集成逆超短基线的自主式水下航行器集群协同定位方法

Cooperative localization methods of autonomous underwater vehicle swarm integrated with inverted ultrashort baseline

构建了集成逆超短基线的自主式水下航行器(AUV)协同定位因子图,提出了两种从AUV协同定位方法,即基于扩展卡尔曼滤波(EKF)协同定位方法和基于粒子置信度传播(BP)协同定位方法,并修正了主AUV定位误差的影响。其中,基于EKF的从AUV协同定位方法利用主AUV的定位信息,重构了从AUV的状态向量实现协同定位;而基于粒子BP的从AUV协同定位方法构建了测量粒子实现协同定位。所提出的方法能够解决传统方法实时性较差的问题,且仅需单程声学定位信号传输测量即可实现从AUV的定位。仿真结果表明,两种协同定位方法适用于不同定位能力的从AUV,均能实现较高精度的协同定位。相比于航位推算方法,两种从AUV协同定位方法的定位误差均减小80%以上,且有效抑制了定位误差累积问题。此外,在水声信道较高丢包率的情况下,所提出的协同定位方法能稳定工作,且定位性能没有明显受到影响。

A cooperative localization factor graph for the autonomous underwater vehicle(AUV) swarm integrated with inverted ultrashort baseline(iUSBL) system is constructed.Two cooperative localization methods of slave-AUV in an AUV swarm are proposed,namely the extended Kalman filter(EKF)-based cooperative localization method and particle belief propagation(BP)-based cooperative localization method,where the effect of the leader-AUV 's positioning error is mitigated.For the EKF-based cooperative localization method,the state of the slave AUV is reconstructed with the positioning information of leader-AUV to achieve cooperative localization.In the particle-BP-based cooperative localization method,the measurement particles are generated to realize cooperative localization.The proposed methods can solve the problem of poor real-time of traditional methods,and only require one-way acoustic positioning transmission measurement to realize the localization of the slave-AUV.Simulation results show that the two proposed cooperative localization methods of slave-AUV in an AUV swarm work well for AUVs with different localization abilities.Compared with the dead-reckoning method,the localization errors of two proposed methods are reduced by more than 80%,and the positioning error accumulation is effectively suppressed.Furthermore,the proposed cooperative localization methods can operate normally under a high packet loss rate of the underwater acoustic channel,and the localization performance is not significantly degraded.