| 题名 | 可变形机器人路径规划与控制方法研究 |
| 作者 | 刘同林 |
| 学位类别 | 博士 |
| 答辩日期 | 2010-06-08 |
| 授予单位 | 中国科学院沈阳自动化研究所 |
| 授予地点 | 中国科学院沈阳自动化研究所 |
| 导师 | 吴成东 ; 李斌 |
| 关键词 | 可变形机器人 可重构 协同 机动性 路径规划 控制 |
| 其他题名 | Path Planning and Control Method Research for a Shape-shifting Robot |
| 学位专业 | 模式识别与智能系统 |
| 中文摘要 | 本文的研究内容是围绕国家“863”重点项目子课题“废墟洞穴搜救机器人研制”和机器人学国家重点实验室开放课题“可变形机器人自适应路径规划方法研究”展开的。以灾难救援为背景，以中国科学院沈阳自动化研究所研制的可变形机器人为平台，主要对以下两个方面进行深入研究。一方面，针对可变形机器人工作环境非结构性和高危险性等特点，研究机器人的通过能力和机动性，以增强其环境适应性和作业能力。另一方面，为实现可变形机器人在复杂环境中能够预测躲避各种威胁，快速准确地完成救援任务，将路径规划方法与机器人具有的变形能力相结合，重点研究可变形机器人特有的路径规划问题，以找到适合可变形机器人的路径规划方法。目前，这方面的研究成果在国内外比较罕见。因此，本论文的选题具有重要的理论意义和应用价值。可变形机器人在结构上具有模块化的特点，为了发挥模块化的优势，设计了分布式模块化体系结构。该系统实现了机器人模块和系统执行单元即插即用，提高了系统的鲁棒性。针对可变形机器人在不平地面变形困难，甚至无法变形的难题，由蛇形机器人的运动特性得以灵感，提出协同变形方法。建立相应数学模型，研究机器人系统各部分在协同变形过程中的运动和力学特性，实现将一部分变形阻力转化为变形动力。针对建立的数学模型非线性化问题，研究了基于摄动分析理论对模型线性化的方法，以降低控制的难度。为评价机器人的变形能力，给出协同变形性能评价指标。在此基础上提出基于模糊控制的协同变形方法，增强了机器人在非结构环境中的变形能力。实验表明协同变形方法能有效发挥机器人的整体性能，提高了机器人完成室外任务的可能性。在复杂环境中机器人的转向性能决定了机器人的灵活性，是确保自身安全的重要保障。研究可变形机器人不同构形的转向能力，对实际应用具有重要意义。为此，提出协同转向方法，对机器人所有构形转向进行建模。分析影响多模块机器人转向的因素，研究机器人多种构形的最佳转向方式，并给出协同转向性能评价指标。结合机器人常用运动构形的结构特点，提出抬头转向方法，实现非对称构形在不平地面小半径转向。通过理论和实验比较了不同构形下的协同转向方式，实验验证了协同转向方法的有效性。为了增强机器人在非结构环境中的越障能力，提出可变形机器人的协同越障方法。建立数学模型，对机器人越障高度与其重心位置的关系进行理论分析。从理论上比较常规越障方法与协同越障方法所能翻越障碍的最大高度。同时提出自主越障过程的控制策略，建立相应的情感模型，利用情感状态的变化对机器人控制策略进行微调。实验验证了协同越障方法及自主越障控制策略的有效性。在掌握可变形机器人性能的基础上，将机器人特有的灵活性和适应性应用到路径规划中，研究复杂环境下可变形机器人的路径规划与应用问题，对废墟探测营救等具有重要的理论和现实意义。针对可变形机器人特有的变形能力，提出面向可变形机器人路径规划的势场法，研究机器人变形对路径规划的影响。对势场法的计算原理进行改进，解决了目标不可达问题和局部极小情况。为了实现机器人自动变形通过狭窄空间，提出狭窄空间检测和机器人变形通过方法。实验表明机器人能够根据环境变化改变自身构形，增强了环境适应性，同时有效地缩短了路径长度。在对可变形机器人局部路径规划方法研究的基础上，为了增强机器人在灾难救援中所发挥的作用，根据可变形机器人的机动性和适应性，提出面向可变形机器人的自适应路径规划方法。对自适应路径规划方法搜索路径的机理进行了研究，提出基于环境的机器人变形通过方法，并给出变形评价，以充分发挥可变形机器人特有的通过性能。采用死锁绕行方法有效地解决了死锁情况，并提出变形记忆方法，以减少死锁绕行，同时帮助机器人寻找新的路径。研究电荷奖惩机制，以提高算法的计算速度。实验结果表明该算法能根据障碍物的分布情况自动调整机器人移动性和安全性之间的关系，并能根据环境和变形评价机制决定自身构形，同时记下变形的相关信息，为后续机器人路径寻优提供重要参考。因此，该算法能充分发挥机器人的变形能力和最优的机动性能，有效地完成自适应路径规划。而且通过与SLAM地图的结合可为操作人员提供重要的参考路径。可变形机器人完成未知环境下的路径规划后，通过已有的地图创建算法可以把环境变为已知。研究可变形机器人在已知环境下的全局路径规划可以弥补局部路径规划无法找到全局最优路径等不足。为此，提出面向可变形机器人路径规划的粒子群优化算法（Particle Swarm Optimization, PSO），对机器人的全局路径规划进行分析。研究粒子搜索区域划分对搜索最优路径的影响，提出基于情感粒子的PSO算法。对基于样条函数的轨迹优化方法进行了研究，并建立了PSO算法的适应函数和机器人变形函数。实验结果表明基于粒子搜索区域优化的PSO算法能够根据机器人的变形能力快速找到全局最优路径，而基于情感粒子的PSO算法更适合在较复杂的环境中完成路径规划任务。 |
| 索取号 | TP242/L74/2010 |
| 英文摘要 | The research of this dissertation is supported by a sub-project of The China National 863 Keystone Project—Research on a Relic Cave Rescue Robot and Open Subject of State Key Laboratory of Robotics—Adaptive Path Planning Method Research for a Shape-shifting Robot. Using disaster relief as the background, a platform which is the shape-shifting robot is developed by Shenyang Institute of Automation, Chinese Academy of Sciences. The major work of this dissertation includes the following two aspects: On one hand, regarding the working environment is non-structural and risky for the shape-shifting robot, the accessibility and mobility of the robot is researched to enhance its environmental adaptability and operation ability. On the another hand, for realizing that in complex environment where the robot can anticipate and avoid various threats to complete the rescue task quickly and accurately, the shape-shifting robot’s specific path planning is mainly studied to find a suitable method of the shape-shifting robot path planning by introducing the reconfigurable ability of robot into the path planning method. Now, the research in this aspect is rare at home and abroad. Thus these issues have great theoretical significance and application value.The