Resumen
With the continuous expansion of the application field of UAV intelligent systems to GNSS-denied environments, the existing navigation system can hardly meet low cost, high precision, and high robustness in such conditions. Most navigation systems used in GNSS-denied environments give up the connection between the map frame and the actual world frame, making them impossible to apply in practice. Therefore, this paper proposes a Lidar navigation system based on global ArUco, which is widely used in large-scale known GNSS-denied scenarios for UAVs. The system jointly optimizes the Lidar, inertial measurement unit, and global ArUco information by factor graph and outputs the pose in the real-world frame. The system includes a method to update the global ArUco confidence with sampling, improving accuracy while using the pose solved from the global ArUco. The system uses the global ArUco to maintain navigation when Lidar is degraded. The system also has a loop closure determination part based on ArUco, which reduces the consumption of computing resources. The navigation system has been tested in the dry coal shed of a thermal power plant using a UAV platform. Experiments demonstrate that the system can achieve global, accurate, and robust pose estimation in large-scale, complex GNSS-denied environments.