R.T. Rajan
info
Please Note
<p>This page displays the records of the person named above and is not linked to a unique person identifier. This record may need to be merged to a profile.</p>
75 records found
1
Gaussian process state-space models are a widely used modeling paradigm for learning and estimation in dynamical systems. Reduced-rank Gaussian process state-space models combine spectral characterization of dynamical systems with Hilbert space methods to enable learning, which scale linearly with the length of the time series. However, the current state of the art algorithms struggle to deal efficiently with the dimensionality of the state-space itself. In this work, we propose a novel algorithm, referred to as Domain-Aware reduced-rank Gaussian Process State-Space Model (DA-GPSSM), which exploits the relationship between state dimensions to model only necessary dynamics resulting in reduced computational cost, by potentially orders of magnitude in comparison to the state-of-the-art. The proposed approach grants modeling flexibility while maintaining comparable performance and thus increasing the applicability of these models. We present implications of the proposed approach and discuss applications where DA-GPSSM can be beneficial. Finally, we conduct simulations to demonstrate the performance and reduced computational cost of our proposed method, compared to the state-of-the-art learning method, and propose future research directions.
...
Gaussian process state-space models are a widely used modeling paradigm for learning and estimation in dynamical systems. Reduced-rank Gaussian process state-space models combine spectral characterization of dynamical systems with Hilbert space methods to enable learning, which scale linearly with the length of the time series. However, the current state of the art algorithms struggle to deal efficiently with the dimensionality of the state-space itself. In this work, we propose a novel algorithm, referred to as Domain-Aware reduced-rank Gaussian Process State-Space Model (DA-GPSSM), which exploits the relationship between state dimensions to model only necessary dynamics resulting in reduced computational cost, by potentially orders of magnitude in comparison to the state-of-the-art. The proposed approach grants modeling flexibility while maintaining comparable performance and thus increasing the applicability of these models. We present implications of the proposed approach and discuss applications where DA-GPSSM can be beneficial. Finally, we conduct simulations to demonstrate the performance and reduced computational cost of our proposed method, compared to the state-of-the-art learning method, and propose future research directions.
Multi-agent systems, such as fleets of robots or drones, are increasingly deployed in logistics, inspection, and surveillance. These systems must reach their targets while maintaining safe separation, even under uncertain dynamics. This is challenging because unmodeled effects, disturbances, and sensor noise can degrade tracking performance and compromise safety. Model Predictive Control (MPC) is well suited for multi-agent navigation since it optimizes trajectories over a prediction horizon while enforcing input and state constraints. However, its performance depends on accurate models, and centralized formulations suffer from poor scalability and a single point of failure. We propose a cooperative Gaussian Process–augmented MPC (GP-MPC) framework that combines learning, chance-constrained safety, and distributed optimization. Each agent uses a Gaussian Process to learn its residual dynamics and quantify local uncertainty, incorporates this uncertainty into a chance-constrained collision-avoidance scheme, and coordinates only with neighbors through an ADMM-based distributed optimization method. This integration provides robustness to model errors and scalability to larger teams. The framework enables collision avoidance using only local uncertainty estimates, removing the need to share covariance information. By restricting computation and communication to each agent’s neighborhood, it maintains scalability and efficiency. Simulations show that the approach yields smoother and more efficient trajectories, faster convergence to targets, and reliable probabilistic safety compared to nominal and nonlinear MPC baselines. Convergence analysis further confirms robust consensus across a range of tuning parameters.
...
Multi-agent systems, such as fleets of robots or drones, are increasingly deployed in logistics, inspection, and surveillance. These systems must reach their targets while maintaining safe separation, even under uncertain dynamics. This is challenging because unmodeled effects, disturbances, and sensor noise can degrade tracking performance and compromise safety. Model Predictive Control (MPC) is well suited for multi-agent navigation since it optimizes trajectories over a prediction horizon while enforcing input and state constraints. However, its performance depends on accurate models, and centralized formulations suffer from poor scalability and a single point of failure. We propose a cooperative Gaussian Process–augmented MPC (GP-MPC) framework that combines learning, chance-constrained safety, and distributed optimization. Each agent uses a Gaussian Process to learn its residual dynamics and quantify local uncertainty, incorporates this uncertainty into a chance-constrained collision-avoidance scheme, and coordinates only with neighbors through an ADMM-based distributed optimization method. This integration provides robustness to model errors and scalability to larger teams. The framework enables collision avoidance using only local uncertainty estimates, removing the need to share covariance information. By restricting computation and communication to each agent’s neighborhood, it maintains scalability and efficiency. Simulations show that the approach yields smoother and more efficient trajectories, faster convergence to targets, and reliable probabilistic safety compared to nominal and nonlinear MPC baselines. Convergence analysis further confirms robust consensus across a range of tuning parameters.
