While the global navigation satellite system (GNSS) has been widely used to provide high-precision location services in many applications, it usually suffers from performance degradation due to non-line-of-sight (NLOS) reception. As the received NLOS signals might have great measurement errors especially in urban canyons, they should be detected to mitigate the errors contaminating the positioning systems. However, the NLOS detection is quite challenging as the accuracy rate is usually highly related to the surrounding environment the receiver is located in. To address this problem, we propose a stacking ensemble learning (SEL) method for the NLOS detection of GNSS. First, satellite measurement features are extracted from the GNSS raw measurements via a designed data processing module. Then, they are input to the SEL module consisting of two levels of machine learning models. In the first level, a support vector machine (SVM) and an extreme gradient boosting (XGBoost) are adopted in parallel, and the outputs of the fist-level models are input to the second-level logistic regression (LR) to obtain NLOS predictions. The proposed SEL module combines the views of different models to the measurement features to address the shortcomings of each single model and improve the model’s generalization. Experimental results on real GNSS observations in urban canyons show that the proposed method outperforms the baseline machine learning methods with obvious detection accuracy improvements.
The horizon line has numerous applications for an unmanned surface vehicles (USV), such as autonomous navigation, attitude estimation, obstacle detection and target tracking. However, maritime horizon line detection is quite a challenging problem. The pixel points of the horizon line features are far fewer than the pixel points of the entire image, on the one hand. Conversely, the detection results might be impacted negatively by the complex maritime environment, waves, light changing, and partial occlusions due to maritime vessels or islands, for example. To solve these problems, a robust horizon line detection method named coarse-fine-stitched (CFS) is proposed in this paper. First, in the coarse step of CFS, a line segment detection approach using gradient features is applied to build a line candidate pool, which probably contains many false detection results. Then, hybrid feature filtering is designed to pick the horizon line segments from the pool in the fine step. Finally, the fine line segments are stitched to obtain the whole horizon line based on random sample consensus (RANSAC). Using real data in the maritime environment, the experimental results demonstrate the effectiveness of CFS, compared to the existing methods in terms of accuracy and robustness.
In the 1980s, Global Positioning System (GPS) receiver autonomous integrity monitoring (RAIM) was proposed to provide the integrity of a navigation system by checking the consistency of GPS measurements. However, during the approach and landing phase of a flight path, where there is often low GPS visibility conditions, the performance of the existing RAIM method may not meet the stringent aviation requirements for availability and integrity due to insufficient observations. To solve this problem, a new RAIM method, named vision-aided RAIM (VA-RAIM), is proposed for GPS integrity monitoring in the approach and landing phase. By introducing landmarks as pseudo-satellites, the VA-RAIM enriches the navigation observations to improve the performance of RAIM. In the method, a computer vision system photographs and matches these landmarks to obtain additional measurements for navigation. Nevertheless, the challenging issue is that such additional measurements may suffer from vision errors. To ensure the reliability of the vision measurements, a GPS-based calibration algorithm is presented to reduce the time-invariant part of the vision errors. Then, the calibrated vision measurements are integrated with the GPS observations for integrity monitoring. Simulation results show that the VA-RAIM outperforms the conventional RAIM with a higher level of availability and fault detection rate.
In this paper, we propose an incremental learning method for end-to-end Automatic Speech Recognition (ASR) which enables an ASR system to perform well on new tasks while maintaining the performance on its originally learned ones. To mitigate catastrophic forgetting during incremental learning, we design a novel explainability-based knowledge distillation for ASR models, which is combined with a response-based knowledge distillation to maintain the original model's predictions and the “reason” for the predictions. Our method works without access to the training data of original tasks, which addresses the cases where the previous data is no longer available or joint training is costly. Results on a multi-stage sequential training task show that our method outperforms existing ones in mitigating forgetting. Furthermore, in two practical scenarios, compared to the target-reference joint training method, the performance drop of our method is 0.02% Character Error Rate (CER), which is 97% smaller than the drops of the baseline methods.
17 October 2024 in IEEE Transactions on Vehicular Technology
The widespread adoption of Global Navigation Satellite Systems (GNSS) has revolutionized high-precision positioning services. However, the accuracy of these systems is significantly compromised by Non-Line-of-Sight (NLOS) signals, especially in urban canyons. Specifically, Line-of-Sight (LOS) signals are obstructed, and the reflections lead to substantial errors. Machine learning has emerged as a promising tool for identifying NLOS signals. However, traditional methods often train models on extensive datasets from various environments and fine-tune them with new data to adjust to new scenarios. This fine-tuning can result in catastrophic forgetting, which impairs the models' generalization performance across different environments. To address the limitations of model fine-tuning, we introduce NICL, an approach for N LOS I dentification with C ontinual L earning. NICL preserves performance on previously encountered scenarios while adapting to new environments by employing simple regularization techniques that do not require access to the data from the old scenarios. Specifically, our method incorporates Kullback-Leibler divergence regularization and explainability-based regularization to maintain the model's prior outputs and the logic behind those outputs, thus mitigating catastrophic forgetting. Experimental results on real-world GNSS datasets have shown that our NICL method outperforms existing baseline methods, achieving an absolute improvement of 5%-12% in NLOS identification accuracy.