Lorenzo Servadei

• Doctor of Engineering
• Senior Researcher at Sony Corporation

Autonomous driving systems require accurate and reliable perception. However, adverse environments, such as rain, snow, and fog, can significantly degrade the performance of LiDAR and cameras. In contrast, 4D millimeter-wave (mmWave) radar not only provides 3D sensing and additional velocity measurements but also maintains robustness in challenging...

In physics and engineering, many processes are modeled using non-differentiable black-box simulators, making the optimization of such functions particularly challenging. To address such cases, inspired by the Gradient Theorem, we propose locality-aware surrogate models for active model-based black-box optimization. We first establish a theoretical...

Radar and LiDAR have been widely used in autonomous driving as LiDAR provides rich structure information, and radar demonstrates high robustness under adverse weather. Recent studies highlight the effectiveness of fusing radar and LiDAR point clouds. However, challenges remain due to the modality misalignment and information loss during feature ext...

This study introduces a novel physically interpretable data augmentation framework to improve the robustness and accuracy of hand gesture recognition using Frequency-Modulated Continuous Wave (FMCW) radar and Convolutional Neural Networks (CNN). The proposed reconfigurable and parametric method modifies specific characteristics of five time-series...

Recently, radar-camera fusion algorithms have gained significant attention as radar sensors provide geometric information that complements the limitations of cameras. However, most existing radar-camera depth estimation algorithms focus solely on improving performance, often neglecting computational efficiency. To address this gap, we propose LiRCD...

Designing integrated circuits involves substantial complexity, posing challenges in revealing its potential applications - from custom digital cells to analog circuits. Despite extensive research over the past decades in building versatile and automated frameworks, there remains open room to explore more computationally efficient AI-based solutions...

Machine learning models are advancing circuit design, particularly in analog circuits. They typically generate netlists that lack human interpretability. This is a problem as human designers heavily rely on the interpretability of circuit diagrams or schematics to intuitively understand, troubleshoot, and develop designs. Hence, to integrate domain...

The increasing demand in artificial intelligence (AI) for models that are both effective and explainable is critical in domains where safety and trust are paramount. In this study, we introduce MIRA, a transparent and interpretable multi-class rule-based algorithm tailored for radar-based gesture detection. Addressing the critical need for understa...

Depth estimation plays a pivotal role in autonomous driving, facilitating a comprehensive understanding of the vehicle's 3D surroundings. Radar, with its robustness to adverse weather conditions and capability to measure distances, has drawn significant interest for radar-camera depth estimation. However, existing algorithms process the inherently...

In recent years, approaches based on radar object detection have made significant progress in autonomous driving systems due to their robustness under adverse weather compared to LiDAR. However, the sparsity of radar point clouds poses challenges in achieving precise object detection, highlighting the importance of effective and comprehensive featu...

Depth estimation is critical in autonomous driving for interpreting 3D scenes accurately. Recently, radar-camera depth estimation has become of sufficient interest due to the robustness and low-cost properties of radar. Thus, this paper introduces a two-stage, end-to-end trainable Confidence-aware Fusion Net (CaFNet) for dense depth estimation, com...

Accurate and continuous monitoring of vital signs like heart and breathing rate is essential for timely diagnosis and early detection of critical health conditions. While traditional wearable devices fulfill this purpose, they can be uncomfortable and impractical for continuous use. In this study, we propose a novel classifier for radar-based vital...

The EU AI Act underscores the importance of transparency, user-centricity, and robustness in AI systems, particularly for high-risk systems. In response, we present advancements in XentricAI, an explainable hand gesture recognition (HGR) system designed to meet these regulatory requirements. XentricAI adresses fundamental challenges in HGR, such as...

This paper introduces a novel system designed to enhance the robustness of airborne ultrasound technology against in-band interference. By integrating MEMS-based transceivers, the system achieves ultrasonic pulse-echo operation up to 100 kHz while maintaining low power consumption. A dedicated hardware readout block effectively addresses the signal...

In this study, we introduce a denoising autoencoder as a solution to mitigate artifacts in MEMS-based airborne ultrasound transceivers. The autoencoder employs a deep neural network architecture to learn a robust representation of ultrasonic echo signals produced by reflecting targets, effectively eliminating unwanted noise. Experimental pulsed-ech...

In many industrial or healthcare contexts, keeping track of the number of people is essential. Radar systems, with their low overall cost and power consumption, enable privacy-friendly monitoring in many use cases. Yet, radar data are hard to interpret and incompatible with most computer vision strategies. Many current deep learning-based systems a...

Accurate and robust object detection is critical for autonomous driving. Image-based detectors face difficulties caused by low visibility in adverse weather conditions. Thus, radar-camera fusion is of particular interest but presents challenges in optimally fusing heterogeneous data sources. To approach this issue, we propose two new radar preproce...