Accurate tracking of targets is vital for safe and reliable operations, particularly in complex and dynamic environments such as urban areas. Traditional tracking methods, including Kalman and particle filters, often perform poorly in real world scenarios, due to inaccurate models and sparse or noisy measurements. Gaussian process (GP) based methods offer a flexible and data driven alternative with uncertainty quantification that does not depend on predefined dynamical equations. However, state of the art GP tracking approaches require expensive hyperparameter optimization, which limits their practicality for real time applications. In this work, we introduce a novel GP mixture based computationally efficient tracking method, which is capable of modeling complex system behavior and adapt to changing dynamics. Our proposed solution, named Multiple Model Recursive Gaussian Process (MM-RGP), adapts continuously to changing dynamics, is capable of modeling complex behavior, and is robust against sparse observation. In addition, the proposed method avoids hyperparameter optimization and adapts to incoming data. We demonstrate the effectiveness of our solution using the example of uncrewed aerial vehicle (UAV) tracking, with both simulated and real datasets, and propose directions for extending our work.
...
Accurate tracking of targets is vital for safe and reliable operations, particularly in complex and dynamic environments such as urban areas. Traditional tracking methods, including Kalman and particle filters, often perform poorly in real world scenarios, due to inaccurate models and sparse or noisy measurements. Gaussian process (GP) based methods offer a flexible and data driven alternative with uncertainty quantification that does not depend on predefined dynamical equations. However, state of the art GP tracking approaches require expensive hyperparameter optimization, which limits their practicality for real time applications. In this work, we introduce a novel GP mixture based computationally efficient tracking method, which is capable of modeling complex system behavior and adapt to changing dynamics. Our proposed solution, named Multiple Model Recursive Gaussian Process (MM-RGP), adapts continuously to changing dynamics, is capable of modeling complex behavior, and is robust against sparse observation. In addition, the proposed method avoids hyperparameter optimization and adapts to incoming data. We demonstrate the effectiveness of our solution using the example of uncrewed aerial vehicle (UAV) tracking, with both simulated and real datasets, and propose directions for extending our work.
Unmanned Aircraft Systems (UAS) have seen a significant growth in civilian space over the past decade. The number one ranked challenge in UAS operations in Europe is regulatory obstacles such as the Specific Operations Risk Assessment (SORA) for 2023–2025. Existing approaches have focused on individual technical solutions (radio technologies, redundancy schemes, or cryptographic protections) or on high-level safety analysis, but have not integrated regulatory compliance, risk assessment, and repeatable systems models that directly support SORA artifact generation and rapid adaptation across BVLOS operational contexts. Thus, the current state-of-the-art apparatus lacks a systematic Model-Based Systems Engineering (MBSE) approach that can cater to Command and Control (C2) data-link design for Beyond Visual Line-of-Sight (BVLOS) missions. In this work, we propose an MBSE methodology designed to assist engineers in designing a C2 data link for BVLOS drone operations that complies with SORA regulations in the Netherlands and Europe. To validate the use of MBSE in a wide range of complex drone operations, we demonstrate how subtle modifications in the proposed engineering models can be made without any major overhaul of new SORA applications, and this is validate these changes through laboratory software tests and simulations.
...
Unmanned Aircraft Systems (UAS) have seen a significant growth in civilian space over the past decade. The number one ranked challenge in UAS operations in Europe is regulatory obstacles such as the Specific Operations Risk Assessment (SORA) for 2023–2025. Existing approaches have focused on individual technical solutions (radio technologies, redundancy schemes, or cryptographic protections) or on high-level safety analysis, but have not integrated regulatory compliance, risk assessment, and repeatable systems models that directly support SORA artifact generation and rapid adaptation across BVLOS operational contexts. Thus, the current state-of-the-art apparatus lacks a systematic Model-Based Systems Engineering (MBSE) approach that can cater to Command and Control (C2) data-link design for Beyond Visual Line-of-Sight (BVLOS) missions. In this work, we propose an MBSE methodology designed to assist engineers in designing a C2 data link for BVLOS drone operations that complies with SORA regulations in the Netherlands and Europe. To validate the use of MBSE in a wide range of complex drone operations, we demonstrate how subtle modifications in the proposed engineering models can be made without any major overhaul of new SORA applications, and this is validate these changes through laboratory software tests and simulations.
