基于SRCKF的多移动机器人协同定位与目标跟踪研究

发布时间：2019-04-15 18:38

【摘要】：随着机器人技术的不断发展,多移动机器人系统因其高运行效率、强鲁棒性和广泛的应用领域受到了越来越多学者的关注和研究。能够对未知复杂外界环境感知、建模并确定自身的位置,是多移动机器人自主导航的前提和基础。现实中的很多任务仅凭借传统的定位问题难以解决,此时就要求将移动机器人定位与目标跟踪方法结合起来。本文致力于解决这两个相互耦合的问题:多移动机器人自身状态估计和对于目标的状态估计。重点围绕协同定位和目标跟踪两项关键内容展开研究,针对这两项内容分别建立了相应的模型,提出了相应的优化算法,并对所提算法进行了仿真实验,分析了实验结果。多机器人协同定位部分,分析了多移动机器人协同系统的体系结构和通信方法,针对单移动机器人在探索未知复杂环境时,存在鲁棒性较差,效率较低等问题以及现有多机器人协同定位算法实时性、和定位精度较差等缺陷,提出基于平方根容积卡尔曼滤波的相对方位多机器人协同定位算法。利用相对方位作为测量值,在更新过程中直接传递目标状态均值和协方差矩阵的平方根因子,精确度更高,更稳定。计算均值和方差时采用基于容积准则的数值积分方法,降低了计算复杂度,实时性强。仿真实验表明了该算法的精确性和有效性。多机器人协同目标跟踪部分,针对移动机器人在未知复杂环境中动态目标追踪存在的数值不稳定、计算量大和精度较差等问题,提出基于平方根容积卡尔曼滤波的移动机器人动态目标跟踪算法,该算法的系统状态由地图环境特征、机器人和目标作为一个整体构成,通过数据关联环节能够有效的降低伪观测值对系统状态估计的影响。仿真结果表明了该算法的合理性和可操作性。针对未知环境下多机器人协同目标跟踪问题,提出基于协方差交集的多机器人协同目标跟踪算法。此算法具有分布式特点,在提高相关对象状态估计准确性的同时,不必对数据信息进行独立性假设,避免了对象状态间的互相关性估计,降低了系统通信能量损耗和计算复杂度。通过仿真实验证明了该算法能够有效解决未知环境下多机器人协同目标跟踪问题。
[Abstract]:With the development of robot technology, the multi-mobile robot system has attracted more and more attention and research due to its high operating efficiency, strong robustness and wide range of applications. It is the premise and foundation of autonomous navigation for multi-mobile robot to be able to perceive the unknown and complex external environment, model and determine its own position. Many tasks in reality are difficult to solve only by traditional positioning problem, so it is necessary to combine mobile robot localization with target tracking method. In this paper, we focus on solving these two coupling problems: state estimation of multi-mobile robot and state estimation of target. Focusing on the two key contents of collaborative positioning and target tracking, the corresponding models are established, and the corresponding optimization algorithms are proposed. Simulation experiments are carried out on the proposed algorithms and the results of the experiments are analyzed. In the part of multi-robot cooperative positioning, the architecture and communication method of multi-mobile robot cooperative system are analyzed. The robustness of single mobile robot in exploring unknown and complex environment is poor. Based on the problems of low efficiency, real-time performance and poor positioning accuracy of existing multi-robot cooperative localization algorithms, a relative azimuth multi-robot co-location algorithm based on square root volume Kalman filter is proposed. Using the relative azimuth as the measured value, the square root factor of the target state mean and covariance matrix is transferred directly during the renewal process. The accuracy is higher and the stability is more stable. When calculating the mean and variance, the numerical integration method based on the volume criterion is adopted, which reduces the computational complexity and has a strong real-time performance. Simulation results show that the algorithm is accurate and effective. In the part of multi-robot cooperative target tracking, aiming at the problems of dynamic target tracking of mobile robot in unknown and complex environment, such as numerical instability, large amount of computation and poor precision, and so on. A moving target tracking algorithm for mobile robot based on square root volume Kalman filter is proposed. The system state of the algorithm is composed of map environment feature, robot and target as a whole. The influence of pseudo-observation value on system state estimation can be effectively reduced by data correlation link. The simulation results show that the algorithm is reasonable and feasible. A multi-robot cooperative target tracking algorithm based on covariance intersection is proposed to solve the problem of multi-robot cooperative target tracking in unknown environment. This algorithm has distributed characteristics. While improving the accuracy of the state estimation of related objects, it does not need to assume the independence of the data information, thus avoiding the cross-correlation estimation between the states of the objects. The energy loss and computational complexity of the system are reduced. The simulation results show that the algorithm can effectively solve the problem of multi-robot cooperative target tracking in unknown environment.
【学位授予单位】：安徽工程大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP242

【参考文献】