With the growing concern for air quality and its impact on human health, interest in environmental gas monitoring has increased. However, chemi-resistive gas sensing devices are plagued by issues of sensor reproducibility during manufacturing. This study proposes a novel approach for detecting sensor-to-sensor variations in sensing devices using th...

In this paper, we propose a non-contact vital sign monitoring algorithm based on 60GHz Frequency Modulated Continuous Wave (FMCW) radar. At the heart of the algorithm is a bank of sinc filters covering the heart and respiratory frequencies of an healthy adult. The filter bank is adapted, in successive iterations, to the specific cardio-respiratory...

Blood pressure monitoring is of paramount importance in the assessment of a human’s cardiovascular health. The state-of-the-art method remains the usage of an upper-arm cuff sphygmomanometer. However, this device suffers from severe limitations—it only provides a static blood pressure value pair, is incapable of capturing blood pressure variations...

Editor’s notes: Deep reinforcement learning is shown to improve the design cost of hardware char63software interfaces within an industrial design framework. Based on optimization preferences specified by a designer, the proposed approach generates optimized solutions. —Ulf Schlichtmann, Technical University of Munich

Vital signs estimation provides valuable information about an individual’s overall health status. Gathering such information usually requires wearable devices or privacy-invasive settings. In this work, we propose a radar-based user-adaptable solution for respiratory signal prediction while sitting at an office desk. Such an approach leads to a con...

This paper proposes a novel pre-processing technique for radar-based short-range gesture sensing using a frequency modulated continuous wave (FMCW) radar. The pre-processing is light-weight and works without Fourier transformation. The signal after pre-processing represents the backscattering central dynamics of the hand as a complex-valued time si...

The quality of voice communication using True Wireless Stereo (TWS) is not ideal in noisy environments. Existing two microphone beamforming solution is unable to isolate the users voice reliably in all possible acoustic environment. Here, the use of bone-conducted voice pickup holds promise for improving voice calling quality. There are some microp...

Introduces multiple state-of-the-art deep learning architectures for mmwave radar in a variety of advanced applications. Methods and Techniques in Deep Learning: Advancements in mmWave Radar Solutions provides a timely and authoritative overview of the use of artificial intelligence (AI)-based processing for various mmwave radar applications. Focu...

Current state‐of‐the‐art deep learning‐based radar applications rely heavily on preprocessed data. On the one hand, preprocessing the data leads to image‐like representations of the radar data, allowing the use of neural network architectures established in computer vision domain, but on the other hand, preprocessing is also a manually chosen and n...

Deep learning based approaches have achieved state‐of‐the‐art results in a variety of tasks recently. In computer vision and natural language processing especially, the large corpus of available training data and often the abstract problems are suitable for data‐driven approaches. Often, however, the available dataset is limited in size. In these c...

This chapter will give the readers an overview on the inherit limitations of conventional deep learning architectures, i.e. convolution neural network (CNN), long short‐term memory (LSTM), and need of Geometric Deep Learning to capture and learn underlying patterns in more complex non‐Euclidean data structure. The learning approach used by CNN/LSTM...

Deep reinforcement learning (DRL) utilizes deep learning techniques to model reinforcement learning (RL) tasks. Recently, DRL has gained significant attention due to its remarkable success. Thanks to brilliant solutions in gaming, supply chain, flight schedule, and many other areas, DRL made a clear movement from academia to industry. In order to s...

In Domain Adaptation, an algorithm learns from a source target distribution to predict on a typically related target data distribution. To this end, we define Domain Adaptation and position it in relation to transfer learning. Additionally, we explain different subcategories of Domain Adaptation and show how these different Domain Adaptation method...

This chapter will give an overview on the fundamental principle of learning theory. Later, keeping goal of learning theory, formulation of deterministic architecture derived followed by its inherent limitation on representation learning. These limitations are due to model architecture and imitated or noisy measurement data. This leads to formulatio...

This chapter provides a fundamental introduction to multimodal deep learning approaches which allows readers to not only have a strong technical understanding of how one can build solutions around a multimodal dataset but also an intuitive understanding about these learning methodologies which hold strong resemblance with how we perceive the world...

There exist several radar applications that aim to classify between a set of predefined classes such as different human activities or hand gestures. However, in real‐world environments more than the predefined classes exist, which can lead to a high number of false alarms and turns the problem into an open‐set classification task. Deep metric learn...

The chapter introduces the fundamentals of radar signal processing, and how a target is detected to track and explains the rationale behind it. The chapter introduces deep learning, its evolution over time, and the different facets that make deep learning so powerful. Various components of conventional convolutional neural networks, recurrent neura...

Nowadays, Deep Learning (DL) methods often overcome the limitations of traditional signal processing approaches. Nevertheless, DL methods are barely applied in real-life applications. This is mainly due to limited robustness and distributional shift between training and test data. To this end, recent work has proposed uncertainty mechanisms to incr...

Data selection is essential for any data-based optimization technique, such as Reinforcement Learning. State-of-the-art sampling strategies for the experience replay buffer improve the performance of the Reinforcement Learning agent. However, they do not incorporate uncertainty in the Q-Value estimation. Consequently, they cannot adapt the sampling...

