Situational awareness for small maritime craft is of vital importance for improved marine safety of both the craft and other water users, as well as being paramount for path planning and obstacle avoidance in the development of small autonomous vessels. As part of a wider experimental and theoretical research program, here we investigate the potential for use of commercial electromagnetic simulation software, to simulate multi-modal radar responses and high-resolution imagery of maritime scenes at mm-wave and sub-THz frequencies, incorporating a developed physical sea surface model. The ultimate goal is to use simulation in the development of radar anomaly detection algorithms and for situations where experimental measurements are logistically challenging. We also outline the procedure for experimentally verifying the simulated data using a wave tank as a controlled repeatable measurement environment.

Reverse forward scatter radar is a novel concept where the target of interest is illuminated against a background reflector(s) and the echo signal is formed by the forward scattering field on the line between the background reflector, the target, and the receiver. This paper is the first to derive and analyse the target signature where a background clutter surface is used as a distributed reflector of opportunity. It is shown that, depending on the topology and target speed, the target signature has a power spectral density of the order of 10Hz which coincides with the signature bandwidth in traditional forward scatter radar. Simulations have been completed for a number of practical scenarios.

Use of sub-THz radar for automotive applications brings many potential game changing advantages, in terms of sensor size and imaging capabilities. We discuss the fundamentals governing these advantages and then focus on the potential to achieve increased azimuth resolution through processing techniques such as synthetic aperture radar and Doppler beam sharpening at sub-THz carriers, in comparison to current automotive frequencies. Examples of theoretical achievable resolutions are given for the automotive scenario and experimentation is carried out showing theoretical resolutions are achievable in ideal conditions. We highlight the factors that may affect the achievable resolutions and thus research areas that need to be further studied for sub-THz image formation and briefly discuss the current state of hardware development.

A universal image segmentation framework, which can be applied to various high-resolution automotive radar imagery produced by different beamforming strategies, is expected in the radar community to provide robust support to the development of autonomous driving. This paper estimates the universality of the segmentation framework, which is developed based on radar data produced by the mechanical steer beamforming, by directly implementing it onto another high-resolution radar imagery produced by the beamforming strategy of MIMO Doppler beam sharpening (DBS). The comparison of the distribution features of two parts of data shows that the return power level shift caused by the resolution difference is the major factor that needs to be compensated for the framework transfer implementation. The details of the universal segmentation framework are given to show that this can significantly simplify the complicated manual labelling and feature extraction. The segmentation results are discussed with the analysis of the performance and the potential future work.

Image segmentation on automotive radar imagery is the key technique for identifying the passable and impassable regions for path planning in autonomous or assistive driving. The availability of consecutive frames which measure the driving scene shifted along with the timeline enables improved segmentation on radar imagery. The frame fusion on automotive radar map is implemented as a two-step procedure: 1) The pixel-to-pixel mapping between consecutive frames is achieved based on an inertial measurement unit (IMU); 2) The information fusion of consecutive frames is achieved based on the Kalman filter. The frame fusion operation leads to correct classification of the initially “unknown” regions and overall improves the confidence of classification compared to single frame segmentation. The segmentation results with frame fusion are presented and compared with the results of single frame segmentation to demonstrate the segmentation improvement.

This paper shows an application of a technique for finding anomalies in an image. The technique is based on transforming the image in a way which reduces the 'normative' form into a compact region within the transform domain. This region can be eliminated and the transform inverted to leave only anomalous features. The technique is applied to range-time data from a millimetre-wave radar observing littoral sea clutter. The wavefronts are approximately straight in such images, so a Radon transform is used to compress straight lines into single points. Removing the brightest points in the transformed image leaves only the anomalies in the scene (i..e. features not corresponding to the waves on the sea).

Reverse forward scatter radar, an unusual alternative form to the traditional forward scatter radar, is investigated to demonstrate its viability and practicality. The power budget is illustrated graphically and derived by finding the maximum possible baselines for various targets, including a human. The two-ray path and the free space propagation model are considered. Reverse forward scatter radar is shown to have clear potential to be used in practice.

This paper presents a solution to the current challenges of the imaging radar to respond the demands of autonomy for detection and classification of targets in radar imagery, which traditionally has been considered as clutter. The proposed object detection method is defined in a new way, as opposed to the traditional object detection methods in the radar related contexts. The current paper presents the first application of this novel approach, based on deep neural networks for object detection, on outdoor radar images, as well as indoor images taken in controlled environment. Object detection was performed using two detectors, Faster R-CNN and SSD and the evaluation proved that this method can be successfully used on radar imagery for autonomous applications.

This paper discusses statistical approaches currently being investigated to perform image segmentation and region classification in high resolution automotive radar imagery in the complex urban environment. The purpose is to identify the free traversable space ahead of the vehicle which would ultimately provide input into autonomous vehicle path planning algorithms. Three general methodologies are described, all based on the image region pixel intensity statistics which vary depending on the features within the imaged scene. Results show the promising potential for segmentation with these methods and some examples of segmentation and identification have been demonstrated for chosen road scene region types of asphalt, grass, shadow and objects/other.

