Research on anomaly data detection algorithms for vision/inertial navigation systems
Fault detection (FD)methods for sensor anomaly data in visual/inertial navigation systems (VINS)are essential for improving the positioning performance and reliability of the system.However,research on anomaly data detection and troubleshooting methods based on visual/inertial navigation systems is scarce.This paper proposes an anomalous data detection algorithm based on sensor measurement residuals (VINS-ORFD)to improve system reliability by actively detecting and filtering camera and IMU anomalous data.Our test results based on the TUM dataset show the algorithm not only achieves fast detection of vision and IMU sensor anomaly data,but also improves the positioning accuracy (RMSE)by at least 22.66％.
国家科技期刊平台
NSTL
万方数据
