【作者】 谢圆；

【导师】 周建亮； 王永；

【作者基本信息】 中国科学技术大学 ， 导航、制导与控制， 2011， 博士

【摘要】 按照我国月球探测工程“三步走”的阶段实施方案,预计于2013年实现探测器在月面的软着陆,并开展月面定位探测和巡视勘察探测。月球车是实现月面巡视勘察的探测器,也将是我国首个在地外天体表面开展工作的移动机器人。一方面由于月球车智能水平有限,另一方面其良好的对地通信条件,使得月球车的控制方式以地面遥操作为主。其中,规划月球车在月面的工作序列,保证其在满足月面环境约束及自身能力约束的前提下,安全有效完成各种工程试验和科学探测目标,是地面遥操作系统的首要任务。因此,任务规划技术是月球车遥操作中的一项关键技术。然而,由于应用对象的特殊性,现有的其他应用领域内的任务规划技术及其他国家或航天机构巡视器的相关研究都无法直接应用在我国月球车的控制上。因此,开展月球车遥操作中的任务规划技术研究具有重大的工程意义和较高的理论价值。本文正是从我国月球车遥操作的实际需要出发,研究了月球车的任务规划技术。基于国内外在环境建模方面的研究进展,建立了面向月球车任务规划的综合环境模型；在该模型上研究了不同于局部导航中传统路径的任务层路径及其规划问题。最终,在该路径规划方法及月球车的行为集的分析基础上,研究了既能充分反映月球车行驶路径和其他行为之间耦合关系,又考虑月球车月面工作不确定性的任务规划技术影响月球车月面工作的环境因素有多种,主要包括月面地形、太阳位置和地球位置等。其中,日、地位置随时间变化。现有的环境建模方法研究主要集中在如何进行地形特征分类及建模上,对其他环境因素考虑较少。本文首先分析了各种月面环境因素对月球车月面工作过程的影响机制,提出了一种综合环境建模方法建立月面环境模型,作为开展月球车任务规划研究的基础。由于月球车在月面上的移动范围大,从一个探测目标点行驶到下一个目标探测点的时间长,整个过程中环境模型会发生显著变化,另外任务规划的目标往往也与时间、能源等相关。因此,月球车的行驶路径不能仅考虑躲避地形障碍,还要考虑其他影响路径选择的因素,这就是任务层路径。本文第4章研究了任务层路径规划问题的问题模型及其规划方法。针对月面环境的时变特点,提出了一种可预测动态模型下的路径规划方法,将动态环境下的路径规划转化为一系列静态环境下的路径全局寻优问题。行驶仅仅是月球车行为集合中的一种,月球车的月面工作过程是由多个行为有序排列而实现的。任务规划的目标就是规划出这样的行为序列,既全程满足所有约束条件,又能最好的实现任务目标。在本文第5章中,描述了月球车的行为集合并根据触发条件进行分类,在第4章提出的任务层路径规划方法的基础上,设计了月球车任务规划的一般方法,最终实现了月面工作行为序列的输出。月球车的月面工作过程往往存在许多不确定性,执行过程中行为时间消耗不可能完全与任务规划输出的结果一致。传统的确定性任务计划在实际应用中可能碰到困难。针对此问题,本文最后提出了与确定性计划相对的弹性计划概念,将行为开始时间由时刻变为时间区间,可大大提高任务规划输出结果在实际工程中的可行性。并且,设计了一种计算序列中各个行为开始时间区间的方法并进行了仿真实验。更多还原

【Abstract】 According to China’s lunar exploration program, it is scheduled to land probes on the moon in 2013, carrying out point detection and roving detection on the surface. The lunar rover is the one which will traverse on the moon, and it’s also the first Chinese mobile robot working on a celestial body besides earth. As a result of the limited intelligence of lunar rover and abundant communication opportunity to earth, ground-based teleoperation is chosen to be the main approach to control the lunar rover. Planning the activities sequence for a rover, ensuring it to achieve engineering and scientific goals without violating any operation constraints, is the primary mission of ground teleoperation system. Therefore, mission planning technology comes to be a key technology of teleoperation. However, as mission planning is a domain-related technology, neither approaches in other domains nor for other rovers (like mars rovers) can be used for lunar rover. So, developing a mission planning technology for lunar rovers is of great engineering significance and theoretic valuable.This research is directly driven by the practical needs of China’s lunar exploration program on planning for the lunar rover. Based on the review of research progress on environment modeling, a synthesized lunar environment model is built for mission planning, in which a mission-level path is proposed and planned. After analyzing the features of activities of lunar rover, a mission planning approach is finally proposed. It considers not only the interaction of mission-level path and path-dependent activity, but also the uncertainty of working on the lunar surface,There are some environmental factors affecting working progress on the moon, including lunar terrain, position of the sun and the moon. Most of the current researches on environment modeling focus on terrain classification and identification. On analyzing the affecting mechanism of environment factors to lunar rover, a synthesized environment modeling approach is proposed to model the lunar environment, which serves as the foundation of mission planning.It usually takes rover a long time to traverse form one goal to another, which is far apart. The environment is continuously varying. Moreover, mission goal is always related to time and energy. Traditional shortest path only avoiding terrain obstacle is not suitable for lunar rover anymore. Traversing on the moon should consider all the global factors. The path is named mission-level path. In chapter 3, mission-level path planning problem is first modeled, followed with a Predictable Dynamic Environment (PDE) path planning method. It turns planning a path on the continuously varying environment to a series of global path searching on static environments.Traverse is one of the activities performing on the moon, but not all. Working progress of lunar rover is directed by a mission plan comprising all kinds of activities. So, activities are descripted and categorized at the beginning of chapter 5. Then, a generalized mission planning method producing activity sequences for lunar rover is finally formulated, based on PDE mission-level path planning approach in chapter 4.Because of the uncertainties of working on the moon, time consumption of every activity during plan execution may not conform to the original plan. Traditional deterministic mission plan will run into difficulties when applying to practical use. To solve this problem, a new flexible plan is proposed, where the beginning time of an activity is of flexibility. It can enhance the practical ability of a mission plan greatly. What’s more, a fashion of computing the flexible interval of every activity in a sequence is given out and simulated.更多还原