Radar sensors are an emerging technology in the context of non-contact monitoring of vulnerable individuals. Radar-based solutions ensure end-user privacy, whilst providing medical professionals and caregivers with key information concerning the subject's well-being. This thesis proposes novel methods for the classification of sequential human activities using a network of radar sensors. Accurate classification of Activities of Daily Life (ADL) can enable for instance the detection of falls and wandering amongst elderly individuals, and can be employed for the recognition of aggressive or otherwise anomalous behaviour for those receiving mental health care.@en

Radar-based sensors are used to perceive their environment and objects of interest in a contactless manner and with robust performance in all weather and light conditions. One of the main drawbacks is the energy needed for the processing of radar data in order to extract its valuable information. Spiking neural networks are an emerging type of neural networks that aim to reduce the energy footprint of their computations while maintaining acceptable performance. To do so, the data is encoded through time in binary spikes to help leverage the low cost of additions. This is in stark opposition to the much higher cost of multiplications that are highly present in conventional artificial neural networks. The drawback of this energy gain is that the rate encoding adds an extra time dimension, hence increasing the latency between the acquisition of the radar data and the recognition of the corresponding gesture class.
More specifically, this work uses an air-marshalling dataset from the literature to exemplify a gesture recognition problem. The first step is to replicate the well-known radar processing pipeline, and classification approach based on conventional neural networks to reach high classification accuracies. A validation accuracy of 98.5% and a test accuracy of 59.8% are reached on the full dataset (11 classes) and 86.7% on their 5 best classes (test set), which is about the same performance reported in the original dataset baseline.
The following steps propose an adaptation of this non-spiking pipeline to its spiking equivalent by optimising the trade-off between the model’s latency, its memory requirements and its accuracy. This work also develops a strategy to tune spiking networks’ thresholds to make the process of developing a spiking equivalent more efficient. For example, the spiking network can reach 94.1% validation accuracy using 100 encoding steps and only 4.7% of the initial memory requirements, and reach 46.8% on the test set. However, this trade-off can be shifted towards lower latency, lower memory, or higher accuracy according to the desired requirements.

Extreme precipitation, like floods and landslides, poses major risks to safety and the economy, underscoring the need for sophisticated weather forecasting to predict these events accurately, enhancing readiness and resilience. Nowcasting, which uses real-time atmospheric data to predict short-term weather, is key in addressing this challenge. Traditional nowcasting systems, reliant on extrapolation from rainfall radar observations and constrained by simplistic physical assumptions, often struggle to detect complex, nonlinear weather patterns. This gap has opened the door for deep learning models, which have shown significant promise in improving the accuracy and reliability of short-term weather predictions, making them a focal point of recent research and the basis of this thesis's approach.

This thesis introduces a deep generative model designed for the nowcasting of extreme precipitation events up to 3 hours ahead, utilizing a Vector-Quantized Variational Autoencoder (VQ-VAE) to compress radar data into a low-dimensional latent representation, and an Autoregressive Transformer for predicting future radar images. Additionally, a binary classifier works in conjunction with the Autoregressive Transformer to identify extreme versus non-extreme weather events, using these classifications to inform an Extreme Value Loss (EVL) function. This loss function aims to improve the accuracy of predicting extreme weather events by addressing the data imbalance between normal and extreme precipitation occurrences. The proposed model displays comparable performance with the state-of-the-art conventional methods and other deep learning nowcasting models in predicting extreme events.

