Nonstationary Target Research Articles

Nonstationary Target Detection Using Time–Frequency Multisqueezing Transform for DRM-Based Passive Radar

Nonstationary Target Detection Using Time–Frequency Multisqueezing Transform for DRM-Based Passive Radar

Cooperative Fusion Localization of a Nonstationary Target for Multiple UAVs Without GPS

Cooperative Fusion Localization of a Nonstationary Target for Multiple UAVs Without GPS

Symmetrical dual-sideband oppositely chirped differential FMCW LiDAR.

A differential FMCW LiDAR for high-precision distance measurements of remote non-stationary targets is proposed and demonstrated experimentally. The required positive and negative symmetrically oppositely chirped laser beams are generated synchronously through a fixed-frequency laser by employing externally unified broadband optical phase modulation and symmetrical dual-sideband optical filtering. After coaxial transmission and reception, orthogonally polarized optical beat signals containing target distance and vector velocity data are de-chirped separately by optical in-phase and quadrature demodulations and then synchronously received by four-channel photoelectric balance detectors. After differential processing of the received beat signals and a fast Fourier transform, it is possible to implement real-time simultaneous range and vector velocity measurements. The inherent symmetrically oppositely chirped optical frequency make it possible to measure the target distance immune to the internal random phase noise introduced by the spectral linewidth of the frequency-swept laser and the external random phase noise introduced by atmospheric turbulence, speckle, and vibration. Meanwhile, the measurement of the target velocity is immune to the nonlinearity of the frequency-swept laser. These results encourage an approach to overcome the barriers of coherence length, nonlinearity, and external noise, and implement simultaneous real-time ranging and velocimetry of long-range, rapid-moving targets.

Distributed Multicircular Circumnavigation Control for UAVs with Desired Angular Spacing

This paper addresses a multicircular circumnavigation control for UAVs with desired angular spacing around a nonstationary target. By defining a coordinated error relative to neighboring angular spacing, under the premise that target information is perfectly accessible by all nodes, a centralized circular enclosing control strategy is derived for multiple UAVs connected by an undirected graph to allow for formation behaviors concerning the moving target. Besides, to avoid the requirement of target’s states being accessible for each UAV, fixed-time distributed observers are introduced to acquire the state estimates in a fixed-time sense, and the upper boundary of settling time can be determined offline irrespective of initial properties, greatly releasing the burdensome communication traffic. Then, with the aid of fixed-time distributed observers, a distributed circular circumnavigation controller is derived to force all UAVs to collaboratively evolve along the preset circles while keeping a desired angular spacing. It is inferred from Lyapunov stability that all errors are demonstrated to be convergent. Simulations are offered to verify the utility of proposed protocol.

State-Following-Kernel-Based Online Reinforcement Learning Guidance Law Against Maneuvering Target

In this article, a state-following-kernel-based reinforcement learning method with an extended disturbance observer is proposed, whose application to a missile-target interception system is considered. First, the missile-target engagement is formulated as a vertical planar pursuit–evasion problem. The target maneuver is then estimated by an extended disturbance observer in real time, which leads to an infinite-horizon optimal regulation problem. Next, utilizing the local state approximation ability of state-following kernels, the critic neural network (NN) and actor NN for synchronous iteration are constructed to calculate the approximate optimal guidance policy. The states and NN weights are proven to be uniformly ultimately bounded using the Lyapunov method. Finally, numerical simulations against different types of nonstationary targets are effectively tested, and the results highlight the role of state-following kernels in the value function and policy approximation.

