Research
Publications
729 publications
- 2026
Sensitivity of Inter-Point Distance Measurement to Control-Point Coordinate Error in a Panoramic Camera(opens in a new window)
Sensors, Vol. 26 (19), p. 6300, MDPI
Abstract
Vision-based traffic monitoring recovers metric quantities such as inter-vehicle distances by mapping the image to the ground plane through a planar homography. In deployed systems, this homography is often fitted from ground control points whose world coordinates are obtained from satellite orthoimagery, where each coordinate is uncertain by about one image pixel. The impact of this control-point coordinate error on the measured distance has not been quantified for a single fixed camera. This paper evaluates a 180-degree panoramic camera (AXIS P3807-PVE) on two adjacent ground-plane regions, each referenced to RTK-surveyed control points and mapped by a separate homography. Control-point coordinates are perturbed with isotropic Gaussian noise, the homography is refitted, and the resulting change in inter-point distance between held-out evaluation points is measured against the unperturbed fit. The distance error is dominated by random scatter rather than systematic bias, and the scatter grows approximately linearly with the coordinate noise, at about 0.7 to 0.8 times the per-coordinate standard deviation within the tested range. The results indicate that distance error can be predicted from control-point coordinate accuracy, and that relative precision of those coordinates matters more than absolute georeferencing.
Scopus percentile: 91
- 2026
ConMonity: An IoT-Enabled LoRa/LTE-M Platform for Multimodal, Real-Time Monitoring of Concrete Curing in Construction Environments(opens in a new window)
Namatēvs, I., Gaigals, G., & Ozols, K.
Sensors, Vol. 26 (1), p. 14, MDPI
Abstract
Monitoring the curing process of concrete remains a challenging and critical aspect of modern construction, often hindered by labour-intensive, invasive, and inflexible methods. The primary aim of this study is to develop an integrated IoT-enabled platform for automated, real-time monitoring of concrete curing, using a combination of LoRa-based sensor networks and an LTE-M backhaul. The resulting ConMonity system employs embedded multi-sensor nodes—capable of measuring strain, temperature, and humidity‒connected via an energy-efficient, TDMA-based LoRa wireless protocol to an LTE-M gateway with cloud-based management and analytics. By employing a robust architecture with battery-powered embedded nodes and adaptive firmware, ConMonity enables multi-modal, multi-site assessments and demonstrates stable, autonomous operation over multi-modal, multi-site assessment and demonstrates stable, autonomous operation over multi-month field deployments. Measured data are transmitted in a compact binary MQTT format, optimising cellular bandwidth and allowing secure, remote access via a dedicated mobile application. Operation in laboratory construction environments indicates that ConMonity outperforms conventional and earlier wireless monitoring systems in scalability and automation, delivering actionable real-time data and proactive alerts. The platform establishes a foundation for intelligent, scalable, and cost-effective monitoring of concrete curing, with future work focused on extending sensor modalities and enhancing resilience under diverse site conditions.
Indexed in Scopus / WoS · Scopus percentile: 88
- 2026
Ultrasonic Monitoring of the Processes of Blast Freezing and Thawing of Meat(opens in a new window)
Alexey Tatarinov 1 , Marija Osipova and Viktors Mironov
Foods, Vol. 15, p. 328
Abstract
Freezing and thawing affect the structural integrity and quality of meat, yet these processes remain difficult to monitor due to spatial temperature gradients and non-uniform phase transitions. This study investigates the ability of ultrasound to detect dynamic freezing and thawing events in pork tissues with different fat contents. Specimens of water, lean meat, marbled meat, layered lean–fat structures, and lard were subjected to controlled freeze–thaw cycles while ultrasonic signals and internal temperatures were continuously monitored. Consistent amplitude drops in the megahertz range at entering the freezing phase formed characteristic signal patterns that differed sharply between lean and fatty tissues. Lean meat, dominated by water content, exhibited rapid signal loss at the onset of ice crystallization and a clear recovery of amplitude once fully frozen. Fat-rich tissues demonstrated prolonged attenuation and near disappearance of high-frequency signals, with incomplete recovery even at deep-frozen states. A jump of ultrasound velocity from 1.4–1.6 km/s in a warm state to 2.6–3.7 km/s in a frozen state indicated complete freezing. Hysteresis between temperature readings and actual phase transition moments was found. Distinct ultrasonic freeze–thaw signatures reflecting tissue composition suggest a novel approach for monitoring the true onset and completion of freezing and thawing in meat.
- 2026
Differentially Private Federated Learning for Remaining Useful Life Prediction(opens in a new window)
Arturs Nikulins, Kārlis Freivalds, Ivars Namatēvs, Kaspars Sudars, Audris Arzovs, Wilhelm Söderkvist Vermelin, Madhav Mishra, Kaspars Ozols.
