Andrew G. Dempster

Sanat Biswas, Li Qiao, Andrew Dempster

The Extended Kalman Filter (EKF) is a well established technique for position and velocity estimation. However, the performance of the EKF degrades considerably in highly non-linear system applications as it requires local linearisation in its prediction stage. The Unscented Kalman Filter (UKF) was developed to address the non-linearity in the system by deterministic sampling. The UKF provides better estimation accuracy than the EKF for highly non-linear systems. However, the UKF requires multiple propagations of sampled state vectors in the measurement interval, which results in higher processing time than for the EKF. This paper proposes an application of two newly developed UKF variants in launch vehicle navigation. These two algorithms called the Single Propagation Unscented Kalman Filter (SPUKF) and the Extrapolated Single Propagation Unscented Kalman Filter (ESPUKF), reduce the processing time of the original UKF significantly and provide estimation accuracies comparable to the UKF. The estimation performance of the SPUKF and the ESPUKF is demonstrated using Falcon 9 V1.1 launch vehicle in CRS-5 mission scenario. The launch vehicle trajectory for the mission is generated using publicly available mission parameters. A SPIRENT GNSS simulator is used to generate the received GPS signal on the trajectory. Pseudo-range observations are used in the EKF, UKF, SPUKF and the ESPUKF separately and the estimation accuracies are compared. The results show that the estimation errors of the SPUKF and the ESPUKF are 15.44% and 10.52% higher than the UKF respectively. The processing time reduces by 83% for the SPUKF and 69.14% for the ESPUKF compared to the UKF.

Sanat Biswas, Li Qiao, Andrew Dempster

Application of two new UKF based estimation techniques with reduced processing time in re-entry vehicle position and velocity estimation problem using ground-based range and elevation measurements is presented. The first method is called the Single Propagation Unscented Kalman Filter (SPUKF) where, the a postiriori state is propagated only once and then the sampled sigma points at the next time state are approximated by the first-order Taylor Series terms. In the second method called the Extrapolated Single Propagation Unscented Kalman Filter (ESPUKF), the sigma points are approximated to the second-order Taylor Series terms using the Richardson Extrapolation. The EKF, SPUKF, ESPUKF and the UKF are utilized in a re-entry vehicle navigation scenario using range and elevation measurements. The estimation accuracies and the processing times for different algorithms are compared for the scenario. The result demonstrates that the UKF provides better accuracy than the EKF but requires more processing time. The SPUKF accuracy is better than the EKF and the processing time is significantly less than the UKF. However, the accuracy of the SPUKF is less than the UKF. The ESPUKF provides estimation accuracy comparable to the UKF and the processing time is also significantly reduced.

Ruida Xie, Andrew G. Dempster

Deep learning-based techniques have been introduced into the field of trajectory optimization in recent years. Deep Neural Networks (DNNs) are trained and used as the surrogates of conventional optimization process. They can provide low thrust (LT) transfer cost estimation and enable more complex preliminary mission designs. However, it is a challenge to efficiently obtain the required amount of trajectory data for training. A Generative Adversarial Network (GAN) is adapted to generate the feasible LT trajectory data efficiently. The GAN consists of a generator and a discriminator, both of which are deep networks. The generator generates fake LT transfer features using random noise as input, while the discriminator distinguishes the generator's fake LT transfer features from real LT transfer features. The GAN is trained until the generator generates fake LT transfers that the discriminator cannot identify. This indicates the generator generates low thrust transfer features that have the same distribution as the real transfer features. The generated low thrust transfer data have a high convergence rate, and they can be used to efficiently produce training data for deep learning models. The proposed approach is validated by generating feasible LT transfers in a Near-Earth Asteroid (NEA) mission scenario. The convergence rate of GAN-generated samples is 84.3%.

Ruida Xie, Andrew G. Dempster

In recent years, deep learning techniques have been introduced into the field of trajectory optimization to improve convergence and speed. Training such models requires large trajectory datasets. However, the convergence of low thrust (LT) optimizations is unpredictable before the optimization process ends. For randomly initialized low thrust transfer data generation, most of the computation power will be wasted on optimizing infeasible low thrust transfers, which leads to an inefficient data generation process. This work proposes a deep neural network (DNN) classifier to accurately identify feasible LT transfer prior to the optimization process. The DNN-classifier achieves an overall accuracy of 97.9%, which has the best performance among the tested algorithms. The accurate low-thrust trajectory feasibility identification can avoid optimization on undesired samples, so that the majority of the optimized samples are LT trajectories that converge. This technique enables efficient dataset generation for different mission scenarios with different spacecraft configurations.

Ullah Ihsan, Ziqing Wang, Robert Malaney et al.

Location information claimed by devices will play an ever-increasing role in future wireless networks such as 5G, the Internet of Things (IoT). Against this background, the verification of such claimed location information will be an issue of growing importance. A formal information-theoretic Location Verification System (LVS) can address this issue to some extent, but such a system usually operates within the limits of idealistic assumptions on a-priori information on the proportion of genuine users in the field. In this work we address this critical limitation by using a Neural Network (NN) showing how such a NN based LVS is capable of efficiently functioning even when the proportion of genuine users is completely unknown a-priori. We demonstrate the improved performance of this new form of LVS based on Time of Arrival measurements from multiple verifying base stations within the context of vehicular networks, quantifying how our NN-LVS outperforms the stand-alone information-theoretic LVS in a range of anticipated real-world conditions. We also show the efficient performance for the NN-LVS when the users' signals have added Non-Line-of-Site (NLoS) bias in them. This new LVS can be applied to a range of location-centric applications within the domain of the IoT.

F. Boray Tek, Andrew G. Dempster, İzzet Kale

This paper presents an effective color normalization method for thin blood film images of peripheral blood specimens. Thin blood film images can easily be separated to foreground (cell) and background (plasma) parts. The color of the plasma region is used to estimate and reduce the differences arising from different illumination conditions. A second stage normalization based on the database-gray world algorithm transforms the color of the foreground objects to match a reference color character. The quantitative experiments demonstrate the effectiveness of the method and its advantages against two other general purpose color correction methods: simple gray world and Retinex.