Long-Term Visitation Value for Deep Exploration in Sparse-Reward Reinforcement Learning

Reinforcement learning with sparse rewards is still an open challenge. Classic methods rely on getting feedback via extrinsic rewards to train the agent, and in situations where this occurs very rarely the agent learns slowly or cannot learn at all. Similarly, if the agent receives also rewards that create suboptimal modes of the objective function, it will likely prematurely stop exploring. More recent methods add auxiliary intrinsic rewards to encourage exploration. However, auxiliary rewards lead to a non-stationary target for the Q-function. In this paper, we present a novel approach that (1) plans exploration actions far into the future by using a long-term visitation count, and (2) decouples exploration and exploitation by learning a separate function assessing the exploration value of the actions. Contrary to existing methods that use models of reward and dynamics, our approach is off-policy and model-free. We further propose new tabular environments for benchmarking exploration in reinforcement learning. Empirical results on classic and novel benchmarks show that the proposed approach outperforms existing methods in environments with sparse rewards, especially in the presence of rewards that create suboptimal modes of the objective function. Results also suggest that our approach scales gracefully with the size of the environment.

Continuous Adaptation with Online Meta-Learning for Non-Stationary Target Regression Tasks

Most environments change over time. Being able to adapt to such non-stationary environments is vital for real-world applications of many machine learning algorithms. In this work, we propose CORAL, a computationally efficient regression algorithm capable of adapting to a non-stationary target. CORAL is based on Bayesian linear regression with a sliding window and offline/online meta-learning. The sliding window makes our model focus on the recently received data and ignores older observations. The meta-learning approach allows us to learn the prior distribution of the model parameters. It speeds up the model adaptation, complements the sliding window’s drawback, and enhances the performance. We evaluate CORAL on two tasks: a toy problem and a more complex blood glucose level prediction task. Our approach improves the prediction accuracy for the non-stationary target significantly while also performing well for the stationary target. We show that the two components of our method work in a complementary fashion to achieve this.

UAV-based autonomous detection and tracking of beyond visual range (BVR) non-stationary targets using deep learning

UAV-based autonomous detection and tracking of beyond visual range (BVR) non-stationary targets using deep learning

An online data driven actor-critic-disturbance guidance law for missile-target interception with input constraints

In this article, we develop an online robust actor-critic-disturbance guidance law for a missile-target interception system with limited normal acceleration capability. Firstly, the missile-target engagement is formulated as a zero-sum pursuit-evasion game problem. The key is to seek the saddle point solution of the Hamilton Jacobi Isaacs (HJI) equation, which is generally intractable due to the nonlinearity of the problem. Then, based on the universal approximation capability of Neural Networks (NNs), we construct the critic NN, the actor NN and the disturbance NN, respectively. The Bellman error is adjusted by the normalized-least square method. The proposed scheme is proved to be Uniformly Ultimately Bounded (UUB) stable by Lyapunov method. Finally, the effectiveness and robustness of the developed method are illustrated through numerical simulations against different types of non-stationary targets and initial conditions.

Lifelong Personalization via Gaussian Process Modeling for Long-Term HRI.

Across a wide variety of domains, artificial agents that can adapt and personalize to users have potential to improve and transform how social services are provided. Because of the need for personalized interaction data to drive this process, long-term (or longitudinal) interactions between users and agents, which unfold over a series of distinct interaction sessions, have attracted substantial research interest. In recognition of the expanded scope and structure of a long-term interaction, researchers are also adjusting the personalization models and algorithms used, orienting toward “continual learning” methods, which do not assume a stationary modeling target and explicitly account for the temporal context of training data. In parallel, researchers have also studied the effect of “multitask personalization,” an approach in which an agent interacts with users over multiple different tasks contexts throughout the course of a long-term interaction and learns personalized models of a user that are transferrable across these tasks. In this paper, we unite these two paradigms under the framework of “Lifelong Personalization,” analyzing the effect of multitask personalization applied to dynamic, non-stationary targets. We extend the multi-task personalization approach to the more complex and realistic scenario of modeling dynamic learners over time, focusing in particular on interactive scenarios in which the modeling agent plays an active role in teaching the student whose knowledge the agent is simultaneously attempting to model. Inspired by the way in which agents use active learning to select new training data based on domain context, we augment a Gaussian Process-based multitask personalization model with a mechanism to actively and continually manage its own training data, allowing a modeling agent to remove or reduce the weight of observed data from its training set, based on interactive context cues. We evaluate this method in a series of simulation experiments comparing different approaches to continual and multitask learning on simulated student data. We expect this method to substantially improve learning in Gaussian Process models in dynamic domains, establishing Gaussian Processes as another flexible modeling tool for Long-term Human-Robot Interaction (HRI) Studies.