Applied Sciences, Vol. 16 (6), p. 2784 (raksta numurs), MDPI
Abstract
Accurate remaining useful life (RUL) prediction is essential for the safe and cost-effective operation of safety-critical systems such as electronic components and engines. While data-driven machine learning approaches have demonstrated strong performance for RUL estimation, their effectiveness is limited by the lack of full run-to-failure data and by strict privacy and intellectual property constraints in industrial settings. Federated learning (FL) enables collaborative model training across multiple data owners without direct data sharing, but it does not, by itself, provide formal privacy guarantees and remains vulnerable to information leakage. This paper presents a privacy-preserving DP-enhanced FL setup for RUL prediction that combines federated learning with differential privacy (DP). We describe an end-to-end implementation based on the Opacus DP library, highlight practical challenges arising from the integration of DP into recurrent neural network architectures, and propose solutions to address them. Using two representative RUL datasets (CMAPSS and SiC MOSFET), we analyze the effect of DP noise on prediction performance and on the functional dependence between the predicted RUL and the already lived life feature. The results demonstrate that differential privacy can be integrated into federated RUL prediction with limited degradation in predictive performance, providing practical insights for deploying privacy-aware collaborative models in industrial environments.
Indexed in Scopus / WoS · Scopus percentile: 79 · SNIP: 0.956
- 2026
Investigation of Audio Feature Application for CO2 Sensor-Based Occupancy Detection Enhancement(opens in a new window)
Marija Skromule, Rainers Kozlovskis, Deniss Tiscenko, Janis Judvaitis
Buildings, MDPI
Abstract
This study investigates the integration of audio features with CO2 sensor data to enhance occupancy detection accuracy in naturally ventilated office environments. Accurate occupancy detection is pivotal for smart building energy management, yet CO2-based methods cannot provide fast enough response times and are sensitive to air circulation changes due to internal convection. In this article we propose a combination of CO2 sensors and audio features from MEMS microphones to improve the occupancy detection accuracy and improve the response times. We use a Random Forest classifier and evaluate the results across two scenarios: CO2-only and CO2 combined with audio features. Results show that incorporating the audio features into the occupancy detection algorithms yields a significant increase in detection accuracy and speed, especially when the environment is subject to frequent air circulation changes due to internal convection, like the opening and closing of windows and doors. Combining the CO2 and audio sensing offers a promising, cost-effective approach to occupancy detection in smart buildings, yet more research on advanced audio processing and feature selection is necessary.
- 2026
Ambient Temperature Impact on the Thermal Behavior and Power Consumption of the NVIDIA Jetson AGX Orin in an Outdoor Enclosure(opens in a new window)
Rihards Krišlauks, Deniss Tiscenko, Vladislavs Medvedevs, Juris Ormanis, Janis Judvaitis.
Electronics, Vol. 15 (11), MDPI
Details
Indexed in Scopus / WoS · Scopus percentile: 83
- 2025
ShapeFuture - Technical Progress After Year 1(opens in a new window)
Druml, N., Gschwandtner, M., Jeannin, M., Matischek, R., Lielamurs, E., Celitans, M., Ozols, K., Demiralay, N., Tayfur, B., Gulbas, I. S., Kucuk, N., Kiyat, I., Nasolo, Y., Brandt, J., Pütz, N. C., Bartz-Beielstein, T., Isola, J., Mandic, N., Flamigni, F., Kuehhas, A., Brilli, G., Burgio, P., Paolieri, G., Villagra, J., Cueto, Á. F., Sánchez, J. A., Sini, J., Violante, M., Giraudi, L., Santero, P., Kölbel, U., Schaffenroth, M., Sjövall, P., Vanne, J., Larsen, M., Yilmaz, N. G., Yengin, Z. U., Dimitrakopoulos, G.
28th Euromicro Conference on Digital System Design (DSD), p. 350-359
Abstract
ShapeFuture will drive innovation in fundamental Electronic Components and Systems (ECS) that are essential for robust, powerful, fail-operational and integrated perception, cognition, AI-enabled decision making, resilient automation and computing, as well as communications, for highly automated vehicles. The overarching vision of ShapeFuture is to bring ECS Innovation to the heart of Europe’s Mobility Transformation, thereby elevating sovereignty by perfecting programmable ECS solutions for intelligent, safe, connected, and highly automated vehicles. In this paper, we detail not only the vision and mission of the ShapeFuture project, but we also showcase the results achieved during the first year.
Indexed in Scopus / WoS
- 2025
Advancing Energy Citizenship: Hindering and Supporting Factors in Latvia’s Energy Transition(opens in a new window)
Lagzdina E., Brizga J., Kudrenickis I., Ikstena R., Ernsteins R.
Section in collective monography “Energy Citizenship Across Europe: Contexts and Conditions for an Emerging Energy Transition” (eds. F.Fahy and E.Vadovics)
- 2025
Economic Assessment of Self-Consumption Rate in the Context of Energy Communities: A Latvian Case Study(opens in a new window)
L. Petrichenko, R. Lazdins, A. Mutule, J. Teremranova
2025 21st International Conference on the European Energy Market (EEM), p. 1-5
- 2025
Utilisation of Deep Neural Networks for Estimation of Cajal Cells in the Anal Canal Wall of Patients with Advanced Haemorrhoidal Disease Treated by LigaSure Surgery(opens in a new window)
Fišere, I.; Edelmers, E.; Svirskis, Š.; Groma, V.
