8 July 2020 in IEEE Sensors Journal
The performance of the integrated global navigation satellite system (GNSS) and inertial navigation system (INS) has been proven to be more accurate, reliable, and continuous than stand-alone systems. The development of micro-electronic mechanical system (MEMS) enables inertial measurement unit (IMU) to meet the low-cost and small-size requirements of vehicular navigation. However, the stochastic error characteristics of the inertial sensors and the instability caused by the GNSS signal outages pose a threat to the MEMS-based GNSS/INS land vehicle navigation system. Within this context, we propose the following two-step GNSS/INS integrated architecture at two levels: 1) enhancing the signal-to-noise ratio (SNR) of MEMS-INS raw measurements utilizing a hybrid denoising algorithm with wavelet transform and support vector machine (SVM) and 2) improving the positioning accuracy by a SVM-based data fusion approach which could predict the accumulated error of the MEMS inertial sensors during the GNSS outages. A rotation platform experiment and a field test were carried out, which suggests that the proposed method can effectively eliminate the stochastic errors of MEMS-IMU, and significantly improve the overall positioning accuracy in land vehicle navigation.
17 June 2020 in IEEE Access
Static environment is a prerequisite for most existing vision-based SLAM (simultaneous localization and mapping) systems to work properly, which greatly limits the use of SLAM in real-world environments. The quality of the global point cloud map constructed by the SLAM system in a dynamic environment is related to the camera pose estimation and the removal of noise blocks in the local point cloud maps. Most dynamic SLAM systems mainly improve the accuracy of camera localization, but rarely study on noise blocks removal. In this paper, we proposed a novel semantic SLAM system with a more accurate point cloud map in dynamic environments. We obtained the masks and bounding boxes of the dynamic objects in the images by BlitzNet. The mask of a dynamic object was extended by analyzing the depth statistical information of the mask in the bounding box. The islands generated by the residual information of dynamic objects were removed by a morphological operation after geometric segmentation. With the bounding boxes, the images can be quickly divided into environment regions and dynamic regions, so the depth-stable matching points in the environment regions are used to construct epipolar constraints to locate the static matching points in the dynamic regions. In order to verify the preference of our proposed SLAM system, we conduct the experiments on the TUM RGB-D datasets. Compared with the state-of-the-art dynamic SLAM systems, the global point cloud map constructed by our system is the best.
2 March 2020 in IEEE Access
How to effectively blend global navigation satellite system (GNSS) and inertial navigation system (INS) data to achieve an optimal solution in harsh environments has always been an urgent task. The main challenges for low-cost GNSS/INS integrated land vehicle navigation system are poor accuracy of GNSS observations in complex urban environments and time-growing position errors of stand-alone micro-electro mechanical system (MEMS)-based inertial sensors during GNSS outages. This paper aims to enhance the positioning accuracy and reliability of low-cost integrated system. To attain this, we propose a two-tier robust fusion scheme with different aspects: 1) The GNSS and INS information are fused through a support vector regression-based adapted Kalman filter (SVR-AKF), with which scaling factors are generated to tune the covariance parameters of KF. 2) The position errors of MEMS-INS during GNSS outages are predicted and compensated by modeling INS error characteristics utilizing an adaptive neuro fuzzy inference system (ANFIS) due to its effectiveness in dealing with the nonlinear and uncertainty problems. To verify the feasibility of the proposed methodology, experimental road tests were performed, which suggested that the proposed methodology can significantly improve the overall reliability and positioning performance of land vehicle navigation in GNSS-challenged environments.