Traditional target tracking using monostatic radar systems typically relies on centralized or decentralized architectures, where all data is transmitted to a fusion center for processing the position and velocity of mobile targets. This approach not only introduces a single point of failure and can lead to increased data transmission times, particularly when the fusion center is distant from individual radar nodes, but it also faces scalability issues and potential bottlenecks when data accumulates at the fusion center. To address these challenges, we introduce a Distributed Alternating Direction Method of Multipliers (DADMM) for target localization within a radar network, wherein each radar node shares its observed data only with immediate neighboring nodes, achieving consensus on the estimated target locations and velocities. Our simulations, which incorporate critical parameters such as the number of radar nodes, radar geometry, and Signal-to-Noise Ratio (SNR), assess their impact on estimation accuracy and convergence speed. The results demonstrate that the proposed DADMM not only effectively eliminates the single point of failure, but also enhances system efficiency and robustness. We also incorporate two distinct stopping criteria for position and velocity estimations, enabling us to promptly fix the first accurately estimated parameter and reallocate computational resources to more effectively refine the remaining parameters, which streamlines computational efforts by focusing on unresolved parameters. We highlight the additional benefits of our proposed framework, and present directions for future work.

4D millimeter-wave radar is increasingly important in advanced driver-assistance systems due to its ability to capture Doppler/velocity information and robustness in low-light or adverse weather conditions. Unlike traditional 3D radar, 4D radar provides elevation information, enhancing 3D spatial perception. Most of the perception tasks using 4D radar tensors focus on target classification using bounding boxes. In contrast, although semantic segmentation is typically used to process images and LiDAR point clouds, it has not been well explored for 4D radar tensors.

This MSc thesis aims to bridge the gap in using 4D radar tensors for semantic segmentation and in generating the required labels for supervision. Specifically, it proposes an automatic approach for generating multi-class point-wise labels for automotive datasets by leveraging the complementary information from the synchronized camera and LiDAR data. Then, 4D radar tensors are used as inputs, supervised by the generated labels, to design a radar semantic segmentation network. Promising results are shown by applying both developed parts to the publicly shared RaDelft dataset. The automatic labeling process demonstrates satisfactory quantitative and qualitative results compared with manual labeling results obtained on randomly chosen scenes. The outputs of the semantic segmentation network achieve more than 65% in overall detection probability, improving by +13.1% in terms of vehicle class detection probability, and reducing 0.54 m in terms of Chamfer distance compared to the variants inspired by the literature.

This thesis focuses on advancing radar technology to meet the growing demands of autonomous driving, particularly regarding angular resolution and target detection. The work acknowledges the critical role of automotive radar in achieving higher levels of vehicle autonomy, where reliable detection, classification, and tracking of objects like pedestrians, vehicles, and infrastructure are critical. The primary focus of the dissertation is to explore novel techniques for enhancing radar system performance through machine learning (ML) and compressive sensing (CS) approaches, addressing the challenges faced by current radar technologies, such as low angular resolution and inefficient target detection in dynamic environments.

The dissertation begins by outlining the challenges in automotive radar systems, especially the need for improved angular resolution without increasing the physical size and complexity of the radar devices. The importance of angular resolution is emphasized for both azimuth and elevation, as modern vehicles must be able to discriminate between various objects, such as distinguishing between two vehicles at a similar distance, or identifying an object’s height to determine whether it can be driven under or must be avoided. Current methods for improving angular resolution, such as increasing the number of transmitters and receivers in multiple input multiple output (MIMO) radars, are costly and increase system complexity, thus requiring novel solutions to meet industry needs. Then, Chapter 2 briefly summarizes the theoretical background of MIMO radars and defines the terminology used in the rest of the dissertation. This is crucial since automotive radar is a multi-disciplinary topic with people from different backgrounds interacting, and often, the same concepts are named differently.

The first research chapter, Chapter 3, introduces a self-supervised learning framework designed to enhance the angular resolution of radar systems without the need for additional physical hardware. A neural network (NN) artificially expands the radar’s aperture by predicting the response of additional antenna elements based on data from radars with larger arrays. This approach leverages the correlation between antenna elements to generate a more detailed angular profile from a smaller, low-resolution radar, allowing for a more accurate estimation of incoming signals' direction of arrival (DoA). Extensive simulations and experimental results demonstrate that this method significantly enhances radar performance in separating closely spaced objects, which is critical in automotive scenarios.