Improved CFAR Detection and Direction Finding on Time–Frequency Plane With High-Frequency Radar

In addition to sea state monitoring, high-frequency surface wave radar (HFSWR) has attracted more and more attention in ship detection and tracking. Constant false alarm rate (CFAR) detectors have been widely used to handle the complex properties of sea clutters. Due to the relatively high detection threshold, the performance of CFAR in detecting weak and nonstationary targets is often not good. To address this problem, time–frequency analysis (TFA) is involved in the ship detection and direction finding (DF) processing. First, the probability distribution model of sea clutter is achieved in the time–frequency (TF) domain. Corresponding to this sea clutter model, the decision thresholds of targets on the TF plane under different false alarm rates <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$P_{\mathrm {fa}}$ </tex-math></inline-formula> are calculated. Next, the array snapshots are formed by the spectral samples along each extracted TF ridge and later used in the DF process to give the direction of arrival (DOA). Experimental results show that, with the automatic identification system (AIS) records as the ground truth, the number of matched targets detected by the proposed TF-CFAR method is 5%–8% greater than that by the cell averaging (CA)-CFAR method. Moreover, the TF multiple-signal classification (MUSIC) also outperforms MUSIC with an improvement of 3.52° in the root-mean-square error (RMSE) of the DOA estimates under a low signal-to-noise ratio (SNR). In conclusion, the involvement of TFA can greatly improve the detection and DF performances of compact HF radar, particularly under the situations of low SNR and target nonstationarity.

A Maximum Likelihood Method for Joint DOA and Polarization Estimation Based on Manifold Separation

The use of the polarization diversity of a target signal at a polarization-sensitive antenna array can enhance the target detection and tracking capabilities of a radar. In this article, the manifold separation steering vector modeling technique is used to develop a maximum likelihood method for joint direction of arrival (DOA) and polarization estimation. Manifold separation can incorporate antenna array nonideal characteristics (e.g., cross polarization, mutual coupling) into the estimation algorithm using array calibration measurements. In the proposed technique, the estimation problem is formulated as a generalized Rayleigh quotient minimization problem that is transformed into a determinant minimization problem. Both the azimuth and elevation angles are estimated using the fast Fourier transform. Unlike the existing manifold separation based polarimetric element space (PES) multiple signal classification method and the PES Capon method, the proposed method can obtain DOA and polarization estimates based on very small-size primary data samples, even with a single sample, which makes the proposed method more suitable for nonstationary target polarization. The performance of the proposed method is demonstrated through simulations. The Cramer-Rao lower bound for joint DOA and polarization is also used for comparison with empirical errors.

Dual Frequency Electronically Controlled Radiation Beam Reconfigurable slotted Antenna for Detection of a Stationary or Nonstationary Target

A dual-frequency and radiation pattern reconfigurable microstrip patch antenna for detecting a stationary as well as a non-stationary target is described. Six angular patches, that collectively form a circular shape, are used. All the six patches radiate one by one after a fixed interval of time and their feed controlling is done by six PIN diodes. The switching of PIN diodes is controlled by an embedded biasing network. This antenna provides radiation beam scanning characteristics. It gives the main lobe scanning at every 60o clockwise (or anticlockwise) continuously by applying a signal to patches one by one. The purpose of introducing the slot is to get the radiation pattern in the desired direction since by changing the length, width, and position of the slot, the direction of the radiation pattern can be controlled. The slotted antenna operates in a C band with two frequencies 4.21 GHz and 4.82 GHz and provides a radiation pattern, 90o apart from each other. The scanning rate of 0.6 deg/ms is obtained; however, the scanning rate can be changed with the help of ATMEGA 2560 microcontroller. This compact Microstrip patch antenna can be widely used for short-range applications i.e. ground surveillance radar, missile control, mobile battlefield surveillance for military and many other applications in a modern wireless communication system. The designed antenna along with the switching application will be able to track the stationary as well as a non-stationary target.

