【作者】 奔粤阳； 郭妍； 阮双双； 洪勇；

【Author】 BEN Yue-yang;GUO Yan;RUAN Shuang-shuang;HONG Yong;College of Automation,Harbin Engineering University;Leador Spatial Information Technology Corporation;

【机构】 哈尔滨工程大学自动化学院； 立得空间信息技术股份有限公司；

【摘要】 为了提高现有分散式数据融合算法在计算量和通信量上的综合性能,提出了一种新的分散式数据融合算法,运用在多无人艇的协同导航中。该算法为了降低计算量,首先借鉴机器人同步构图与定位,利用联合分布状态将关键历史状态保留在滤波中,避免时间更新的复杂计算;然后利用滤波信息参数的稀疏性,减小滤波所涉及的计算复杂度;最后根据Cholesky矩阵分解的特殊性质,进一步减少计算复杂度。同时该算法为了降低通信量,通过环形多节点通信方式和删除冗余信息等操作,保持单个无人艇有限的存储容量,实现通信的有效管理。结果表明,在多无人艇协同导航系统中,该算法较传统的分散式卡尔曼滤波算法,定位达到同等精度,但是其通信量更小,计算量更少,证明了算法的有效性和正确性。更多还原

【Abstract】 In order to improve the overall performance of the existing decentralized data fusion algorithm both in the calculation and the communication,a new decentralized data fusion is proposed used for the cooperative navigation algorithm of unmanned surface vessels.Firstly,in order to reduce the amount of the calculation,the algorithm makes use of the joint distribution state to remain the critical history state in the filter,which is based on the Simultaneous Localization and Mapping for robots; and then,taking advantage of the sparsely of the information parameter for reducing the computational complexity; finally,according to the special property of the Cholesky matrix decomposition,the algorithm further reduces the computational complexity.Meanwhile,the algorithm achieves the effective communication management in the way of maintaining the limited storage of a single unmanned surface vessel,through using annular multi-node communication and removing redundant information.The results showed that in the system of the cooperative navigation algorithm of multi-USV,the algorithm achieves the same positioning accuracy,but smaller traffic communication and fewer calculations compared with the traditional decentralized Kalman filter algorithm,which proves the validity and correctness of the algorithm.更多还原