DS-ASFGCEE: A Novel Multi-Node Adaptive Cooperative Localization Algorithm Under Unknown Time-Varying Noise

The underwater cooperative localization system for multiple Autonomous Underwater Vehicles (AUVs) has been widely applied in marine operations and explorations. However, the significant amount of uncertainty information present in the system can affect its localization performance. Therefore, this paper proposes the dynamic structure-based adaptive selection factor graphs and center error entropy (DS-ASFGCEE) cooperative localization algorithm for multi-AUV suitable for complex underwater environments. In this paper, the Center Error Entropy (CEE) criterion is first introduced to optimize the factor graph-based cooperative localization system, reducing the impact of non-Gaussian noise in measurement information on localization performance. Additionally, measurement information received multiple times is modeled, and adaptive estimation is performed on unknown measurement noise variances. Finally, the factors influencing observability in the multi-AUV cooperative localization system are analyzed. By constructing a node selection matrix, follower AUVs are controlled to adaptively select high-quality nodes for localization. Through comprehensive simulation experiments and analyses, the proposed algorithm demonstrates higher robustness and accuracy compared to the comparison algorithms when facing various complex and extreme underwater environments.