shape-shifting robot has the modular characteristic in structure, so the distributed modular architecture is designed to play the modular advantages. This system implements the plug and play function of robot module and system execution units, therefore the system’s robustness is increased.Referring to the problem that reconfiguration of the shape-shifting robot is difficult or impossible to achieve on uneven ground, a cooperative reconfiguration method is proposed according to inspiration of the motion characteristics of snake robot. The mathematical model is established to study the kinematical and mechanical properties of each section of the robot, and a part of resistance is transformed into drive force of reconfiguration. Because of the nonlinear characteristics of the mathematical model, the linearization of non-linear model based on a perturbation analysis theory is studies to reduce the difficulty of control. An evaluation criterion of cooperative reconfiguration is proposed for evaluating the reconfiguration capacity of the robot. On this basis, a cooperative reconfiguration method based on fuzzy control is proposed to enhance the reconfiguration ability of the robot in unstructured environment. Experiments show that the cooperative reconfiguration method can effectively enhance the robot’s overall performance, and the possibility that the robot completes outdoor tasks. In complex environment, the turning performance of the robot determines the flexibility of the robot and is an important safeguard to ensure its own security. This dissertation studies the turning ability of the different configurations of the shape-shifting robot, which has important significance for practical application. Therefore, a cooperative turning method is proposed and a mathematical model of robot’s entire configuration turning is established. The impact factors of multi-module robot turning are analyzed and the best turning methods of different configurations of the robot is studied. An evaluation criterion of cooperative turning is proposed. A lift-turning method is proposed by combining with the structural characteristics of robot’s usual motion configuration. Small radius turning of unsymmetrical configuration is realized. Cooperative turning methods under various configurations are compared by theoretical analysis and experiments. Experimental results prove the validity of the cooperative turning method.A cooperative negotiation method of the shape-shifting robot is proposed to reinforce the capability of obstacle negotiation of robot in unstructured environment. A mathematical model is established. The relationship between the height that the robot can overcome and angle with gravity offset’s variation is analyzed theoretically. The maximum heights that conventional negotiation method and cooperative negotiation method can overcome are compared. Control strategy of autonomous negotiation is presented and the emotion model is established. The robot’s control strategy is fine-tuned according to the change of the emotion. Experimental results prove the validity of the cooperative negotiation method and autonomous control strategy.On the basis of the unique flexibility and adaptability of the robot, the path planning in complex environments is studied. This work has important theoretical and applicable significance in the area of the ruins detection and rescue.According to the unique reconfiguration of the shape-shifting robot, a potential field which is a classical method in path planning for shape-shifting robot is introduced and modified