节点文献

空间机器人考虑接触碰撞的轨迹规划与跟踪控制研究

Planning and Control for Space Robot with Contact and Collision

【作者】 李晨;

【导师】 袁建平;

【作者基本信息】 西北工业大学 , 飞行器设计, 2022, 博士

【摘要】 随着航天事业的发展,在轨服务技术已成为日益重要的研究方向并引起了世界范围内的广泛关注。空间机器人技术作为在轨装配、维修、升级、燃料加注以及碎片清理等各类在轨服务任务的关键技术,有着至关重要的意义。传统的空间机器人研究尽力回避接触碰撞的影响,很少分析接触碰撞的作用,更不会利用接触碰撞进行操作,使得空间机器人局限在先稳固连接后实施操作这一传统操作模式中,既受制于抓捕机构与目标表面构造,也不能充分利用空间机器人的动态特性。随着机器人理论与硬件的快速发展,一种流畅且充分利用接触碰撞作用与系统动态特性的操作模式引起了机器人领域的关注。考虑到空间环境的微重力特性,这类基于接触碰撞的操作模式在空间任务中有着更良好的实施环境与广阔的应用前景。本论文针对空间机器人利用接触碰撞进行空间操作的任务,开展了从可操作性度量、轨迹规划到跟踪控制的一系列研究,尝试为这类新兴操作模式提供一套完整的实施方法。本文的主要内容和创新性的研究成果如下:(1)针对空间多臂机器人利用摩擦接触操作空间目标的组合体系统,考虑接触力的单边约束、摩擦力与内力关系、系统等效动力学等约束,研究了组合体系统的动力学可操作性并提出一种动力学可操作性度量方法。建立摩擦接触下空间操作系统的动力学模型,并将其转化为一个将空间机器人基座驱动和关节力矩映射到空间目标加速度、基座加速度和内力的显式表达式,用以依照可操作性的原始定义给出动力学可操作性度量。针对自由漂浮系统、自由飞行系统等空间操作的典型控制场景,分别推导了空间多臂机器人利用摩擦接触操作目标的动力学可操作性度量。所提出的动力学可操作性度量能够用于在轨服务任务设计阶段的机械臂参数优化及任务实施阶段的最优构型选择,也能够为控制参数设计提供参照。(2)针对空间机器人考虑接触碰撞的空间操作系统,开展了模态不变的轨迹优化方法研究。基于互补约束建模方法构造了空间机器人系统在接触环境下的隐式模态直接法优化问题。基于泊松假说,将接触环境下的优化问题构造方法推广至碰撞环境,构造了空间机器人系统在碰撞环境下的隐式模态直接法优化问题。考虑到所构造的优化问题非线性较强,求解困难,引入弹性放缩等技术,采用序列二次规划求解该优化问题。所得轨迹优化方法可以同步优化接触或碰撞系统的模态切换序列和局部最优轨迹,无需外部输入模态切换序列,所得轨迹能够充分利用系统的动态特性。(3)针对空间机器人利用反复碰撞进行在轨消旋的任务,考虑实际轨迹与参考轨迹碰撞时刻不一致的难点,开展了基于参考扩展误差和在线重规划的轨迹跟踪控制策略研究。实现通过空间机器人机械臂对自由漂浮环境中的被动目标施加反复碰撞使之跟踪其参考轨迹,同时空间机器人松散跟踪其参考轨迹。碰撞消旋系统被描述为混合系统,控制策略由预测、重规划和跟踪构成,为了克服碰撞时刻不一致的问题,还引入了参考扩展误差。考虑到空间目标的状态只能够通过其与空间机器人的碰撞改变,预测工作主要通过目标的当前状态及期望状态推测碰撞时刻的目标状态进而求解空间机器人在碰撞时刻的状态;在线重规划工作针对空间机器人,通过以其当前状态和碰撞时刻状态为边值约束的优化问题,在线重新规划空间机器人的参考轨迹并进行扩展;跟踪器针对空间机器人,用于对重规划所得的空间机器人轨迹进行跟踪控制。所设计的跟踪控制策略,能够在反复碰撞后,维持空间目标在其参考轨迹附近。(4)针对空间机器人利用接触进行在轨操作的任务,考虑系统长期行为与短期的收敛性与稳定性,开展了基于模型预测控制方法和控制障碍函数方法的分层异频轨迹跟踪控制策略研究。基于局部线性化的模型,设计基于混合整数规划模型预测控制的上层规划器,用以获得合适的系统长期行为及相应的接触模态切换序列。基于完整模型的各个单一模态,考虑模态约束、误差约束、系统原有约束等,设计了各单一模态下基于控制障碍函数的下层跟踪器,依照上层规划器所得接触模态切换序列选择下层跟踪器模态,实现上层规划器更新间隔内的下层跟踪器模态不变的控制。所设计的分层跟踪控制策略,能够兼顾系统长期行为的预测与短期内的稳定性,实现对空间机器人利用接触进行在轨操作任务轨迹的跟踪。