Another key contribution of the dissertation is presented in Chapter 4, with the application of Bayesian compressive sensing (BCS) to automotive radar, which exploits the sparse nature of the data in the DoA domain. The BCS approach uses probabilistic models to estimate the DoA while also providing uncertainty measures, offering both accuracy and reliability in angular estimation. The research further explores how array topologies can be optimized for BCS-based DoA estimation, demonstrating that a carefully designed antenna array can achieve better performance with fewer elements, thus reducing system costs. Additionally, the work presents a computationally efficient BCS algorithm that dramatically reduces the time needed for DoA estimation without compromising accuracy. This is a crucial advancement for real-time applications, where fast processing is required for decision-making in autonomous driving.

In addition to BCS, in Chapter 5, total variation compressive sensing (TVCS) is applied to the problem of radar imaging. TVCS enforces sparsity in the gradient of the signal rather than in the signal itself, which proves particularly effective in estimating the shape of extended targets, such as vehicles or pedestrians. By applying TVCS to 2D and 3D radar data, the dissertation demonstrates that this method can reconstruct objects' shapes more accurately than traditional methods, thereby enhancing the radar's ability to classify and understand the surrounding environment. The application of TVCS marks a significant step forward in radar-based shape estimation, especially for imaging radars used in automotive systems.

The dissertation also addresses the limitations of conventional target detection methods, particularly the widely used window-based constant false alarm rate (CFAR) detectors. Window-based CFAR detectors struggle with dynamic and unpredictable environments, which are common in road traffic scenarios. Moreover, they are unsuitable for extended targets with very different sizes, such as the ones encountered in automotive radar. To overcome this, in Chapter 6, a deep learning-based detector is proposed, trained using a newly developed dataset, RaDelft, which includes synchronized radar and lidar data. This deep learning detector outperforms traditional CFAR detectors by significantly improving the probability of detection and the Chamfer distance, especially in complex and cluttered environments. The RaDelft dataset itself is another important contribution of the dissertation, providing the research community with a well-curated, large-scale, multi-sensor dataset for further exploration and development of radar-based detection and classification systems.

In conclusion, this dissertation presents a comprehensive study of methods to enhance the angular resolution, detection capabilities, and efficiency of automotive radar systems through a combination of machine learning and compressive sensing. It provides practical solutions verified with experimental data to overcome existing limitations in automotive radar technology, particularly in the areas of angular resolution, target detection, and data processing speed. These advancements contribute to the broader goal of achieving fully autonomous driving by improving the ability of radar systems to perceive and interpret complex environments.@en

Autonomous driving is one of the most popular research topics. Radar technology is used for many applications of ADAS and is considered one of the key technologies for HAD. It has unique advantages compared with other sensors, especially its capabilities during adverse weather conditions and Doppler information extraction. Required by autonomous applications, radar has to change its historical role from a simple detector to an imaging sensor, which requires not only the range and Doppler resolution ability but also a high spatial resolution, i.e., the azimuth and the elevation angle resolution. To address this problem, in this thesis, new signal processing algorithms are proposed, which pave the way to improved performance of the automotive radar sensor.
The FMCW waveform is widely used in current automotive applications due to its low cost and simplicity. MIMO array techniques exploit the spatial diversity of transmit and receive antenna arrays and have been exploited in current automotive radar because of their ability to achieve high angular resolution with a few antennas. Platform movement is one of themain characteristics of automotive radar, which introduces movement uncertainty compared with radars at fixed locations but provides an opportunity to use the movement to boost the angular resolution. Thus, FMCW waveform and MIMO antenna array are the main research subjects in this thesis.....@en