Distributed affine formation control (AFC) enables unmanned aerial vehicle (UAV) swarms to achieve coordinated motion while maintaining a desired geometric configuration, thereby enhancing their maneuverability. However, conventional leader-follower-based distributed AFC remains vulnerable to dynamic changes in the network topology, where UAVs may be temporarily unavailable due to maintenance in real-world applications. In this work, we propose a reliable maintenance policy that enables individual UAVs to detach from the swarm without compromising the formation stability. Our policy introduces an agent homogenization strategy that replaces the conventional leader UAVs with virtual leaders, thereby ensuring all operational UAVs are followers and thus eligible for maintenance. A relative affine localization (RAL) technique is employed, which allows the remaining UAVs to estimate the relative positions of missing neighbors by leveraging the formation geometry. Our proposed framework is validated through a series of experiments with a swarm of Crazyflie quadrotors in an indoor environment, which demonstrate the effectiveness of our policy that allows individual UAVs to be removed and returned in sequence while the rest of the swarm maintains its target configuration with high accuracy. Our proposed maintenance policy enables robust and long-duration deployments of UAV swarms in inaccessible and harsh environments.
...
Distributed affine formation control (AFC) enables unmanned aerial vehicle (UAV) swarms to achieve coordinated motion while maintaining a desired geometric configuration, thereby enhancing their maneuverability. However, conventional leader-follower-based distributed AFC remains vulnerable to dynamic changes in the network topology, where UAVs may be temporarily unavailable due to maintenance in real-world applications. In this work, we propose a reliable maintenance policy that enables individual UAVs to detach from the swarm without compromising the formation stability. Our policy introduces an agent homogenization strategy that replaces the conventional leader UAVs with virtual leaders, thereby ensuring all operational UAVs are followers and thus eligible for maintenance. A relative affine localization (RAL) technique is employed, which allows the remaining UAVs to estimate the relative positions of missing neighbors by leveraging the formation geometry. Our proposed framework is validated through a series of experiments with a swarm of Crazyflie quadrotors in an indoor environment, which demonstrate the effectiveness of our policy that allows individual UAVs to be removed and returned in sequence while the rest of the swarm maintains its target configuration with high accuracy. Our proposed maintenance policy enables robust and long-duration deployments of UAV swarms in inaccessible and harsh environments.
Field estimation in spatio-temporally evolving environments remains challenging, particularly when limited sensor resources must capture dynamic features while contending with modeling errors and measurement noise e.g., in environmental monitoring using aerial vehicles, where system dynamics interact with practical sensing limitations. In this work, we consider a scenario where a network of mobile sensor nodes measure an advection-diffusion field, where the sensor locations can be dynamically optimized based on PDE residuals e.g., sensors on-board drones. Our novel two-stage framework strategically integrates Gaussian Process regression with PDE constraints. An initial inference stage estimates key parameters (e.g., advection velocity, diffusion coefficient) through stationary sensor measurements and finite-difference derivative approximations, while a subsequent mobility stage employs forward-Euler time-stepping to dynamically relocate the sensors toward regions of high PDE residual. Simulations based on a 2D advection-diffusion field experiment reveals upto an order magnitude improvement in field reconstruction error, as compared to information theoretic deployments. We conclude with future directions of extending our work and suggest applications.
...
Field estimation in spatio-temporally evolving environments remains challenging, particularly when limited sensor resources must capture dynamic features while contending with modeling errors and measurement noise e.g., in environmental monitoring using aerial vehicles, where system dynamics interact with practical sensing limitations. In this work, we consider a scenario where a network of mobile sensor nodes measure an advection-diffusion field, where the sensor locations can be dynamically optimized based on PDE residuals e.g., sensors on-board drones. Our novel two-stage framework strategically integrates Gaussian Process regression with PDE constraints. An initial inference stage estimates key parameters (e.g., advection velocity, diffusion coefficient) through stationary sensor measurements and finite-difference derivative approximations, while a subsequent mobility stage employs forward-Euler time-stepping to dynamically relocate the sensors toward regions of high PDE residual. Simulations based on a 2D advection-diffusion field experiment reveals upto an order magnitude improvement in field reconstruction error, as compared to information theoretic deployments. We conclude with future directions of extending our work and suggest applications.
Affine formation control (AFC) is a subset of formation control methods that enables coordinated multiagent movement while preserving affine relationships, and has recently gained increasing popularity due to its utility across diverse applications. AFC is inherently distributed, where each agent's local controller relies on the relative displacements of neighboring agents. The unavailability of these measurements in practice, due to node or communication failures, leads to a change in the underlying graph topology and subsequently causes instability or sub-optimal performance. In this work, each edge in the graph is modeled using a state-space framework, allowing the corresponding edge-states to be estimated with or without up-to-date measurements. We then propose a Kalman-based estimation framework where we fuse both temporal information from agents' dynamics and spatial information, which is derived from the geometry of the affine formations. We give convergence guarantees and optimality analysis on the proposed algorithm, and numerical validations show the enhanced robustness of AFC against these topology changes in several practical scenarios.