【Abstract】 On-orbit servicing technology has become an increasingly important research direction and has attracted worldwide attention with the development of space science and technology.Onorbit manipulation technology is of great significance as a key technology for on-orbit servicing missions,such as assembly,repair,upgrade,refueling,and active debris removal.Traditional on-orbit manipulation technologies sidestep the analysis of contacts by trying to avoid contacts and collisions,leading to a capture-and-operation schedule,which cannot take full advantage of the dynamics of space robots.A smooth operation,which makes full use of contact,collision,and dynamic characteristics of the system,has attracted the attention of the robotic community,with the rapid development of robotic theory and hardware.Contact and collision based on-orbit manipulations will have a better implementation environment and broader application prospects under a microgravity environment.In this dissertation,a series of studies from manipulability,trajectory optimization to tracking control are carried out for contact and collision based onorbit manipulation,in an attempt to provide a complete set of methods for such tasks.The main contents and innovations of the dissertation are as follows:(1)A dynamic manipulability measure method for contact based multi-arm on-orbit manipulation is proposed,considering unilateral constraints of contact forces,constraints between friction and internal forces,and equivalent dynamics of the system.The system is modeled under unilateral constraints of frictional contacts.A mapping from base actuation and joint torques to target acceleration,base acceleration,and internal forces is formulated based on the system dynamic.The dynamic manipulability measure is defined according to the original definition based on the mapping.The dynamic manipulability measures of multi-arm space robots are derived respectively for typical control scenarios of on-orbit manipulation system,such as free-floating system and free-flying system.The proposed dynamic manipulability measure can be used to optimize the manipulator parameters in mission design and find the optimal configuration in operations,and can also provide hint for the design of control parameters.(2)Direct trajectory optimization methods are developed for contact and collision based onorbit manipulation.Based on the complementary constraints formulation of contacts,a direct trajectory optimization method with an implicit contact model for an on-orbit manipulation system is formulated.Based on Poisson’s hypothesis,the direct trajectory optimization method is modified for collision based on-orbit manipulation.Several techniques are introduced to the sequential quadratic programming to solve the resulted nonlinear optimization problem,which is difficult to solve.The proposed trajectory optimization method can simultaneously optimize the contact/collision mode schedules and the local optimal trajectory of contact and collision based operation.The obtained trajectory can make full use of the dynamic characteristics of the system without external input of mode schedules.(3)A tracking control strategy based on the reference spreading error and online replanning is developed for the repeated-impact-based detumbling of space target,considering the collision time inconsistency between the actual trajectory and the reference trajectory.The goal of the control strategy is for the passive target in the free-floating environment to track the reference trajectory by repeated impacts from robotic manipulators,while robotic manipulators track their reference trajectories loosely.The repeated-impact-based detumbling system is described as a hybrid system,and the control strategy consists of prediction,replanning,and tracking.The reference spreading error is introduced to overcome the problem of inconsistent collision times.The prediction is about the states of the target and the chaser at collision moments calculated from current states and reference states,considering that the state of the target can only be changed by impacts of manipulators.Reference trajectories of the chaser system are then replanned online and extended forward and backward.A tracking law is employed to track the extended replanned trajectory of the chaser system.The proposed tracking control strategy can make the space target track its reference trajectory after repeated collisions.(4)A multi-rate hierarchical tracking control strategy leveraging model predictive control and control barrier functions is developed for the contact based space operation,considering the long-term behavior and short-term convergence and stability of the system.A highlevel planner,leveraging mixed-integer programming model predictive control based on a linearized model of the system,is designed to obtain the appropriate long-term behavior of the system.A low-level tracker,leveraging control barrier functions based on each contact mode of the system,is designed considering mode constraints,error constraints,and system constraints.The contact mode of the low-level tracker is optimized by the high level planner to realize mode invariant control in the continue flow of the system between the update interval of the high level planner.The designed multi-rate hierarchical tracking control strategy can take into account the long-term behavior prediction and short-term stability of the system,and realize the tracking of contact based on-orbit manipulation task.

  • 【分类号】TP242;V448
节点文献中: 

本文链接的文献网络图示:

本文的引文网络