With the aging population, the demand for healthcare and related services is increasing and, for this reason, technologies for remote patient monitoring are developing, aiming at indoor scenarios. Remote patient monitoring can help capture the clinical data of patients at home, which can save time and money, specifically reducing the need for hospitalization by potentially detecting health-related issues before they become too serious.
The non-contact radar-based technology can be applied in the remote patient monitoring system for detecting vital signs. Radars are suitable for applications at home because they are non-invasive, robust in changing lighting and temperature, and suitable for patients with skin irritation.
Heartbeat and respiration are critical clinical data for the diagnosis of the disease. The study of respiration frequency estimation was explored by previous work, such as the MSc thesis in \cite{Maxthesis}. Building on that work, this project proposes a pipeline to measure the heartbeat frequency and cancel the random body movement. The impact of different orientations is also studied. The phase history difference of the chest displacement due to vital signs is extracted, and the wavelet transform is used to separate heartbeat and respiration signals. Different methods are tested to calculate the heartbeat frequency in the time and frequency domain. The RBM is detected by the energy threshold of the phase difference, and the intervals with the RBM are discarded.
The simulation and experimental results indicate that the proposed processing pipeline can work on the radar data.

According to the World Health Organization, approximately 15 million neonates are born prematurely each year, which is more than 1 in 10 newborns. Premature birth disrupts the normal development of the neonate in the womb, which can result in various complications and health issues, necessitating continuous monitoring of vital signs to assess their health status.

Currently, the predominant monitoring approach involves the use of contact-wired sensors, which offer precise readings but are accompanied by specific drawbacks. One notable limitation is the utilization of strong adhesives to attach the sensor to the neonate's skin, causing pain and stress during removal. Additionally, the wired sensor hinders skin-to-skin contact care, which is proved to be crucial for the neonate's growth and development.

To address these drawbacks, a non-contact wireless approach using FMCW radar is proposed in this thesis project. The main aim is to develop a processing pipeline to monitor the vital signs of neonates with the radar. The core concept involves estimating the vital signs from the phase information corresponding to the chest movement of the monitoring target. This will be achieved through the implementation of an algorithm based on the short-time Fourier transform. Furthermore, this study also explores an approach to detect body movement, select near-steady state data segment and determine the range bin that is most relevant for the vital signs estimation.

The proposed pipeline aims to provide a more comfortable and efficient monitoring system for neonates, potentially overcoming the limitations of conventional contact-wired sensors and promoting better care for their overall well-being and development.

Autonomous driving can bring revolutionary advancements in transportation, by making vehicles navigate and operate independently without human intervention. To achieve this, autonomous vehicles are equipped with sensors, including cameras, Light Detection and Ranging (LiDAR), and radar, which provide them with a comprehensive view of their surroundings. Frequency-Modulated Continuous Wave (FMCW) radar is the most popular radar sensor employed on autonomous vehicles (AVs), for its long working distance, simultaneous accurate measurements of range and radial velocity of the target, and its robustness for all weather conditions. However, there are still many problems to be solved for FMCW radar for detecting vulnerable road users (VRUs), such as pedestrians. The Constant False Alarm Rate (CFAR) detector plays a crucial role in FMCW radar signal processing, for its adaptive capability to estimate multiple potential targets versus variable clutter and noise backgrounds, and it is often the first step of processing in many automotive radar pipelines.
There have been several published methods for detecting pedestrians. However, most methods hardly consider public real-world radar datasets for their performance evaluation. Meanwhile, conventional CFAR used in autonomous driving, such as Cell averaging CFAR (CA-CFAR) and Ordered-statistic CFAR (OS-CFAR) are not specially designed for Doppler-spread targets (DSTs), while pedestrians are typical DSTs which shows Doppler extension in range-Doppler-map (RDM). This is due to the movement of the arms and legs of pedestrians that make them not only extended targets at mm-wave frequencies but also present a spread Doppler signature.
Therefore, in this thesis a proposed CFAR detector for DSTs enhances the probability of detection compared to two-dimensional (2D) CA-CFAR and two-dimensional (2D) OS-CFAR. The parametric study is conducted on CA-CFAR and OS-CFAR detectors. Additionally, the computation time is reduced dramatically.

