Abstract:

In response to the challenges of positioning, mapping, and path planning for UAV inspections caused by the complex structures and strong electromagnetic interferences in substations, an autonomous UAV inspection method based on factor graph optimization for multi-sensor fusion positioning and multi-level trajectory planning was innovatively proposed. At the positioning level, the factor graph optimization framework was utilized to deeply couple laser odometry, inertial measurement unit (IMU) pre-integration, global navigation satellite system (GNSS)/real-time kinematic (RTK) observations, and loop detection, forming a multi-sensor simultaneous localization and mapping (SLAM) system with stable anti-interference capabilities and high precision characteristics. At the planning level, after generating a smooth global flight path using the minimum snap polynomial, cardinal splines were introduced to refine and adjust the trajectory, and real-time obstacle avoidance and dynamic feasibility constraints were integrated into the system. The results show that this method can maintain centimeter-level positioning accuracy and low map registration error in indoor and outdoor substation environments of different voltage levels, and it significantly improves stability, flight efficiency, and robustness compared to single-sensor or conventional satellite navigation solutions. The study indicates that the combination of factor graph optimization-driven multi-sensor deep fusion and multi-level trajectory planning provides a new path for autonomous UAV inspection in strong electromagnetic interference scenarios, with both high precision and high robustness.

Key words: strong electromagnetic interference, substation, factor graph optimization, unmanned aerial vehicle (UAV), inspection