【作者】 陈勇平；

【导师】 周海波； 李志强；

【作者基本信息】 中南大学 ， 机械（专业学位）， 2025， 硕士

【摘要】 随着智能制造技术的快速发展,传统机器人因其安装与部署过程过于繁杂且柔性化程度低,在大表面检测中存在显著局限性。其作业效率不仅受限于人工操作周期,还存在检测路径规划困难,缺陷图像数据不一致的问题,难以满足高效快速的检测要求,已经逐步被由移动底盘与机械臂构成的复合机器人替代。这种结构形式打破了传统机械臂固定工位的局限性。本文针对飞机外壳检测场景的应用需求,搭建了一套复合机器人图像采集系统,主要内容如下:(1)搭建了面向大表面的复合机器人缺陷图像采集系统,并开发了控制软件。通过对复合机器人的工作场景进行分析,将相关需求量化为设计指标,完成了主要硬件的选型与设计。本文设计的机器人转弯半径为1151mm,其搭载的机械臂最大运动范围为1350mm。同时,为了降低复合机器人系统的使用难度,软件部分对自动手眼标定、检测路径自动生成、设备离线检测、数据管理等功能进行了模块化集成。(2)基于齐次坐标变换理论,采用D-H参数法建立了复合机器人的正运动学模型,并通过相机标定、手眼标定得到了相机最佳成像距离相对于移动底盘的位姿变换关系,实现了从移动底盘到相机坐标系中各个坐标系的对齐。(3)提出多尺度梯度方差图像清晰度评价函数,通过融合图像梯度的统计特征与图像的多尺度特征,增加图像清晰度评价函数对不同光照环境、不同缺陷图像内容的适应性。基于所提出的图像清晰度评价函数,实现了缺陷图像采集过程的闭环,提高了缺陷图像采集的一致性,同时将其应用在手眼标定过程中以提高手眼标定的效率与鲁棒性,实验表明,相较于人工标定,自动手眼标定将标定效率提升50%,标定的重投影误差分布空间从0.8×0.8×0.8mm降低到0.6×0.6×0.4mm,重投影平面法向量夹角误差均值从0.0524°降低到0.0391°。(4)提出了基于飞机外壳点云数据的全覆盖路径规划方法,采用参数映射的方法对飞机外壳点云数据进行降维,并将相机视野投射到降维之后的飞机外壳表面,实现对其表面图像采样的全覆盖,随后采用分层规划,局部优化的方法,对复合机器人进行运动规划。通过分析复合机器人机械臂的工作空间,对待遍历路径点进行分割,确定机器人的停靠位置,最后通过构建多目标优化模型,对复合机器人进行协同运动规划,完成机器人最优逆解求解,并通过实验验证了该方法的有效性。图98幅,表21个,参考文献80篇更多还原

【Abstract】 With the rapid development of intelligent manufacturing technology,traditional robots have significant limitations in large surface detection because of their complicated installation and deployment process and low flexibility.Its operation efficiency is not only limited by the manual operation cycle,but also has the problems of difficult detection path planning and inconsistent defect image data.It is difficult to meet the requirements of efficient and rapid detection,and has been gradually replaced by a composite robot composed of a mobile chassis and a manipulator.This structural form breaks the limitations of the traditional mechanical arm fixed station.In this thesis,a set of composite robot image acquisition system is built for the application requirements of aircraft shell detection scene.The main contents are as follows:(1)A composite robot defect image acquisition system for large surface is built,and the control software is developed.Through the analysis of the working scene of the composite robot,the relevant demand is transformed into the design index,and the selection and design of the main hardware are completed.The turning radius of the robot designed in this thesis is 1151 mm,and the maximum motion range of the manipulator is1350 mm.At the same time,in order to reduce the difficulty of using the composite robot system,the software part integrates the functions of automatic hand-eye calibration,automatic generation of detection path,offline detection of equipment,data management and so on.(2)Based on the homogeneous coordinate transformation theory,the forward kinematics model of the composite robot is established by using the D-H parameter method.The pose transformation relationship between the optimal imaging distance of the camera and the moving chassis is obtained by camera calibration and hand-eye calibration,and the alignment of each coordinate system from the moving chassis to the camera coordinate system is realized.(3)A multi-scale gradient variance image sharpness evaluation function is proposed.By fusing the statistical features of image gradient and the multi-scale features of the image,the adaptability of the image sharpness evaluation function to different illumination environments and different contents is increased.Based on the proposed image sharpness evaluation function,the closed-loop of the defect image acquisition process is realized,and the consistency of the defect image acquisition is improved.At the same time,it is applied to the hand-eye calibration process to improve the efficiency and robustness of the hand-eye calibration.Experiments show that compared with manual calibration,automatic hand-eye calibration improves the calibration efficiency by 50%.The calibrated reprojection error distribution space is reduced from 0.8×0.8×0.8mm to0.6×0.6×0.4mm,and the mean angle error of the reprojection plane normal vector is reduced from 0.0524°to 0.0391°.(4)A full coverage path planning method based on the point cloud data of the aircraft shell is proposed.The parameter mapping method is used to reduce the dimension of the point cloud data of the aircraft shell,and the camera field of view is projected onto the surface of the aircraft shell after dimension reduction to realize the full coverage of the surface image sampling.Then,the hierarchical planning and local optimization method are used to plan the motion of the composite robot.By analyzing the workspace of the composite robot manipulator,the traversal path points are segmented to determine the parking position of the robot.Finally,the multi-objective optimization model is constructed to plan the cooperative motion of the composite robot and solve the optimal inverse solution of the robot.The effectiveness of the method is verified by experiments.更多还原