Radar micro-Doppler signatures are powerful indicators of target movements and activities, enabling the extraction of valuable information about various objects' internal and external dynamics. Consequently, classifying these signatures has become crucial in numerous applications, ranging from target recognition in surveillance, to biomedical sensing and interaction with smart sensors.
In this thesis, an evaluation of classification performances for a wide variety of orthogonal moments, when applied to micro-Doppler classification problems, is presented. A pipeline is proposed to evaluate all moments commonly used in image processing, but not routinely employed in radar-based classification.
The evaluation results are compared with other state-of-the-art classification approaches, such as using micro-Doppler signatures directly as the input of Convolutional Neural Networks. The influence of noise in the data on the classification performance is also shown.
The classification results demonstrate the different moments' capabilities with a variety of publicly available datasets containing human micro-Doppler signatures, resulting in a very well performing classification pipeline for this type of classification problem, and novel insights into the potential of these moments for radar classification problems.

Small UAVs and in particular the class of micro-UAVs, whose mass is below 2 kg, are constantly rising in popularity for personal as well as professional use, since they are beneficial in many fields such as defense, transportation, monitoring and agriculture. In spite of their advantages, UAVs can be used for terrorist attacks to fly over restricted areas, transport illegal materials or cause an accident in crowded areas. Thus, in the event of non-cooperative drone users, radars can be ideal detection devices, due to their capability to operate day and night in all-weather conditions. However, due to birds and UAVs having similar altitude, speed and radar cross section, many false alarms can occur. For this reason, it is urgent to develop techniques to distinguish UAVs and birds amongst other radar contacts that may be present.

The classification of drones and birds is not unprecedented. Plenty of acoustic, camera and radar solutions are available in the market today. Radar applications frequently rely specifically on the differences in micro-Doppler signature between these two classes. However, limited solutions exist based on the differences in flight behavior, which is referred as "track behavior" in the radar domain. Assuming that drones and birds have different kinematic characteristics, the question arises, whether these can be tracked by surveillance radars in such a way that radar processing can recognize either or both, based on a given tracking time interval. In particular, possible methods include the exploration of the flights of drones and birds, the kinematic features extraction from tracking their trajectories, their implementation to machine learning models in order to classify these two flying objects, as well as the evaluation of their classification accuracy.

The methodology of this thesis project is divided into two classification cases. First, a 6-Degree-of-Freedom quadcopter simulator is used to generate drone trajectories, while the bird data are created from real bird GPS tracks. The second part of the work classifies real bird and drone trajectories tracked by a man-portable scanning surveillance radar. The current research can provide valuable information for the development of classification algorithms for existing radars, and fill the gaps for future efforts to improve the classification of birds and drones.

Today, with increasingly aging population, healthcare systems in many countries need to improve their effectiveness, and the automatic HAR technology can be beneficial. This can provide early diagnosis of changes in behavioral patterns in the home environment, without hospitalization, and detect critical events such as falls in a timely manner. In this area, radar-based HAR solutions are attracting the researchers' attention because no optical images are captured by radars, and thus respect of privacy and functionality in darkness can be guaranteed. Furthermore, no sensors need to be worn by the person being monitored.
Most previous work related to radar-based HAR employs image-like data representation such as spectrograms, range profiles, and snapshots of point clouds, and the information contained in these data representation is limited. For instance, spectrograms and range profiles cannot reflect the body shape of the subjects, while snapshots of point cloud do not contain Doppler or intensity information.
To overcome the limitation of these data representations, we propose to utilize the data from a mm-wave FMCW MIMO radar to create a novel data representation of point cloud with Doppler and intensity values, plus temporal information to achieve accurate HAR.
Specifically, this thesis work focuses on the high dimensional radar point clouds and on a pipeline to generate and process this novel data representation. The proposed method combines the spatial information of point clouds with other features like Doppler, intensity/SNR, and time, expanding each point from 3D coordinates to a 6D vector.
Hence, the movement of every part of the body can be expressed by those points. A module consisting of adaptive noise cancelation, frame selection, and resampling is proposed to process point clouds to match the input of subsequent classifiers.
Considering that the core of the self-attention concept matches well the information within point clouds, we investigate three self-attention based models as classifiers. These models can learn the spatial distribution of point clouds with their extra features with self-attention mechanism. The best combination of different input features, the positive contribution of the proposed adaptive noise cancelation method, and the performance of these three models are studied with experimental data from the MMActivity dataset and a purposely collected TU Delft (TUD) dataset.