Pedestrian localisation in the typical indoor environments

The world is adopting complete wireless infrastructure to cover small size buildings as well as large geographical regions. Indoor Positioning System (IPS) is used to locate stationary as well as non-stationary objects in the wireless domain. The indoor localisation of a pedestrian remains an open problem in the noisy environment. Researchers are trying to develop low-complexity approach without depending on building infrastructure to achieve accurate and reliable results for pedestrian localisation in noisy environment. We are proposed the problem associated with improving localisation scalability and accuracy by considering the noisy environment. Our propose methodology exhibits robustness and portability with respect to the number of experiments in noisy environment with the help of captured signal strength. The propose methodology is easily applied in various indoor environments (i.e. different building designs) to locate the stationary and non-stationary object. The collected multimodal data of non-stationary targets in the noisy environment is required to enhance further. The collected multimodal data is used to explore specific movement of the pedestrian in the noisy environment. We consider numerical methods to make the shape of hot-spot contour for a specific target. Principal Component Analysis (PCA) based data science is used to obtain the predominant components in the collected information and make a compact contour in the noisy environment. The generated model depicts a highly sensitive region in the noisy environment due to the presence of thick walls inside the building and multipath fading. Our novel methodology exhibits efficient localisation of the non-stationary target or precisely identifies the moving object, on the floor of a multi- storey building. Our novel approach enhances the technique to improve surveillance and security for the noisy indoor environment.

Circumnavigation on Multiple Circles Around a Nonstationary Target With Desired Angular Spacing

In this article, the objective for a group of nonholonomic agents is to achieve multicircular circumnavigation with any desired angular spacing around a nonstationary target. A cooperative protocol is proposed to achieve this objective and, additionally, generalizes the circumnavigation problem of unicycle vehicles around a target (moving and stationary). Due to the nonholonomic constraints, existing protocols cannot be extended directly to achieve this objective. Thus, the proposed algorithm is worked out from the desired geometry to achieve the objective. A fixed-time consensus estimator is designed under a cyclic digraph to arrive at the desired interagent angular separation in the target-centric frame. The target information and desired formation parameters are assumed to be known to only one agent partially. Due to uncertainty in the target's motion, it is assumed that the target's acceleration and angular velocity are unknown. Fixed-time estimators under a digraph address the lack of information. The tracking controller drives the agents to the desired position around the target, thus making the errors go to zero instead of any finite bounds. The controller guarantees the bounded control effort irrespective of the error magnitude. The numerical examples are presented to show the effectiveness of the algorithm.

All-Aspect Guidance With Impact Angle Constraint Against Unknown Target Maneuver

This paper presents an all-aspect guidance law that allows the missile to achieve the interception of a target at the prescribed impact angle for any initial conditions. The proposed guidance law is designed to intercept a maneuvering target without requiring information of the normal acceleration of the target. In order to develop a guidance law that ensures all-aspect capability, the kinematic conditions that always guarantee the interception of the maneuvering target at the desired impact angle are defined. The defined kinematic conditions are expressed as two error variables for simultaneously satisfying the homing and impact angle control, and the guidance law is derived by stabilizing both error variables near zero based on the backstepping control technique. The analysis of the overall closed-loop dynamics verifies that the proposed law achieves the interception of the maneuvering target at the prescribed impact angle from any initial heading and boresight conditions, despite the absence of the information of the target acceleration. Numerical simulations also demonstrate that the proposed law achieves an acceptable miss distance at the desired impact angle for various initial conditions against different types of nonstationary targets: nonmaneuvering targets, constant-maneuvering targets, and weaving targets.

