毕业生知识库

目前，大部分喷涂生产线采用机器人手动编程的方法预先设置机器人喷涂路径，该方法手动编程时间长，喷涂的工件类型固定，不适用于批量喷涂定制化产品。在个性化喷涂领域中，主要采用“人工+模板”的方法进行喷涂，此方法需要根据图案设计不同模板，操作更为繁琐，并且工人会长时间处于漆雾环境中，不利于工人健康。为了解决这些问题，本文以国产防爆喷涂机器人为基础，提出了一种基于点云模型与矩形喷嘴喷枪的数码迷彩自动喷涂系统，该系统通过对复杂待喷物体表面进行多视角点云扫描与数据配准，建立待喷对象模型；结合喷枪工作特点和机器人工作约束进行路径规划，生成面向多色数码迷彩喷涂的最优喷涂路径，据此控制喷涂机器人进行自动喷涂。本文围绕所提出的数码迷彩自动喷涂系统开展模型构建与路径规划技术研究，主要工作成果与贡献包括以下四个方面：

（1）根据数码迷彩喷枪工作特点提出一种基于矩形喷嘴喷枪的数学模型，通过推导给出了面向二维平面和三维曲面喷涂时的喷枪涂层厚度累积速度分布函数。基于所构建的喷枪模型，分别在同一曲面内以及不同曲面衔接处，以保持涂层厚度精度和均匀性为目标，提出了相邻路径间隔、喷枪喷涂速度、喷枪切换角度等路径参数的优化计算方法。实验结果与实际应用效果表明：所建立的喷枪数学模型对数码迷彩喷涂具有更好的适用性，依据所提出的参数优化方法进行喷涂参数优化可以有效提高喷涂效率，有较好的适用性。

（2）针对具有较小重叠区域的多视角点云配准问题，提出一种基于结构特征描述子的点云配准方法，可以对不同视角采集的点云对进行配准，且对点云的初始姿态无限制要求。该方法将点云分割为若干子结构，并利用最小包围盒获取每个子结构局部形状特征，进而提出一种基于最小包围盒的形状特征描述子，在利用该描述子对点云的结构特征进行表征的同时，实现了基于该描述子的小重叠区域点云配准。此外，针对现有小重叠区域点云数据集较少的问题，构建了以实际室内外场景和大型车辆为对象的小重叠区域点云数据集，并借助于该数据集对所提算法开展了性能评估实验，以确保提出的配准算法能够得到有效评估。实验结果和实际应用效果表明，所提出的配准方法可以获得较高的配准成功率，通过配准所构建的特征车辆点云模型能够满足实际喷涂的精度要求。

（3）针对数码迷彩喷涂中待喷车辆CAD模型较少以及现有路径规划方法在规划复杂曲面时效率低下的问题，提出了一种基于点云分割和路径序列优化的最优喷涂路径规划方法。该方法利用基于最小曲率点的区域生长算法将点云模型分割成不同的子曲面，并在每个子曲面上利用提出的平面与点云相交求解的方法生成最优喷涂路径。为了规划出各子曲面间的喷涂路径，提出了一种基于进化粒子的序列优化算法，进而得到面向车体整体的最优喷涂路径。为了进一步生成不同颜色的数码迷彩喷涂路径，借助于提出的基于包围盒的模型重建算法把点云转为若干10cm×10cm的网格面片，把具有相同颜色面片的中心点作为新点云，进而通过对不同颜色点云进行独立喷涂路径规划获得面向多色数码迷彩喷涂的喷涂路径。实验结果和实际应用效果表明，所提喷涂路径规划方法可有效提高喷涂效率，满足实际数码喷涂要求。

（4）为了解决目前人工喷涂数码迷彩中存在的喷涂效率低、施工难度大等问题，以上述工作为基础构建了数码迷彩自动喷涂系统。该系统采用双臂双轨结构，通过控制机械臂夹持激光扫描仪对待喷车辆进行多视角扫描；借助于本文所提出的基于结构特征描述子的的点云拼接算法构建其待喷表面模型；借助于本文所提出的基于点云分割和路径序列优化的路径规划方法获取多色喷涂路径，并在综合考虑机器人约束和工艺要求的情况下，通过机器人自编程实现了给定多色数码迷彩喷涂的自动喷涂控制；编制完成了具备碰撞检测功能和动态效果实时展示功能的喷涂路径仿真与修正软件，基于该软件可实现喷涂过程中的人机交互验证，可保证喷涂路径的安全性和可靠性。面向多种类型车辆的实际喷涂效果证明，所开发的数码迷彩自动喷涂系统满足数码迷彩的工艺要求，且能够在很大程度上提高喷涂效率。