...
Affine formation control (AFC) is a subset of formation control methods that enables coordinated multiagent movement while preserving affine relationships, and has recently gained increasing popularity due to its utility across diverse applications. AFC is inherently distributed, where each agent's local controller relies on the relative displacements of neighboring agents. The unavailability of these measurements in practice, due to node or communication failures, leads to a change in the underlying graph topology and subsequently causes instability or sub-optimal performance. In this work, each edge in the graph is modeled using a state-space framework, allowing the corresponding edge-states to be estimated with or without up-to-date measurements. We then propose a Kalman-based estimation framework where we fuse both temporal information from agents' dynamics and spatial information, which is derived from the geometry of the affine formations. We give convergence guarantees and optimality analysis on the proposed algorithm, and numerical validations show the enhanced robustness of AFC against these topology changes in several practical scenarios.
Estimation of the relative positions of N static nodes in D-dimensional space given the pairwise distances between them is a well-studied problem in literature. However, for a network of mobile nodes, the existing solutions proposed in literature rely either on the knowledge of absolute positions of some nodes or enforce constraints on the motion of individual nodes to achieve a unique solution. In this work, we consider an anchorless environment and propose a time-varying Grammian-based data model which relates the relative positions of the mobile nodes to the pairwise distances between them. Given the data model, we propose algorithms to estimate the relative positions, velocity and other higher order derivatives, referred to as relative kinematics, associated with the network of mobile nodes. We further consider a scenario where accelerometers are on-board on all the mobile nodes, and investigate the inclusion the accelerometer measurements in the proposed model. The Cramér-Rao lower bound for the proposed data models are derived and compared with the performance of the estimators using Monte-Carlo simulations. We further compare and analyze the performance of the proposed estimators against the state-of-the-art methods, and present research directions for future work to further improve the proposed approach.
...
Estimation of the relative positions of N static nodes in D-dimensional space given the pairwise distances between them is a well-studied problem in literature. However, for a network of mobile nodes, the existing solutions proposed in literature rely either on the knowledge of absolute positions of some nodes or enforce constraints on the motion of individual nodes to achieve a unique solution. In this work, we consider an anchorless environment and propose a time-varying Grammian-based data model which relates the relative positions of the mobile nodes to the pairwise distances between them. Given the data model, we propose algorithms to estimate the relative positions, velocity and other higher order derivatives, referred to as relative kinematics, associated with the network of mobile nodes. We further consider a scenario where accelerometers are on-board on all the mobile nodes, and investigate the inclusion the accelerometer measurements in the proposed model. The Cramér-Rao lower bound for the proposed data models are derived and compared with the performance of the estimators using Monte-Carlo simulations. We further compare and analyze the performance of the proposed estimators against the state-of-the-art methods, and present research directions for future work to further improve the proposed approach.
Traditional target tracking using monostatic radar systems typically rely on centralized or decentralized architectures, where all data is transmitted to a fusion center for estimating the position and velocity of mobile agents. This approach introduces a single point of failure and can significantly increase communication costs, particularly when the fusion center is far from individual radar nodes. To overcome these issues, we introduce a distributed Alternating Direction Method of Multipliers (ADMM) for target localization using a radar network, wherein each radar node shares its observed data only with its immediate neighboring nodes, and achieves consensus with the radar network on the estimated target locations and velocities. We perform simulations incorporating critical system parameters such as the number of radar nodes and Signal-to-Noise Ratio (SNR) to assess their impact of estimation accuracy and convergence speed of the proposed distributed ADMM algorithm. We highlight the additional benefits of our proposed solution, and present directions for future work.
...
Traditional target tracking using monostatic radar systems typically rely on centralized or decentralized architectures, where all data is transmitted to a fusion center for estimating the position and velocity of mobile agents. This approach introduces a single point of failure and can significantly increase communication costs, particularly when the fusion center is far from individual radar nodes. To overcome these issues, we introduce a distributed Alternating Direction Method of Multipliers (ADMM) for target localization using a radar network, wherein each radar node shares its observed data only with its immediate neighboring nodes, and achieves consensus with the radar network on the estimated target locations and velocities. We perform simulations incorporating critical system parameters such as the number of radar nodes and Signal-to-Noise Ratio (SNR) to assess their impact of estimation accuracy and convergence speed of the proposed distributed ADMM algorithm. We highlight the additional benefits of our proposed solution, and present directions for future work.
