基于采样的采摘机器人机械臂避障算法研究

发布时间：2018-05-25 09:34

本文选题：采摘机器人 + 机械臂　；参考：《中南林业科技大学》2017年硕士论文

【摘要】：采摘机器人的应用有利于缓解农业劳动力日趋不足的问题,能够大大提高农业产品的市场竞争力。但是采摘过程中障碍物阻碍采摘机器人机械臂的问题严重制约了采摘机器人在农业生产中的应用和发展。因此,开展采摘机器人机械臂避障路径规划方法的研究,有着十分重要的现实意义和广阔的应用前景。本文主要从三个方面展开研究:快速构形空间构建算法;机械臂碰撞模型建立方法;高维空间路径规划算法。快速构形空间构建算法。分析现有避障算法所存在的不足,结合油茶果采摘机器人机械臂的结构特点,选取构形空间作为本文的避障路径规划空间。对传统的机械臂构形空间构建方法所存在的不足进行分析,在此基础上提出一种由离线阶段和在线阶段构成的快速构形空间构建方法。算法在离线阶段通过遍历机械臂构形完成外部碰撞数据库和自碰撞数据库的建立。通过在线阶段计算出实时工作场景中的障碍物所占据的工作空间单元,并根据单元序号索引离线阶段所建立的两大数据库,快速构建机械臂在实时场景中的构形空间模型。机械臂碰撞模型的建立方法。通过对常用碰撞检测技术和空间包围盒优缺点的分析,确定以方向包围盒(Oriented Bounding Box,OBB)作为机械臂的碰撞检测拟合模型。利用固连在方向包围盒上的局部坐标系对方向包围盒各顶点坐标进行描述,并通过几何分析法和空间矢量法详细描述局部坐标系的建立过程,最后利用局部坐标系在基础坐标系中的齐次变换矩阵将方向包围盒在局部坐标系中的描述转换到基础坐标系中。高维空间路径规划算法。对路径规划中基于图形,基于启发,基于采样以及智能路径规划算法的优劣势进行详细的分析,在此基础上选择基于采样的路径规划算法进行本文的避障路径搜索。对双向快速扩展随机树算法(RRT-Connect)以及渐进快速扩展随机树算法(Asymptotically Optimal Rapidly-exploring Random Tree,RRT*)进行融合改进,得到一种双向渐进最优快速扩展随机树算法。通过与原始的RRT-Connect和RRT*算法的性能比较验证了融合算法的优势。本文通过虚拟仿真系统和采摘机器人机械臂实体样机进行方法的验证。结果显示,本文所提出的方法能够快速、有效地进行避障路径规划。
[Abstract]:The application of picking robot can alleviate the shortage of agricultural labor and improve the market competitiveness of agricultural products. However, obstacles in the picking process hinder the robotic arm of the picking robot, which seriously restricts the application and development of the picking robot in agricultural production. Therefore, the research on obstacle avoidance path planning of picking robot has very important practical significance and broad application prospect. This paper focuses on three aspects: fast configuration space construction algorithm, robot arm collision model building method and high-dimensional spatial path planning algorithm. Fast configuration space construction algorithm. Based on the analysis of the shortcomings of the existing obstacle avoidance algorithms and the structural characteristics of the robot arm of Camellia oleifera picking robot, the configuration space is selected as the obstacle avoidance path planning space in this paper. Based on the analysis of the shortcomings of the traditional configuration space construction method, a fast configuration space construction method is proposed, which is composed of off-line and on-line phases. The external collision database and the self-collision database are established by traversing the configuration of the manipulator in the off-line phase. The workspace units occupied by obstacles in real time working scenes are calculated by on-line stage. According to the two databases established in off-line stage of unit ordinal index, the configuration space model of manipulator in real time scene is constructed quickly. The method of establishing collision model of manipulator. Based on the analysis of common collision detection techniques and advantages and disadvantages of space bounding box, the directional bounding box (OBB) is chosen as the collision detection fitting model of robot arm. The local coordinate system attached to the directional bounding box is used to describe the coordinates of each vertex of the bounding box, and the establishment process of the local coordinate system is described in detail by means of geometric analysis and space vector method. Finally, the description of the direction bounding box in the local coordinate system is transformed into the basic coordinate system by using the homogeneous transformation matrix of the local coordinate system in the basic coordinate system. High dimensional spatial path planning algorithm. The merits and demerits of path planning algorithms based on graph, heuristic, sampling and intelligent path planning are analyzed in detail. On this basis, the path planning algorithm based on sampling is selected to search the obstacle avoidance path in this paper. The bidirectional fast extended random tree algorithm (RRT-Connect) and the asymptotic fast extended random tree algorithm (Asymptotically Optimal Rapidly-exploring Random tree RRTT) are fused to obtain a bidirectional asymptotically optimal fast extended random tree algorithm. Compared with the original RRT-Connect and RRT * algorithms, the advantages of the fusion algorithm are verified. The method is verified by virtual simulation system and the robot arm prototype. The results show that the proposed method can plan obstacle avoidance path quickly and effectively.
【学位授予单位】：中南林业科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP242

【相似文献】