This paper demonstrates the ability of an active 300GHz radar system to perform 3D imaging non-coherently with as few as three receivers. The large bandwidth of the radar system facilitated full 3D reconstruction using time of flight information alone, applying single-look direct backprojection at a series of azimuth positions.

In this paper we apply the Doppler beam sharpening technique of azimuth refinement to the emerging area of low-THz radar. This improves the image quality and thus aids object classification in low-THz radar imaging systems, for example for autonomous platforms. The paper briefly explains the theory behind the process of Doppler beam sharpening; this is then experimentally tested and verified against two closely space corner reflector reference targets.

Knowledge on radar reflectivity of typical road targets is essential to develop robust detection algorithms for automotive sensor system. Measurement results of a typical passenger car at 300 GHz are presented for the first time and compared with measurement results at the reference frequency of 24 GHz. Measurements are undertaken in a typical road environment with the aid of a computer controlled turntable to collect backscatter data at 360 aspect angles of the car. The measurement methodology with an analysis of the experimental setup and calibration procedure are presented.

The applicability of Doppler beam sharpening (DBS) is assessed for the use in a passive bistatic system using a spaceborne transmitter in the Inmarsat constellation (Alphasat) as the illuminator of opportunity, for potential application in low cost maritime early warning systems in vessels. The effect of sharpening for both stationary and moving targets is discussed in theory, with models to show the effects for stationary targets that predicted a large improvement in angular resolution through the use of DBS. Experimental results show the effect in action, with three distinguishable targets found within 10 degrees using an antenna with a full beamwidth of 30 degrees, for a receiver moving at a mean of 11.6m/s over a coherent integration time of 1s, a carrier wave wavelength of 0.193m, a target range of 2.5-3km, and target look angle of 20-30 degrees. The results from the experiment are shown to agree with a simulation of a similar system, and a visual demonstration of the results overlaid on a satellite map is also displayed.

This paper presents an initial experimental investigation into multi-sensor imaging of a test scene through artificially generated fog. The sensor set consists of a high-resolution low-THz imaging radar, a lidar and a stereo optical camera. Images are obtained from each sensor in varying densities of fog and presented herein, along with a description and comparison of the findings. The reason for such a study is to highlight the requirement for inclusion of low-THz imaging radar as part of the next generation of automotive sensing, in order to provide terrain imaging in adverse weather conditions, when optical systems may fail to be effective.

In this paper we present a bistatic passive maritime surveillance system based on the use of signals from Inmarsat I-4 satellite Broadband Global Area Network (BGAN) band. An experimental set-up is presented with a stationary passive receiver designed for acquisition and processing of Inmarsat signals. A description of signal processing aspects is provided, including the clutter cancellation, range-Doppler cross correlation, and experimental analysis of the ambiguity function. For the first time the detection and bistatic range-Doppler target location by reflected Inmarsat signals are presented for a typical maritime target.

This paper focuses on the estimation of target's motion parameters in a moving transmitter/moving receiver forward scatter radar scenario. An extension of the quasicoherent processing initially developed for stationary FSR mode of operation is introduced. This offers an estimation of the target's trajectory in a general case where transmitter and receiver can be moving. Simulated and real data are used to analyse the performance of the presented processing.

In this paper we present a series of road scene images recorded with a prototype high resolution low-THz imaging radar. The aim of this study is to assess the images in order to identify and extract features of interest for scene characterisation to aid the development of future autonomous vehicles. Images were recorded at multiple radar heights in order to assess the effect of illumination angle on the selected features. At this stage of research the recorded images are presented and more qualitative analysis of a subset of the features is given.

This paper focuses on the estimation of target motion parameters in forward scatter radar. A new method based on the use of the Doppler signature spectrogram has been investigated. This allows a rough estimation of target motion parameters even in the presence of significant clutter, when the Doppler signature cannot be used to extract accurate kinematic information. The spectrogram characteristics change depending on target's speed, crossing angle and crossing point, this will be shown in order to explain how it is possible to estimate kinematic information analyzing the spectrogram trend. The performance of this new method is analyzed at the initial stage of the research.

The focus of this paper is on the estimation of the kinematic parameters of moving targets via a MIMO Forward Scatter Radar (FSR) system. A sub-optimum estimation technique is considered that exploits the information concerning the time instants at which the target crosses the individual baselines to retrieve the motion parameters. The accuracy of such technique is firstly investigated from a theoretical point of view and then the effectiveness of the proposed approach is demonstrated by applying it to live MIMO FSR data. Shown results prove the practical applicability of the proposed technique.

This paper concerns with the passive forward scatter radar (FSR). An algorithm to extract the Doppler signature of a target is presented for the first time. It allows removal of the background modulation of the transmitted signal, so that the signature can be extracted within a narrow band. Receiver architecture, corresponding to the proposed algorithm experimental set-up and experimental results are shown, where a DVB-T signal has been used to test the passive FSR algorithm. The performance of the proposed algorithm is analyzed at this initial stage of the research.