SARCL: A Search-and-Rescue Oriented Event-Triggered Resilient Collaborative Localization Scheme Against Random Packet Dropout for AAV Platoon Under Limited Communication Conditions

The limited communication resources and the degradation of global navigation satellite system (GNSS) have posed formidable challenges for robust localization of search-and-rescue autonomous aerial vehicles (SAR-AAVs), which is an essential prerequisite to successful SAR missions. Consequently, depending on the mutual geometry relationships of a SAR-AAV platoon, this article proposes an event-triggered resilient collaborative localization scheme by leveraging intervehicle relative range-azimuth-elevation observations under GNSS-denied or anchor-free conditions. More specifically, we first integrate the event-triggered mechanism into a Bernoulli packet-loss model to enhance the collaborative localization resilience against random packet dropout of cooperative messages, as well as optimize the communication resource utilization. On this basis, for addressing the unsolvable problem of error covariance matrix, we further derive a stochastic distributed Kalman filter (SDKF)-based collaborative localization algorithm, where the minimum upper bound of error covariance matrix is employed to obtain the recursive state estimation formulas of a SAR-AAV platoon instead of the unsolvable analytical solution. In addition, the convergence conditions of SDKF are also provided by deducing the stochastic boundedness of estimation error. Finally, three P600 AAVs equipped with the onboard ultra-wideband module, which can jointly obtain the mutual relative range-azimuth-elevation information, are presented to verify the effectiveness of the proposed method, and the corresponding experimental results show that the proposed method can achieve resilient state estimates subject to random packets dropout compared with the state-of-the-art methods.