基于分散式增广信息滤波的无人艇协同导航算法研究

To the problems of computational complexity，communication complexity and poor balance for the regular method about cooperative navigation such as KF.The subject studies unmanned surface vessels,useing decentralized data fusion skill to set algorithm framework ,introducing MRF to aid design and managementing communication with accumulation chained,the distributed incremental Cholesky modification algorithms and the method of decentralized state recovery to solve the problem of cooperative navigation of multi-vessels based on DAIF . Then we analysis the performance of the algorithm，to determine the algorithm's accuracy and superiority in improving the navigation's performance. The project is expected to reveal the great value and broad prospects of decentralized augmented information filter in the application of cooperative navigation of unmanned surface vessels. It will build a series of performance and indicators for DAIF, and achieve the computational complexity and communication complexity' reduction and the workload's balance of every unmanned surface vessel. The significance of the research results about this project is important for popularizing the applications of cooperative navigation in both military and civilian fields. So it can provide a new scientific basis for researching decentralized data fusion skills and cooperative navigation of unmanned surface vessels.

针对协同导航目前常用滤波方法的计算量和通信量过于复杂、工作均衡性不佳等问题，本课题以水面无人艇为研究对象，研究基于分散式增广信息滤波（DAIF）的多艇协同导航方法，建立算法框架，引入马尔可夫随机场进行辅助设计，提出分布式递增Cholesky修正算法，使用"累积链式"通信管理，完成状态恢复在动态分布式环境中的实现。分析该算法的性能，确定本算法在提高系统导航性能方面的正确性和可行性。项目预期将实现对计算复杂度和通信复杂度的降低，平衡各无人艇的工作负载，揭示分散式增广信息滤波在无人艇协同导航应用方面的巨大价值和广阔前景。项目的研究成果对于推广无人艇协同导航在军事和民用等领域的应用具有重要意义，可以为研究分散式数据融合技术和无人艇的协同导航应用提供新的科学依据。

以水面无人艇编队导航为应用背景，基于增广信息滤波设计一种分布计算型协同导航算法。本算法在建立状态空间模型和概率图模型的基础上，提出动态分块矩阵的递增克勒斯克修正算法，以及通信规则设计，并引入马尔卡夫随机场进行辅助设计，通过艇间协作实现状态恢复，从而完成增广信息滤波在多艇协同导航分布式环境中的实现。建立计算复杂度等指标分析多艇协同导航框架下DAIF算法性能。本课题所提出的算法经过了水上实验的验证，证明在保证协同定位精度的条件下，有效的降低运算量和工作负担。本课题的研究可以有效推动分布式传感器网络信息融合技术的发展。