信息受限条件下多自主水下航行器三维运动目标围捕控制方法

Three-dimensional Encirclement Control of Multiple AUVs for Moving Target under Limited Information

摘要:

针对水下环境中自主水下航行器(Autonomous underwater vehicle,AUV)无法获取绝对位置信息以及部分AUV因感知受限难以持续获取目标信息的问题,本文提出一种信息受限条件下多AUV三维运动目标球状围捕控制方法,以实现对目标的近距离全方位监测。首先,基于有向环通信拓扑设计分布式目标相对位置–速度估计器,使目标不可见AUV通过局部信息交互获得目标状态估计;其次,以目标运动方向为虚拟围捕旋转轴构建辅助系统,建立多AUV三维球状围捕队形;在此基础上,设计基于局部相对位置信息的围捕运动学控制器,并结合AUV动力学模型设计速度跟踪控制器。理论分析证明,在理想条件下局部跟踪误差指数收敛,考虑目标运动扰动和状态估计误差时,闭环系统误差最终有界。仿真结果表明,在目标信息完全可得和部分缺失2种情况下,多AUV均能够形成稳定的三维球状围捕队形,当目标信息部分缺失时,相对位置–速度估计误差可快速收敛至零附近,围捕半径和分离角均趋近于期望值。研究结果表明,所提方法能够在缺乏绝对位置信息和目标信息不完全可得条件下实现三维运动目标稳定围捕,为复杂水下环境中的多AUV协同目标监测提供了有效解决方案。

Abstract:

The three-dimensional encirclement control problem of a moving target was investigated by multiple autonomous underwater vehicles (AUVs) under limited information conditions, where absolute position information was unavailable and some AUVs cannot continuously obtain target measurements due to sensing constraints. To achieve close-range omnidirectional monitoring of the target, a distributed relative position – velocity estimator was firstly developed over a directed ring communication topology, enabling AUVs without direct target observations to estimate the target state through local information exchange. Subsequently, a virtual auxiliary system was constructed by taking the target motion direction as the virtual encirclement axis, thereby generating a three-dimensional spherical encirclement formation. Based on the estimated relative states, a kinematic encirclement controller using only local relative position information was designed, and a velocity tracking controller was further developed by incorporating the AUV dynamic model. Stability analysis showed that the local tracking errors converged exponentially under ideal conditions, while the closed-loop system remained ultimately bounded in the presence of target motion disturbances and estimation errors. Simulation results demonstrated that the proposed method enabled multiple AUVs to form and maintain a stable three-dimensional spherical encirclement formation under both complete and partial target information scenarios. Moreover, when target information was partially unavailable, the relative position – velocity estimation errors rapidly converged to a neighborhood of zero, and both the encirclement radius and separation angles approached their desired values. The proposed framework provided an effective solution for three-dimensional moving-target encirclement under limited information conditions, featuring an integrated design of distributed target-state estimation and spherical encirclement control without relying on absolute position information.