Currently, most spraying production lines use a robot manual programming method to preset the robot spraying path. This method takes a long time to manually program, and the type of workpiece being sprayed is fixed, which is not suitable for batch spraying customized products. In the field of personalized spraying, the "manual+template" method is mainly used for spraying. This method requires different templates to be designed according to the pattern, making the operation more cumbersome. Moreover, workers are exposed to paint mist for a long time, which is not conducive to their health. In order to solve these problems, this paper proposes a digital camouflage automatic spraying system based on a domestic explosion-proof spraying robot and a rectangular nozzle spray gun. The system builds a model of the object to be sprayed by performing multi view point cloud scanning and data registration on the surface of the complex object to be sprayed; Combining the working characteristics of the spray gun and the working constraints of the robot, path planning is performed to generate the optimal spray path for multicolor digital camouflage spraying, and based on this, the spray robot is controlled to perform automatic spraying. This article focuses on the research of model construction and path planning technology for the proposed digital camouflage automatic spraying system. The main work achievements and contributions include the following four aspects:

(1) According to the working characteristics of digital camouflage spray guns, a mathematical model based on rectangular nozzle spray guns is proposed. The cumulative velocity distribution function of spray gun coating thickness for spraying on two-dimensional and three-dimensional surfaces is derived. Based on the constructed spray gun model, an optimization calculation method for path parameters such as adjacent path spacing, spray gun spraying speed, and spray gun switching angle was proposed to maintain the accuracy and uniformity of coating thickness within the same surface and at the joints of different surfaces, respectively. The experimental results and practical application results show that the established spray gun mathematical model has better applicability for digital camouflage spraying. Spraying parameter optimization based on the proposed parameter optimization method can effectively improve spraying efficiency and have good applicability.

(2) Aiming at the problem of multi view point cloud registration with small overlapping regions, a point cloud registration method based on structural feature descriptors is proposed, which can register pairs of point clouds collected from different perspectives without limiting the initial pose of the point cloud. This method divides a point cloud into several sub structures, and uses the minimum bounding box to obtain the local shape features of each sub structure. Then, a shape feature descriptor based on the minimum bounding box is proposed. While using the descriptor to characterize the structural features of the point cloud, point cloud registration in small overlapping regions is implemented based on the descriptor. In addition, in view of the problem that there are few existing point cloud datasets in small overlapping areas, a small overlapping area point cloud dataset is constructed based on actual indoor and outdoor scenes and large vehicles, and performance evaluation experiments are conducted on the proposed algorithm using this dataset to ensure that the proposed registration algorithm can be effectively evaluated. Experimental results and practical application results show that the proposed registration method can achieve a high registration success rate, and the feature vehicle point cloud model constructed by registration can meet the accuracy requirements of actual spraying.

(3) Aiming at the problems of fewer CAD models of vehicles to be sprayed in digital camouflage spraying and the inefficiency of existing path planning methods in planning complex surfaces, an optimal spraying path planning method based on point cloud segmentation and path sequence optimization was proposed. This method uses a region growth algorithm based on minimum curvature points to divide the point cloud model into different sub surfaces, and generates the optimal spray path on each sub surface using the proposed method of intersection between the plane and the point cloud. In order to plan the spray path between each sub surface, an evolutionary particle based sequential optimization algorithm was proposed to obtain the optimal spray path for the entire vehicle body. In order to further generate digital camouflage spray paths with different colors, a model reconstruction algorithm based on bounding boxes is proposed to transform a point cloud into several 10cm × 10cm  mesh patches, using the center point of the same color patch as a new point cloud, and then obtaining a spray path for multicolor digital camouflage spray through independent spray path planning for different color point clouds. Experimental results and practical application results show that the proposed spray path planning method can effectively improve spray efficiency and meet practical digital spray requirements.

(4) In order to solve the problems of low spraying efficiency and difficult construction in the current manual spraying of digital camouflage, an automatic digital camouflage spraying system was constructed based on the above work. The system adopts a dual arm and dual rail structure, and controls the mechanical arm to hold the laser scanner to scan the vehicle from multiple perspectives; The surface model to be sprayed is constructed using the point cloud stitching algorithm based on structural feature descriptors proposed in this paper; Using the path planning method based on point cloud segmentation and path sequence optimization proposed in this paper, the multi color spray path is obtained. Under the comprehensive consideration of robot constraints and process requirements, automatic spray control for a given multi color digital camouflage spray is realized through robot self programming; A spray path simulation and correction software with collision detection and real-time display of dynamic effects has been developed. Based on this software, man-machine interaction verification during the spray process can be achieved, ensuring the safety and reliability of the spray path. The actual spraying results for various types of vehicles have proven that the developed digital camouflage automatic spraying system meets the technological requirements of digital camouflage, and can greatly improve the spraying efficiency.