Nowadays, deep neural networks are widely used in a variety of fields that have a direct impact on society. Although those models typically show outstanding performance, they have been used for a long time as black boxes. To address this, Explainable Artificial Intelligence (XAI) has been developing as a field that aims to improve the transparency...

Nowadays, the digital chip design flow starts with formal specifications, which are mapped to Register Transfer Level (RTL) models using different underlying (micro-) architectures. By doing so, a hardware designer predicts and resolves time-critical parts to achieve an RTL-design that meets all constraints after synthesis. However, wrong predictio...

Driven by Moore’s law, the chip design complexity is steadily increasing. Electronic Design Automation (EDA) has been able to cope with the challenging very large-scale integration process, assuring scalability, reliability, and proper time-to-market. However, EDA approaches are time and resource-demanding, and they often do not guarantee optimal s...

Gesture recognition is one of the most intuitive ways of interaction and has gathered particular attention for human computer interaction. Radar sensors possess multiple intrinsic properties, such as their ability to work in low illumination, harsh weather conditions, and being low-cost and compact, making them highly preferable for a gesture recog...

Commercial radar sensing is gaining relevance and machine learning algorithms constitute one of the key components that are enabling the spread of this radio technology into areas like surveillance or healthcare. However, radar datasets are still scarce and generalization cannot be yet achieved for all radar systems, environment conditions or desig...

Multi-target tracking with radars is a highly challenging problem due to detection artifacts, sensor noise, and interference sources. The traditional signal processing chain is, therefore, a complex combination of various algorithms with several tunable tracking-parameters. Usually, these are initially set by engineers and are independent of the sc...

Technological advances and scalability are leading Human-Computer Interaction (HCI) to evolve towards intuitive forms, such as through gesture recognition. Among the various interaction strategies, radar-based recognition is emerging as a touchless, privacy-secure, and versatile solution in different environmental conditions. Classical radar-based...

Machine Learning (ML) methods have become state of the art in radar signal processing, particularly for classification tasks (e.g., of different human activities). Radar classification can be tedious to implement, though, due to the limited size and diversity of the source dataset, i.e., the data measured once for initial training of the Machine Le...

In this paper, we introduce the Label-Aware Ranked loss, a novel metric loss function. Compared to the state-of-the-art Deep Metric Learning losses, this function takes advantage of the ranked ordering of the labels in regression problems. To this end, we first show that the loss minimises when datapoints of different labels are ranked and laid at...

Malware applications are one of the major threats that computing systems face today. While security researchers develop new defense mechanisms to detect malware, attackers continue to release new malware families that evade detection. New defense mechanisms must therefore be developed to effectively counter malware. Hardware Performance Counters (H...

The “HPI Future SOC Lab” is a cooperation of the Hasso Plattner Institute (HPI) and industry partners. Its mission is to enable and promote exchange and interaction between the research community and the industry partners. The HPI Future SOC Lab provides researchers with free of charge access to a complete infrastructure of state of the art hard an...

Hardware/software co-designs are usually defined at high levels of abstractions at the beginning of the design process in order to provide of options of how to realize a system. This allows for design exploration which relies on knowing the costs of different design configurations (with respect to hardware usage and firmware metrics). To this end,...

In order to continuously increase design productivity, engineers and researchers rely on automation frameworks for hardware and firmware design purposes. However, a major problem for optimizing the generated design is retrieving data from which a prediction function (e.g., area, speed) could be learned correctly (since complete generation and synth...

Advancements of Machine Learning (ML) in the field of computer vision have paved the way for its potential application in many other fields. Researchers and Hardware domain experts are exploring possible applications of Machine Learning in optimizing many aspects of the Hardware development process. In this paper, we propose a novel approach for pr...

E-health is a key factor in the E-society. E-health, in fact, enhances the efficiency and reduces the costs of the health services. In the diagnostic field, E-health can avail of several Machine Learning (ML) algorithms, as Artificial Neural Networks (ANNs) for instance, which often demonstrated a high classification accuracy. Although ANNs have al...

Predictive analytics has become a widely accepted and adopted method for workload and demand. It refers to the application of “[…] statistical models and other empirical methods that are aimed at creating empirical predictions (as opposed to predictions that follow from theory only), as well as methods for assessing the quality of those predictions...

E-health is a key factor in the E-society. E-health, in fact, enhances the efficiency and reduces the costs of the health services. In the diagnostic field, E-health can avail of several Machine Learning (ML) algorithms, as Artificial Neural Networks (ANNs) for instance, which often demonstrated a high classification accuracy. Although ANNs have al...

The aim of the project is to create an easy and helpful car pooling web portal which guarantees integrity and preservation of released personal information and be a trustworthy platform for drivers as well as passengers. To create such a web portal, it has been necessary the use of distinct and various material and methods, first in phase of appreh...