Affine formation control (AFC) is a distributed networked control system that has recently received increasing attention in various applications. AFC is typically achieved using a generalized consensus system where the stress matrix, which encodes the graph structure, is used instead of a graph Laplacian. Universally rigid frameworks (URFs) guarantee the existence of the stress matrix and have thus become the guideline for such a network design. In this work, we propose a convex optimization framework to design the stress matrix for AFC without predefining a rigid graph. We aim to find a resulting network with a reduced number of communication links, but still with a fast convergence speed. We show through simulations that our proposed solutions can yield a more sparse graph, while admitting a faster convergence compared to the state-of-the-art solutions.
...
Affine formation control (AFC) is a distributed networked control system that has recently received increasing attention in various applications. AFC is typically achieved using a generalized consensus system where the stress matrix, which encodes the graph structure, is used instead of a graph Laplacian. Universally rigid frameworks (URFs) guarantee the existence of the stress matrix and have thus become the guideline for such a network design. In this work, we propose a convex optimization framework to design the stress matrix for AFC without predefining a rigid graph. We aim to find a resulting network with a reduced number of communication links, but still with a fast convergence speed. We show through simulations that our proposed solutions can yield a more sparse graph, while admitting a faster convergence compared to the state-of-the-art solutions.
One of the key challenges for multi-agent systems is collision free navigation in an unknown environment. In this work, we propose a unified framework for joint localization, control, and collision avoidance of multi-agent systems navigating in an unknown environment in the presence of dynamic obstacles. The cooperative agents rely on information from immediate neighboring agents within their communication neighborhood, and the dynamic obstacles are modelled as non-cooperative agents. The agents achieve localization by exploiting the individual agent dynamics, and pairwise distance measurements with agents in the sensing neighborhood of each cooperative agent. To ensure collision-free navigation, we exploit a Model Predictive Control (MPC) for each agent, with avoidance constraints using safety radius between pairwise agents. Futhermore, to avoid single point of failure, we propose Cooperative Positioning, Control and Collision Avoidance (CPCCA), which is based on distributed Method of Multipliers methods. We validate our framework and algorithms through simulations, demonstrating its effectiveness in real world scenarios, and propose directions for future work.
...
One of the key challenges for multi-agent systems is collision free navigation in an unknown environment. In this work, we propose a unified framework for joint localization, control, and collision avoidance of multi-agent systems navigating in an unknown environment in the presence of dynamic obstacles. The cooperative agents rely on information from immediate neighboring agents within their communication neighborhood, and the dynamic obstacles are modelled as non-cooperative agents. The agents achieve localization by exploiting the individual agent dynamics, and pairwise distance measurements with agents in the sensing neighborhood of each cooperative agent. To ensure collision-free navigation, we exploit a Model Predictive Control (MPC) for each agent, with avoidance constraints using safety radius between pairwise agents. Futhermore, to avoid single point of failure, we propose Cooperative Positioning, Control and Collision Avoidance (CPCCA), which is based on distributed Method of Multipliers methods. We validate our framework and algorithms through simulations, demonstrating its effectiveness in real world scenarios, and propose directions for future work.
Journal article
(2025)
-
Metin Calis, Massimo Mischi, Alle-Jan van der Veen, Raj Thilak Rajan, Borbàla Hunyadi
Tensor decomposition methods for signal processing applications are an active area of research. Real data are often low-rank, noisy, and come in a higher-order format. As such, low-rank tensor approximation methods that account for the high-order structure of the data are often used for denoising. One way to represent a tensor in a low-rank form is to decompose the tensor into a set of orthonormal factor matrices and an all-orthogonal core tensor using a higher-order singular value decomposition. Under noisy measurements, the lower bound for recovering the factor matrices and the core tensor is unknown. In this paper, we exploit the well-studied constrained Cramér-Rao bound to calculate a lower bound on the mean squared error of the unbiased estimates of the components of the multilinear singular value decomposition under additive white Gaussian noise, and we validate our approach through simulations.
...
Tensor decomposition methods for signal processing applications are an active area of research. Real data are often low-rank, noisy, and come in a higher-order format. As such, low-rank tensor approximation methods that account for the high-order structure of the data are often used for denoising. One way to represent a tensor in a low-rank form is to decompose the tensor into a set of orthonormal factor matrices and an all-orthogonal core tensor using a higher-order singular value decomposition. Under noisy measurements, the lower bound for recovering the factor matrices and the core tensor is unknown. In this paper, we exploit the well-studied constrained Cramér-Rao bound to calculate a lower bound on the mean squared error of the unbiased estimates of the components of the multilinear singular value decomposition under additive white Gaussian noise, and we validate our approach through simulations.
