【作者】 王晓丽；

【导师】 吴迪；

【作者基本信息】 大连理工大学 ， 计算机应用技术， 2015， 硕士

【摘要】 近年来,大型起重机吊装行业迅猛发展,超大型吊装日渐普遍,推动了用于辅助吊装方案设计人员快速有效制定作业方案的计算机辅助吊装仿真技术的广泛应用,而现存的仿真工具在吊装动作序列规划上的低效性问题也日益突出。本文重点研究了基于随机采样思想的高自由度动作搜索算法——RRTConnect,针对其在动作规划中因树生长缺乏方向性、所得路径迂回严重、平滑性较差而导致路径代价大、规划时间长、路径质量差的问题,提出了一种新型的算法用于起重机的动作序列规划。算法通过在采样策略、代价度量方式等方面的改进以及对既得路径的平滑处理,使随机采样具有了特定的方向性,从而能在给定的吊装环境中根据被吊物的初始和就位位置规划出一条安全、无碰撞的、满足起重性能约束的、相对平滑的高质量可行动作序列。并且,本文将研究成果嵌入课题组的吊装仿真平台,通过几个作业案例验证了所提方法的高效性。本文首先对目前吊装仿真软件及动作规划算法存在的问题进行研究,分析其优点与不足,在发扬优点、尽量克服不足的基础上,依据改善仿真效果、提高算法性能的思路设定研究内容,组织文章结构。然后,作为研究算法与完善仿真平台的基础,本文对移动式起重机的作业原理进行分析。定义了吊装系统的刚体模型、位姿表达向量、位姿变换矩阵、动作代价度量函数等,进而建立了求解起重机吊装动作序列规划问题的完整数学模型。并且,本文对吊装过程中存在的起重性能约束、非完整性运动学约束、碰撞约束、单封闭链约束等各项约束的处理思路进行了介绍,保障了仿真的真实性。进而,本文提出了一种新型的基于随机采样的树生长算法。基于椭球模型的采样策略的应用,使随机树的扩展具有了一定的方向性,缩减了初始点到达目标点的时间,大大提高了动作规划的效率。接着,我们提出了一种考虑动作安全度的序列平滑算法,通过定义一种带优先级系数的代价度量函数以及基于节点增强法的序列平滑算法的应用,尽量避免了危险动作的执行。同时,使所得序列尽量平滑,从而帮助我们规划出一条从起始状态到就位状态的安全、无碰撞的高质量动作序列。最后,通过对仿真平台下几个吊装案例的介绍,验证了所提算法的效率和性能。更多还原

【Abstract】 In recent years, along with the rapid development of heavy lifting, the Computer Aided Lift Motion Plan Design technology (CALMPD) which can assist designers to quickly plan a lifting project is widely used. However, almost all of the existing simulation tools are less efficient ones. In this paper, we focus on the study of RRTConnect, a motion plan algorithm in high-dimensional space based on random sampling. And to solve its disadvantages in motion planning such as high path cost, expensive planning time and low path quality which result from detour problem and poor path smoothness, a newly developed algorithm is put forward in this paper. With the help of a novel sampling strategy, a new way to calculate the cost and the smoothness of obtained path and so on, we can quickly find a smoother motion sequence with higher quality between the initial and goal configurations. And results from the simulation experiments validate the effectiveness and usability of the algorithm proposed.To begin with, we analyze the advantages and disadvantages of the current simulation softwares and motion planning algorithms and according to which we carry out our study and organize the paper.Then, we introduce the working principle of mobile crane, define the rigid body model, state vector, state transferring matrix and motion cost function of the lift system and establish an integral mathematical model to solve the motion plan problem. And also, several constraints that may appear in the lift process such as capability constraint, nonholonomic constraint, collision constraint and closed-chain constraint are considered in the paper.Next, we propose a new algorithm based on RRTConnect. Through the using of a new sampling strategy, the random tree can grow with a certain direction and as a result, the planning time is rapidly reduced.Furthermore, we introduce a smoothening algorithm considering the operations’safety. By defining a fresh cost metric function which takes safety of the crane’s operation into account and using a smoothening algorithm based on nodes’ enhancement, some dangerous operations are avoided and the performance of the algorithm proposed is obviously improved. Finally, results from the simulation experiments validate the effectiveness and usability of our algorithm.更多还原