Research roundup: GNSS in urban canyons

GNSS researchers presented hundreds of papers at the 2022 Institute of Navigation (ION) GNSS+ conference, which took place Sept. 19-23, 2022, in Denver, Colorado, and virtually. The following four papers focused on the use of GNSS in urban environments. The papers are available here.

GPS World will be attending this year’s ION conference in Denver, Colorado, on Sept. 11-15.

FGO-based GNSS/INS integration improves performance in urban canyons in Hong Kong

The integration of GNSS and inertial navigation systems (INS) has the potential to improve performance due to their complementariness. In this paper, the authors investigated positioning based on the integration of GNSS and INS using factor graph optimization (FGO). This ultimately showed improved performance in urban canyons in Hong Kong. The effectiveness of the proposed method was verified using challenging datasets collected using two automobile-level GNSS receivers in the urban canyons of Hong Kong.

For the experiment conducted in this paper, only the GNSS pseudorange measurement was utilized in the existing FGO-based GNSS/INS integration. The overall potential of the Doppler frequency and carrier-phase measurements has yet to be explored by the authors. To fill this gap, the authors proposed a tightly coupled GNSS/INS integration, using FGO, by exploiting the potential of diverse raw GNSS measurements. The GNSS pseudorange, Doppler frequency, and time-differenced carrier-phase measurements were integrated with the INS, using FGO.

The authors believe the improved performance using FGO-based GNSS/INS integration positioning was due to the global optimization property and the increased measurement redundancy of FGO, compared with the method based on extended Kalman filtering.

Weisong, Hsu; “Factor Graph Optimization for Tightly-Coupled GNSS Pseudorange/Doppler/Carrier Phase/INS Integration: Performance in Urban Canyons of Hong Kong.”

3D mapping in urban environments aided by surround mask GNSS/lidar SLAM

Automatic driving with coupled GNSS/INS and lidar sensors has been implemented in many urban environments successfully over the years. However, this technology is still prone to errors. These potential errors are especially evident in challenging environments, such as urban canyons with several moving objects and building layouts that provide unexpected and abnormal features for lidar sensors and multi-path for GNSS signals.

To address these error challenges in urban environments, the authors of this paper proposed a surround mask that explores error sources from surrounding environments, which could subsequently improve the performance of an integrated mapping system. The surround mask in this experiment extracted a two-layer factor, including non-line-of-sight detection and static objects detection, to collectively compensate for the specific drawbacks of the lidar-based SLAM and the navigation system.

The authors explain that the surround mask eliminated the need to apply complex post-processing to eliminate the accumulated error for each observing unit.

The experimental results demonstrated that the proposed surround mask detected the represented error sources in the local coordinate and provided environment-awareness information for the integrated mapping system.

Ai, Luo, El-Sheimy; “Surround Mask Aiding GNSS/LiDAR SLAM for 3D Mapping in the Dense Urban Environment.”

Novel process noise model helps GNSS Kalman filter degradation in busy cities

Improving the accuracy of GNSS positioning in urban environments is difficult, especially when using low-cost GNSS receivers. In this paper, the authors showed that if the process noise covariance is turned up in a “naïve” manner for poor satellite geometry, the estimation-error covariance could become unintentionally large in a certain direction.

The unintentional inflation of estimation-error covariance could cause the degradation of accuracy. The authors also proposed a fictitious process noise covariance based on an extension of a novel process noise model, which was proposed in their previous work.

The authors stated that in Kalman filter for GNSS positioning, the process noise covariance is often bumped up to avoid the filter divergence in the presence of unknown model errors, by assuming there is a fictitious process noise in addition to the nominal process noise. In this study, the fictitious noise covariance is determined based on the observation matrix, step-by-step, and it reduced the estimation errors without causing the unintentional inflation of estimation-error covariance.

The effectiveness of the derived process noise model is demonstrated for the data sets that simulate GNSS signals from the antenna that moves from open sky areas to urban areas. The estimation errors with the derived process noise model were significantly reduced, compared to the ones with other two process noise models.

Ai, Luo, El-Sheimy; “Surround Mask Aiding GNSS/LiDAR SLAM for 3D Mapping in the Dense Urban Environment.”

3D lidar-aided GNSS RTK positioning for increased accuracy mapping in urban canyons

The GNSS real-time kinematic (RTK) positioning technique has shown centimeter-level absolute results in open-sky areas; however, it can suffer from polluted GNSS measurements and poor satellite geometry in urban environments. This is due to the non-line-of-sight (NLOS) and multipath reception caused by signal blockage and reflection.

In this paper, the authors stated that lidar sensors integrated with odometry systems that include an inertial measurement unit (IMU) provided a precise environment description and short-term accurate relative positioning capabilities that could be utilized for aiding GNSS-RTK to obtain better performance.

While 3D lidar-aided GNSS RTK positioning methods detect the GNSS NLOS receptions via an incrementally built map and improve the satellite geometry using the low-lying virtual satellite from lidar features, the high-elevation angle NLOS receptions cannot be fully detected, and the multipath signals cannot be effectively mitigated.

In response to this, the authors proposed a 3D lidar-aided GNSS RTK positioning method with iterated coarse to fine batch optimization by a global 3D NLOS exclusion aided by a point cloud map, which enables the detection of high-elevation angle NLOS receptions. Additionally, the authors proposed iterated batch optimization based on a devised, tightly coupled, factor graph that fully exploited the global consistency among the constraints of lidar, IMU and GNSS RTK to exclude potential multipath signals.

The proposed method aimed to achieve lifelong accurate positioning performance in deeply urbanized areas. The effectiveness of the proposed method has been proved by the evaluation conducted on the author’s open-source challenging dataset, UrbanNav, which contains various sequences collected by automobile-level low-cost GNSS receivers in urban canyons of Hong Kong.

Liu, Wen, Hsu; “3D LiDAR Aided GNSS Real-time Kinematic Positioning via Coarse-to-fine Batch Optimization for High Accuracy Mapping in Dense Urban Canyons.”