A collaborative SLAM method for dual payload-carrying UAVs in denied environments

In this paper, we investigate the problem of collaborative localization in search tasks in denied environments, particularly when traditional visual-inertial localization techniques reach their limits. A novel fusion localization method is proposed. It couples dual Payload-Carrying unmanned aerial vehicles (UAVs) using collaborative simultaneous localization and mapping (SLAM) techniques. This method aims to improve the system's search range and payload capacity. The paper utilizes SLAM technology to achieve self-motion estimation, reducing dependence on external devices. It incorporates a collaborative SLAM backend that provides the necessary information for system navigation, path planning, and motion control, ensuring consistent localization coordinates among the UAVs. Then, a joint localization optimization method based on Kalman filtering is introduced. By fusing the localization information from the visual sensors located beneath the UAVs and using the baseline variation between the 2 UAVs as a reference, the method employs a recursive prediction approach to jointly optimize the self-estimated states and the collaborative SLAM state estimates. Experimental validation demonstrates a 31.6% improvement in localization accuracy in complex tasks compared to non-fusion localization method. Furthermore, to address the cooperative trajectory tracking problem of UAVs after system path planning, a baseline-predicting fuzzy Proportional-Integral-Derivative flight controller is designed. Compared to conventional methods, this controller takes into account delays and system oscillations, achieving better tracking performance and dynamic adjustments.