Building Blocks for the Dark-Ages EXplorer (DEX)
Enabling a Lunar Radio Telescope and Advancing Multi-Purpose Infrastructure for Sustainable Lunar Presence
Conference paper
(2025)
-
J. Lazendic-Galloway, M. Bentum, U. Johannsen, C. Brinkerink, M. Klein Wolt, L. V.E. Koopmans, A. J. Boonstra, J. Carpenter, R. T. Rajan, D. Prinsloo
The deployment of a large radio telescope array on the Moon represents a transformative leap for both scientific discovery and technological innovation. The Dark-Ages EXplorer (DEX) concept envisions a large-scale, low-frequency radio array on the lunar surface, capable of conducting groundbreaking observations of the early Universe. Achieving this ambitious goal requires an array of 1000-100,000 antennas, along with novel hardware and software platforms, posing significant engineering challenges. Historically, radio astronomy has been a catalyst for technological progress, and the advancements required for DEX could serve as foundational technologies for a wide range of applications. These innovations aim to enable new scientific discoveries while also supporting a sustainable human presence on the Moon and terrestrial applications. In this paper, we present key technological challenges identified in the recent Concurrent Design Facility (CDF) study, done in collaboration with the European Space Agency (ESA). Technological developments needed to bring DEX to reality have broader applications for future research and commercial activities on the Moon, including energy distribution, autonomous systems, thermal management, communications networks, software development, data management, signal processing, AI/ML and distributed optimisation. By addressing these challenges, we aim to foster cross-sector collaboration and accelerate the development of technologies essential for a sustainable and scientifically productive future on the Moon. Thus, DEX serves not only as an observatory but also as a building block for sustainable lunar exploration and development.
...
The deployment of a large radio telescope array on the Moon represents a transformative leap for both scientific discovery and technological innovation. The Dark-Ages EXplorer (DEX) concept envisions a large-scale, low-frequency radio array on the lunar surface, capable of conducting groundbreaking observations of the early Universe. Achieving this ambitious goal requires an array of 1000-100,000 antennas, along with novel hardware and software platforms, posing significant engineering challenges. Historically, radio astronomy has been a catalyst for technological progress, and the advancements required for DEX could serve as foundational technologies for a wide range of applications. These innovations aim to enable new scientific discoveries while also supporting a sustainable human presence on the Moon and terrestrial applications. In this paper, we present key technological challenges identified in the recent Concurrent Design Facility (CDF) study, done in collaboration with the European Space Agency (ESA). Technological developments needed to bring DEX to reality have broader applications for future research and commercial activities on the Moon, including energy distribution, autonomous systems, thermal management, communications networks, software development, data management, signal processing, AI/ML and distributed optimisation. By addressing these challenges, we aim to foster cross-sector collaboration and accelerate the development of technologies essential for a sustainable and scientifically productive future on the Moon. Thus, DEX serves not only as an observatory but also as a building block for sustainable lunar exploration and development.
Recent Advances in Autonomous Systems for Inspection and Predictive Maintenance of Infrastructures
An Overview of the Special Session
Autonomous systems are artificial systems capable of performing a variety of tasks with a high degree of autonomy. Cognitive Dynamic Systems (CDSs) are one of the possible approaches that allow us to face the challenges of autonomous systems design. CDSs aim to develop rules of behavior over time through learning from continuous experiential interactions with the surroundings. By exploiting these rules, CDSs can deal with environmental dynamics and uncertainties, and have therefore leveraged the automation of tasks with complex perception-action cycles including surveillance, inspection, predictive maintenance, cognitive radio, traffic control, and robot-mediated industrial and domestic applications. This paper presents an overview of this Special Session, featuring works in the fields of inspection and predictive maintenance of infrastructures, that address challenges associated with autonomy, including perception, decision-making, and adaptation.
...
Autonomous systems are artificial systems capable of performing a variety of tasks with a high degree of autonomy. Cognitive Dynamic Systems (CDSs) are one of the possible approaches that allow us to face the challenges of autonomous systems design. CDSs aim to develop rules of behavior over time through learning from continuous experiential interactions with the surroundings. By exploiting these rules, CDSs can deal with environmental dynamics and uncertainties, and have therefore leveraged the automation of tasks with complex perception-action cycles including surveillance, inspection, predictive maintenance, cognitive radio, traffic control, and robot-mediated industrial and domestic applications. This paper presents an overview of this Special Session, featuring works in the fields of inspection and predictive maintenance of infrastructures, that address challenges associated with autonomy, including perception, decision-making, and adaptation.