For the development of automatic People Counting systems, radar is increasingly becoming a popular technology because of the increasingly stringent privacy requirements for people demographic information and the requirement to operate in a challenging environment. Because of the complexity of multi-target movement and the diversity of application scenarios, Radar-based People Counting methods are required to have sufficient robustness. However, based on the review of the current literature, the grouping phenomenon (i.e., multiple individuals moving close together as a single group) was not often considered in the experimental scenarios.

This thesis aims to study one of the most complex motions of individuals, grouping, and address the People Counting problem more in general, including the cases of grouping of multiple individuals. After studying the characteristics of Group People (defined as a group of people sharing neighboring, adjacent locations and moving together), with the help of multiple-input-multiple-output (MIMO) frequency-modulated continuous wave (FMCW) Radar, the combination of the Range-Azimuth map and spectrogram/cadence velocity diagram (CVD) is proposed to solve Group People Counting.

Algorithm-wise, there are two categories of existing Radar-based People Counting methods, namely, tracking for counting methods and feature-based counting methods. It was found that these two categories of methods have complementary strengths. Thus, the proposed method combines the tracking for counting approach and feature-based counting approach into a new processing pipeline to estimate the number of people in each group in the scene while tracking each group. Based on it, the proposed method achieves "Beyond classification", which is the output the unlabeled classes not defined at the training stage. Moreover, compared with other state-of-the-art (SOTA) Radar-based People Counting methods, the proposed method outperforms them, and thus it is proved that the grouping problem can be solved in the Radar-based People Counting field.

In the Roadmap of 2025, EURO NCAP has announced that it will reward car manufacturers that include Child Presence Detection (CPD) technologies in their vehicles starting from 2022. Additionally, seat belt detectors are currently based on pressure sensors which can be falsely triggered with large objects placed on the seats. By basing the detection of people on human vital signs presence instead, these errors can be avoided. Therefore, in recent years, there has been an increasing need to find a solution for multiple people detection and localization in vehicles for both CPD and seat-belt reminder systems in the automotive industry.

With Ultra Wide-Band (UWB) radar, non-invasive human detection is possible through the identification of vital signs characteristics in the radar data. This work aims to improve existing literature by developing a network of UWB radars to perform multiple people detection and localization.

Specifically, an algorithm for de-centralized vital signs detection is proposed, based on the analysis of a novel model for radar signatures of vital signs. Additionally, a centralized association block is developed to fuse the detections from all radars using machine learning-based cost-matrix computation. The performance of the proposed processing pipeline is tested experimentally with a multistatic radar network. A simulation framework is developed for radar data generation to evaluate the results obtained in the experiments, and to propose variations on the evaluated radar topologies.

It can be concluded that the detection and localization of humans in the environment is possible with the proposed framework, with localization RMSE of 16cm for single and double target scenarios. The distribution of multiple focus points and the introduction of bistatic radars enhances the detection, and thus localization, w.r.t. current methods based on monostatic radars and MIMO radars.

