Yahui Zhang, Linxuan Wang, Aimin Li, Yongsheng Zheng, and Mingzhuang Wu, 195-203.
State estimation, sensor fusion, tight coupling, precise positioning
Aiming at the problem that the state estimation of visual- inertial-odometre (VIO) is affected by long-term trajectory drift accumulation, a VIO compact coupling system based on optimisation is proposed. To estimate all the initial state variables of the system effectively, a method combining static initialisation and dynamic initialisation is proposed. To realise the transformation between local frames and local Cartesian coordinates coordinate system (ENU) coordinates of the local Cartesian coordinates coordinate system, local state and global navigation satellite system (GNSS) measurement information are fused. A noise reestimation method is used to ensure the simultaneity of inertial measurement unit (IMU) and wheel speedometre. We evaluated the proposed system on a public data set and tested it in a real-world scenario. The experimental results show that the VIO tightly coupled system based on the optimisation can present better positioning effect and improve the precision of global fusion. The system can achieve accurate attitude estimation in the outdoor environment.
Important Links:
Go Back