This article proposes a novel method of coverage path planning for the purpose of scanning an unstructured environment autonomously. The method uses the morphological skeleton of a prior 2D navigation map via SLAM to generate a sequence of points of interest (POIs). This sequence is then ordered to create an optimal path based on the robot’s current position. To control the high-level operation, a finite state machine (FSM) is used to switch between two modes: navigating toward a POI using Nav2 and scanning the local surroundings. We validate the method in a leveled, indoor, obstacle-free, non-convex environment, evaluating time efficiency and reachability over five trials. The map reader and path planner can quickly process maps of widths and heights ranging between [196,225] pixels and [185,231] pixels in 2.52ms and 1.7ms, respectively. Their computation time increases with 22.0ns/pixel and 8.17 μs/pixel, respectively. The robot managed to reach 86.5% of all waypoints across the five runs. The proposed method suffers from drift occurring in the 2D navigation map.
...
This article proposes a novel method of coverage path planning for the purpose of scanning an unstructured environment autonomously. The method uses the morphological skeleton of a prior 2D navigation map via SLAM to generate a sequence of points of interest (POIs). This sequence is then ordered to create an optimal path based on the robot’s current position. To control the high-level operation, a finite state machine (FSM) is used to switch between two modes: navigating toward a POI using Nav2 and scanning the local surroundings. We validate the method in a leveled, indoor, obstacle-free, non-convex environment, evaluating time efficiency and reachability over five trials. The map reader and path planner can quickly process maps of widths and heights ranging between [196,225] pixels and [185,231] pixels in 2.52ms and 1.7ms, respectively. Their computation time increases with 22.0ns/pixel and 8.17 μs/pixel, respectively. The robot managed to reach 86.5% of all waypoints across the five runs. The proposed method suffers from drift occurring in the 2D navigation map.
Other
(2025)
-
C. D. Brinkerink, M. J. Bentum, A. J. Boonstra, B. Cecconi, L. I. Gurvits, M. Klein-Wolt, J. Lazendic-Galloway, Z. Paragi, R. T. Rajan, More Authors...
Conference paper
(2025)
-
S. Zhao, R. T. Rajan, A. N. Tallarico, M. Millesimo, V. Volosov, A. Imbruglia, J. Dauwels
The accurate prediction of Gallium Nitride High-Electron Mobility Transistors (GaN HEMTs) lifetime is essential for ensuring the reliability of power electronics. However, the complex and often competing degradation mechanisms within a single GaN-based transistor make lifetime extrapolation particularly challenging, especially under limited-data scenarios. In this work, we explore two machine learning approaches, i.e., XGBoost Regression and Gaussian Process Regression (GPR), for static gate lifetime prediction based on early measurements of current and ON-state resistance. In particular, we use features derived from empirical models to improve accuracy and model-specific methods to estimate uncertainty. We compare bootstrapped XGBoost ensembles, which yield empirical confidence intervals, with GPR, which provides analytical uncertainty estimates. Experiments on a time-dependent gate breakdown (TDGB) dataset spanning 16 voltage–temperature combinations show that GPR achieves an SMAPE of 8.8% and ECE of 0.028, outperforming XGBoost in Leave-One-Condition-Out Cross-Validation. These results highlight the feasibility of our proposed uncertainty-aware gate-lifetime prediction for Schottky p-GaN gate HEMTs in small-sample settings, and provide a basis for extending the framework towards time-dependent degradation modeling.
...
The accurate prediction of Gallium Nitride High-Electron Mobility Transistors (GaN HEMTs) lifetime is essential for ensuring the reliability of power electronics. However, the complex and often competing degradation mechanisms within a single GaN-based transistor make lifetime extrapolation particularly challenging, especially under limited-data scenarios. In this work, we explore two machine learning approaches, i.e., XGBoost Regression and Gaussian Process Regression (GPR), for static gate lifetime prediction based on early measurements of current and ON-state resistance. In particular, we use features derived from empirical models to improve accuracy and model-specific methods to estimate uncertainty. We compare bootstrapped XGBoost ensembles, which yield empirical confidence intervals, with GPR, which provides analytical uncertainty estimates. Experiments on a time-dependent gate breakdown (TDGB) dataset spanning 16 voltage–temperature combinations show that GPR achieves an SMAPE of 8.8% and ECE of 0.028, outperforming XGBoost in Leave-One-Condition-Out Cross-Validation. These results highlight the feasibility of our proposed uncertainty-aware gate-lifetime prediction for Schottky p-GaN gate HEMTs in small-sample settings, and provide a basis for extending the framework towards time-dependent degradation modeling.