Direction of Arrival (DoA) estimation is an important topic in radar application and has significant importance for advanced driver assistant systems in the automotive industry. While there is an increasing need for higher resolution and increased target detection and DoA estimation performance, such improvements often require increased hardware cost, complexity and size. With the increase in computational power of modern systems, Compressive Sensing methods have become more attractive as alternative methods for DoA estimation to the established ones, which often rely on uniform linear arrays (ULA) and the acquisition of multiple snapshots to provide good performance. Compressive Sensing methods have been shown to fit very well into the DoA estimation framework and have the ability to use far fewer snapshots, provide super resolution capabilities and by nature, utilise sparse spatial sampling, i.e. sparse antenna arrays. The latter point is the key incentive of this thesis.

Specifically, Bayesian Compressive Sensing (BCS) which in addition to point estimates also provides measures of uncertainty is used in this thesis, to generate and use sparse
linear array structures for DoA estimation. In particular, the entropy of the recovered coefficient vector is reduced in each step. Two array generation algorithms are proposed
building on the same concept to generate sensor arrays for the consideration of a uniformly spaced, linear grid of possible sensor locations and for a Multiple In Multiple Out
(MIMO) array setup. Utilising sparse arrays with BCS has the potential to reduce the hardware complexity of the circuit board, reduce energy consumption and heat generation, as well as ultimately saving costs in production and operation.

The proposed array generation algorithms are first tested and assessed with simulated Frequency-Modulated Continuous-Waveform (FMCW) radar data, where it is shown that
the generated algorithms achieve good estimation and detection performance with a heavily reduced number of sensors compared to their fully filled template arrays. Moreover, they are shown to outperform randomly generated arrays in most cases that have been studied. To add practical insight, the generated antenna arrays are tested with measured data that has been captured in two measurement campaigns with a Texas Instruments Cascade Evaluation board, featuring an 86 element virtual ULA, which has been used as the grid of possible sensor positions for the array generations. The simulated results are affirmed by the measured data, although more sensors tend to be required depending on the clutter present in the scene. It is shown, that BCS can work very well with the proposed, heavily sparse arrays tested on both simulated and measured data, which translates directly to a possible reduction in required hardware antennas. Although in this thesis the possible sensor positions have been confined to a grid of positions spaced by half the wavelength, it is easily possible to extend the procedure to a more finely divided search space.

Drone detection and tracking systems are nowadays a requirement in most public, private and political events, because of the increasing risk of unintentional or malicious misuse of these platforms. Moreover, in order to ensure adequate protection, full spatial coverage is a must for every such system. However, the research literature focuses on staring radars that have a limited field of view, but which yield rich target information via time-frequency distributions that facilitate the target recognition task. In this thesis, surveillance radars that offer full spatial coverage are presented, albeit their usage for classification is made more complex because of the rotating nature of their antennas which limits the dwell time on targets.

Additionally, due to the incredible fast growth of the drone market, novel counter-drone radars that are able to jointly localize and classify small targets while on-the-move now represent a highly in-demand remote sensing system. Nonetheless, surveillance sensors anchored on moving vehicles are a brand-new technology that is currently being developed. This work therefore investigates surveillance systems in a novel scenario, and presents the technological challenges alongside the proposed solutions to achieve reliable object detection via grounded counter-drone radars on-the-move. Specifically, the required pre-processing steps to remove the clutter from the data while the radar is rotating and moving on the ground are developed and discussed.

In the end, the joint detection and classification problem is traditionally solved separately by different algorithms due to the computational complexity of the task. This thesis project presents a novel framework that localizes and labels drones in an unified pipeline under the umbrella of object detection via computer vision, and that is able to operate while being static or on-the-move. Thus, an end-to-end radar data processing architecture that is robust against homogeneity constraints and based on You Only Look Once (YOLO) model is used to perform object detection in real-time. In brief, this work opens new avenues towards multi-class and multi-instance plot-based target detection and classification by transferring cross-disciplinary algorithms from computer vision into remote sensing.