Multi-Sensor Fusion and Real-Time FDE for UGVs Based on Factor Graph Optimization

Jiahao Xu, Hongxia Wang, Zhiqiang Dan, Zhipeng Wang, Yanbo Zhu

Abstract:	As an emerging application, unmanned ground vehicles (UGVs) travel in complex environments, such as open-sky roads, urban canyons, tunnels or underground garages. It is not enough to simply combine global navigation satellite system (GNSS), inertial navigation system (INS) and light detection and ranging (Lidar) together. To achieve sub decimeter level position accuracy and higher reliability, a multi-sensor fusion and real-time fault detection and exclusion (FDE) for UGVs based on factor graph optimization is proposed. After executing FDE algorithm, six types of sensors are combined, including GNSS, INS, Lidar, wheel speedometer, barometric altimeter and magnetometer. To avoid trajectory drift caused by accidental GNSS failure, loop closure constraints have been added as extra measurement. To execute FDE algorithm based on factor graph, test statistic based on factor measurement residual have been constructed. For situations where multiple sensors measure at the same time, the sensor measurement with a higher false alarm rate will be used to construct test statistic first. Two real-world experiments are used to verify our algorithm. Finally, experimental results show that our algorithm can achieve sub decimeter level position accuracy in complex urban canyons, and effectively detect and exclude faults for higher reliability.
Published in:	Proceedings of the 36th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2023) September 11 - 15, 2023 Hyatt Regency Denver Denver, Colorado
Pages:	748 - 762
Cite this article:	Xu, Jiahao, Wang, Hongxia, Dan, Zhiqiang, Wang, Zhipeng, Zhu, Yanbo, "Multi-Sensor Fusion and Real-Time FDE for UGVs Based on Factor Graph Optimization," Proceedings of the 36th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2023), Denver, Colorado, September 2023, pp. 748-762. https://doi.org/10.33012/2023.19238
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In