Adaptive Square-Root Cubature Kalman Filter Based Low Cost UAV Positioning in Dark and GPS-denied Environments
Beiya Yang,Erfu Yang,Haobin Shi,Leijian Yu,Cong Niu
DOI: https://doi.org/10.1109/tiv.2024.3457678
IF: 8.2
2024-01-01
IEEE Transactions on Intelligent Vehicles
Abstract:Routine inspection inside the water tank, pressure vessel, penstocks and boiler which present dark and global positioning system (GPS) denied environment always plays an important role for the safety storage and transportation. The conventional inspection conducted by the skilled workers is highly expensive, time consuming and may cause the safety and heath problem. Nowadays, the emerging unmanned aerial vehicle (UAV) based techniques make it possible to replace human to do the periodical inspection in these environments. However, how to obtain the reliable, high accuracy and precise position information of the UAV becomes a challenging issue, as the GPS is unable to provide the accurate position information in these environments. In order to resolve this problem, an adaptive square-root cubature Kalman filter (ASRCKF) based low cost UAV positioning system is designed. Through the combination of the inertial measurement unit (IMU), ultra-wideband (UWB), the cubature rule, the adaptively estimated noise model and weighting factors, the potential degradation and oscillation for the system performance which caused by the linearisation process, the variation of the measurement noise and the manually adjusted noise model are solved. Finally, the 0.081m median localisation error, 0.172m 95 $^{th}$ percentile localisation error and 0.045m average standard deviation (STD) can be attained, which can support the UAV to achieve the autonomous inspection in dark and GPS-denied environments.