to adapt to the new applied area. The calculation theory of the potential field is improved to solve Goal-Unreachable with Nearby Obstacles (GUWNO) and the local minimum problem effectively. Detection of the narrow space and robot reconfiguration accessibility method is proposed to realize that the robot automatically changes configuration to pass through the narrow space. Experiments show that the robot can change its own configuration to complete path planning corresponding to the environmental variation. As a result, the environmental adaptability has been enhanced and the time of path planning has been reduced effectively.On the basis of previous research of the local path planning method of the shape-shifting robot, an adaptive path planning method for the shape-shifting robot is proposed to enhance the robot’s role in disaster rescue corresponding to the mobility and the adaptability of the shape-shifting robot. A robot reconfiguration accessibility method based on environmental evaluation is proposed and a reconfiguration evaluation is provided so the unique accessibility of the shape-shifting robot is fully used. The problem of deadlock is avoided by using the boundary following method. Further, a reconfiguration memory method is provided to optimize the trace of the boundary following, and it can help the robot to search new path. The reward and punishment mechanism of charge is studied to improve the computational effectiveness of the algorithm. Experiment results show that this method can automatically adjust the relation between the rapid movement and the secure mobile position of the robot based on the distribution of obstacles around the robot, and this method can decide the configuration according to environment and reconfiguration evaluation. At the same time, it can record reconfiguration information to provide the important reference for the follow-up robot. Therefore, the robot can make good use of the reconfiguration ability and flexible performance to effectively complete the adaptive path planning by using this method. And a combination with the SLAM map will provide important reference path for the operators. The environment will be known by existing map building algorithm, after the shape-shifting robot completes the path planning in unknown environment. In known environment the global path planning of the shape-shifting robot will be used to make up the shortage of the local path planning. Therefore, a particle swarm optimization algorithm for the shape-shifting robot is proposed to analyze the global path planning of the robot. The affection of the partition of particle search region on the search of optimal path is studied, and PSO algorithm based on emotional particle is proposed. The trajectory optimization method based on string of cubic splines is researched, and the fitness function of PSO algorithm and reconfiguration function of the robot is established. Experiment results show that PSO algorithm based on optimization of particle search region can quickly find the optimal path by using the robot configuration, while the PSO algorithm based on emotional particle is more suitable for the completion of path planning tasks in the complex environment. |
| 语种 | 中文 |
| 产权排序 | 1 |
| 公开日期 | 2012-07-27 |
| 分类号 | TP242 |
| 内容类型 | 学位论文 |
| 源URL | [http://ir.sia.ac.cn/handle/173321/9391]  |
| 专题 | 沈阳自动化研究所_机器人学研究室 |
| 推荐引用方式 GB/T 7714 | 刘同林. 可变形机器人路径规划与控制方法研究[D]. 中国科学院沈阳自动化研究所. 中国科学院沈阳自动化研究所. 2010. |
| 个性服务 |
| 查看访问统计 |
| 相关权益政策 |
| 暂无数据 |
| 收藏/分享 |
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。
修改评论