Impact of Polarization Distortions on Geometrical Structure Retrieval of Moving Man-Made Targets in ISAR Images

In this paper, the impact of several polarization distortions on geometrical structure retrieval of man-made targets in inverse synthetic aperture radar (ISAR) images is assessed. For polarimetric ISAR, the polarization distortions are usually composed of crosstalk and channel imbalance. For non-stationary targets, a residual phase error exists between two columns of the scattering matrix in the alternate transmission and simultaneous reception (ATSR) mode, especially when the radar frequency is high. These two kinds of distortions have varying influences on different targets, as well as different decomposition methods. Some theoretical derivations and numerical analyses are given to reveal the impact of these distortions on Cameron decomposition. To evaluate the impact of these distortions on geometrical structure retrieval, we use the numerical results and real data of an unmanned aerial vehicle (UAV) called Frontier to perform error analysis of both the decomposition and the retrieved size of the designated structures of the UAV. Some numerical evaluations were conducted from the perspective of system design, and these results can be useful for the practical radar system.

Sliding Mode Guidance Law for Impact Time Control Without Explicit Time-to-Go Estimation

This paper presents an impact time control guidance law that does not perform explicit time-to-go estimation. In order to satisfy the interception and the desired impact time constraint simultaneously, a sliding surface variable formulated as a sum of the relative range and the desired time-to-go is defined, weighted by two nonzero weighting functions. Then, the achievement of the sliding mode satisfies the following equivalence: The relative range is zero if and only if the elapsed time equals to the desired impact time. It means that both the interception and impact time control can be satisfied at the same time in the sliding mode. The impact time control guidance law is derived to enforce the defined surface variable to the sliding mode. Because the law is designed based on the capture condition without separate time-to-go estimation process, the achievement of the sliding mode always guarantees the interception of the target at the desired impact time. In addition, the proposed law can be applied to an engagement considering nonstationary targets in a straightforward manner because the corresponding time-to-go estimation is not needed. Simulation results demonstrate that the proposed guidance law has better performance in comparison with the existing guidance law, in respect of satisfying the desired impact time constraint against a nonstationary target.

Non-Stationary Platform Inverse Synthetic Aperture Radar Maneuvering Target Imaging Based on Phase Retrieval.

As a powerful signal processing tool for imaging moving targets, placing radar on a non-stationary platform (such as an aerostat) is a future direction of Inverse Synthetic Aperture Radar (ISAR) systems. However, more phase errors are introduced into the received signal due to the instability of the radar platform, making it difficult for popular algorithms to accurately perform motion compensation, which leads to severe effects in the resultant ISAR images. Moreover, maneuvering targets may have complex motion whose motion parameters are unknown to radar systems. To overcome the issue of non-stationary platform ISAR autofocus imaging, a high-resolution imaging method based on the phase retrieval principle is proposed in this paper. Firstly, based on the spatial geometric and echo models of the ISAR maneuvering target, we can deduce that the radial motion of the radar platform or the vibration does not affect the modulus of the ISAR echo signal, which provides a theoretical basis for the phase recovery theory for the ISAR imaging. Then, we propose an oversampling smoothness (OSS) phase retrieval algorithm with prior information, namely, the phase of the blurred image obtained by the classical imaging algorithm replaces the initial random phase in the original OSS algorithm. In addition, the size of the support domain of the OSS algorithm is set with respect to the blurred target image. Experimental simulation shows that compared with classical imaging methods, the proposed method can obtain the resultant motion-compensated ISAR image without estimating the radar platform and maneuvering target motion parameters, wherein the fictitious target is perfectly focused.

Reconstruction of Wideband Radar Signal Based on Compressed Sensing

The matched filter is often used for the radar receiving terminal, but such application in wideband radar may cause complex system and high cost. The adoption of compressed sensing can not only remove the matched filter and the high-speed analogdigital converter, but also realize high-resolution radar imaging. In consideration of the sparseness of the radar target scene, a compressed sensing method used for chaotic radar is proposed in this article, wherein the chaotic frequency-modulated signal is generated by Bernoulli mapping and the randomness of such signal can be directly used for constructing the observation matrix. Meanwhile, the under-sampled echo signal can be used for radar scene reconstruction through the optimization technology. According to the simulation of stationary and non-stationary target radar scenes, compared with the matched filter processing technology, the proposed method has simple processing procedure and optimal performance, and the feasibility of this method is also verified by the simulation result.