1)、8 mm^(-1)的约束.同时最短距离的全区间检测及标记结果也表明本文方法可以在满足曲率要求的前提下尽可能不碰撞管道内壁,验证了本文方法的有效性和可行性.Curved pipeline structure is common in industrial detection and medical diagnosis.Traditional robot path planning is mostly based on the extraction of pipeline centerline while ignoring the real parameter constraints in a narrow space.This can easily bend or squeeze the end of the tool,thereby affecting the operational efficiency and in-creasing its operational risk.To solve the problems mentioned above,in this study,a path planning method based on real curvature and boundary distance parameter constraints was proposed.First,a dense center point set of the pipeline was generated in the visual three-dimensional model,and a two-step space feature point identification strategy was proposed for the point set.Additionally,curvature extreme and subsection points were identified according to the curvature mean value,extreme value,and difference relation between adjacent points.Then,inflection points,wide domain nodes,and over-dense points were further identified by traversing the coarse scanning feature points.The identified feature points were used as the model value points,the curvature values of the NURBS curve and the corresponding interpolation points were obtained via inverse calculation,and a boundary distance unqualified point set of the curve interpolation points was obtained through model slice search.Moreover,the extreme points were added to the feature point set.Then,the extreme points were adjusted to a fitting circle through the curvature unqualified point set obtained through searching,and the feature point set was updated.After the completion of the feature point iterations of the adaptive rounds,the interval set of unqualified points of the full interval parameters was detected and marked as empty or continuous invariant.The experimental results showed that the path planning of the curved cylinder simulation and the CT reconstruction venal lumen models meet the curvature constraint of 10 mm^(-1)and 8 mm^(-1)through iterations.Furthermore,the shortest distance detection and labeling results demonstrated

关 键 词：路径规划 弯曲管道 NURBS曲线插值 特征点识别 曲率约束 边界距离约束 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]