Intelligent Fault Detection (IFD) has garnered significant attention, with recent advances in AI-empowered predictive maintenance. A key challenge in applying IFD models lies in the interpretability of the methods, since the mechanisms are typically complex and difficult to integrate with data-driven approaches. In addition, the integration of edge devices is an emerging trend, which ensures fault detection and subsequent decision making on the edge, and thus offering an instant response as compared to a conventional centralized server-based architecture. However, to realize Edge-based IFD the primary constraints are low storage capacity and limited computational resources. In this paper, we address various critical challenges in automatic Edge-based IFD for motors in industrial settings, focusing on three key constraints, i.e., (a) limited availability of training data, (b) the lack of method interpretability, and (c) the computational and storage limitations of edge devices. To overcome these challenges, we propose a suite of light weight Physics-Informed (PI) AI algorithms to achieve Edge-based IFD - without compromising detection performance. We validate our proposed methods on experimental data for motor fault detection, and additionally present results from the implementation of these methods on an edge device. We discuss the benefits of our proposed solutions, and give directions for future work.
...
Intelligent Fault Detection (IFD) has garnered significant attention, with recent advances in AI-empowered predictive maintenance. A key challenge in applying IFD models lies in the interpretability of the methods, since the mechanisms are typically complex and difficult to integrate with data-driven approaches. In addition, the integration of edge devices is an emerging trend, which ensures fault detection and subsequent decision making on the edge, and thus offering an instant response as compared to a conventional centralized server-based architecture. However, to realize Edge-based IFD the primary constraints are low storage capacity and limited computational resources. In this paper, we address various critical challenges in automatic Edge-based IFD for motors in industrial settings, focusing on three key constraints, i.e., (a) limited availability of training data, (b) the lack of method interpretability, and (c) the computational and storage limitations of edge devices. To overcome these challenges, we propose a suite of light weight Physics-Informed (PI) AI algorithms to achieve Edge-based IFD - without compromising detection performance. We validate our proposed methods on experimental data for motor fault detection, and additionally present results from the implementation of these methods on an edge device. We discuss the benefits of our proposed solutions, and give directions for future work.
In this paper, we propose a new method for joint ranging and Phase Offset (PO) estimation of multiple transponder-equipped aviation vehicles (TEAVs), including Manned Aerial Vehicles (MAVs) and Unmanned Aerial Vehicles (UAVs). The proposed method employs the overlapping uncoordinated Automatic Dependent Surveillance-Broadcast (ADS-B) packets broadcasted by the TEAVs for joint range and PO estimation prior to ADS-B packet decoding; thus, it can improve air safety when packet decoding is infeasible due to packet collision. Moreover, it enables coherent detection of ADS-B packets, which can result in more reliable multiple target tracking in aviation systems using cooperative sensors for sense and avoid. By minimizing the Kullback-Leibler Divergence (KLD), we show that the received complex baseband signal, coming from K uncoordinated TEAVs, which is corrupted by Additive White Gaussian Noise (AWGN) at a single antenna receiver can be approximated by an independent and identically distributed (i.i.d.) Gaussian Mixture (GM) with 2K mixture components in the two-dimensional plane. The proposed estimator employs the Expectation-Maximization (EM) algorithm to estimate the modes of the 2D Gaussian mixture followed by a reordering estimation technique to jointly estimate range and PO. Simulation results show that the proposed joint estimator outperforms excising methods, such as the time segmentation method and the blind adaptive beamforming.
...
In this paper, we propose a new method for joint ranging and Phase Offset (PO) estimation of multiple transponder-equipped aviation vehicles (TEAVs), including Manned Aerial Vehicles (MAVs) and Unmanned Aerial Vehicles (UAVs). The proposed method employs the overlapping uncoordinated Automatic Dependent Surveillance-Broadcast (ADS-B) packets broadcasted by the TEAVs for joint range and PO estimation prior to ADS-B packet decoding; thus, it can improve air safety when packet decoding is infeasible due to packet collision. Moreover, it enables coherent detection of ADS-B packets, which can result in more reliable multiple target tracking in aviation systems using cooperative sensors for sense and avoid. By minimizing the Kullback-Leibler Divergence (KLD), we show that the received complex baseband signal, coming from K uncoordinated TEAVs, which is corrupted by Additive White Gaussian Noise (AWGN) at a single antenna receiver can be approximated by an independent and identically distributed (i.i.d.) Gaussian Mixture (GM) with 2K mixture components in the two-dimensional plane. The proposed estimator employs the Expectation-Maximization (EM) algorithm to estimate the modes of the 2D Gaussian mixture followed by a reordering estimation technique to jointly estimate range and PO. Simulation results show that the proposed joint estimator outperforms excising methods, such as the time segmentation method and the blind adaptive beamforming.