Cells 2025, Vol. 14 (7), p. 550, MDPI
Abstract
Interstitial cells of Cajal (ICCs) play a key role in gastrointestinal smooth muscle contractions, but their relationship with anal canal function in advanced haemorrhoidal disease (HD) remains poorly understood. This study uses deep neural network (DNN) models to estimate ICC presence and quantity in anal canal tissues affected by HD. Haemorrhoidectomy specimens were collected from patients undergoing surgery with the LigaSure device. A YOLOv11-based machine learning model, trained on 376 immunohistochemical images, automated ICC detection using the CD117 marker, achieving a mean average precision (mAP50) of 92%, with a recall of 86% and precision of 88%. The DNN model accurately identified ICCs in whole-slide images, revealing that one-third of grade III HD patients and 60% of grade IV HD patients had a high ICC density. Preoperatively, pain was reported in 35% of grade III HD patients and 41% of grade IV patients, with a significant reduction following surgery. A significant decrease in bleeding (p < 0.0001) was also noted postoperatively. Notably, patients with postoperative bleeding, diagnosed with stage IV HD, had high ICC density in their anorectal tissues (p = 0.0041), suggesting a potential link between ICC density and HD severity. This AI-driven model, alongside clinical data, may enhance outcome prediction and provide insights into HD pathophysiology.
Indexed in Scopus / WoS · Scopus percentile: 84
- 2025
An Efficient Communication Protocol for Real-Time Body Sensor Data Acquisition and Feedback in Interactive Wearable Systems(opens in a new window)
Armands Ancans, Modris Greitans, Sandis Kagis
Journal of Sensor and Actuator Networks, Vol. 14 (1), MDPI
Abstract
We introduce a novel wired communication approach for interactive wearable systems, employing a single signal wire and innovative group addressing protocol to reduce overhead. While wireless solutions dominate body sensor networks, wired approaches offer advantages for interactive applications that require low latency, high reliability, and communication with high-density nodes; yet they have been less explored in the context of wearable systems. Many commercial products use wired connections without disclosing technical details, limiting broader adoption. To address this gap, we present and test a new group addressing protocol implemented using Universal Asynchronous Receiver–Transmitter (UART) hardware, disclosing frame diagrams and node architectures. We developed a prototype interactive jacket with nine sensor/actuator nodes connected via three wires for power supply and data transmission to a wireless gateway. Mathematical analysis showed an overhead reduction of approximately 50% compared to traditional individual addressing. Our solution is the most wire-efficient among wired interactive wearable systems reviewed in the literature, using only one signal wire; other methods require at least two wires and often have overlapping topologies. Performance experimental evaluation revealed a total feedback delay of 2.27 ms and a maximum data frame rate of 435.4 Hz, comparable to the best-performing products and leaving room for twice the performance calculated theoretically. These results indicate that the proposed approach is suitable for interactive wearable systems, both for real-time applications and high-resolution data acquisition.
Indexed in Scopus / WoS · Scopus percentile: 93
- 2025
Privacy Auditing in Differential Private Machine Learning: The Current Trends(opens in a new window)
Ivars Namatevs, Kaspars Sudars, Arturs Nikulins, Kaspars Ozols
Applied Sciences, Vol. 15 (2), p. 647, MDPI
Abstract
Differential privacy has recently gained prominence, especially in the context of private machine learning. While the definition of differential privacy makes it possible to provably limit the amount of information leaked by an algorithm, practical implementations of differentially private algorithms often contain subtle vulnerabilities. Therefore, there is a need for effective methods that can audit (ϵ,δ) differentially private algorithms before they are deployed in the real world. The article examines studies that recommend privacy guarantees for differential private machine learning. It covers a wide range of topics on the subject and provides comprehensive guidance for privacy auditing schemes based on privacy attacks to protect machine-learning models from privacy leakage. Our results contribute to the growing literature on differential privacy in the realm of privacy auditing and beyond and pave the way for future research in the field of privacy-preserving models.
Indexed in Scopus / WoS · Scopus percentile: 75
- 2025
Adapting Classification Neural Network Architectures for Medical Image Segmentation Using Explainable AI(opens in a new window)
Arturs Nikulins, Edgars Edelmers, Kaspars Sudars, Inese Polaka
Journal of Imaging, Vol. 11 (2), p. 55, MDPI
Abstract
Segmentation neural networks are widely used in medical imaging to identify anomalies that may impact patient health. Despite their effectiveness, these networks face significant challenges, including the need for extensive annotated patient data, time-consuming manual segmentation processes and restricted data access due to privacy concerns. In contrast, classification neural networks, similar to segmentation neural networks, capture essential parameters for identifying objects during training. This paper leverages this characteristic, combined with explainable artificial intelligence (XAI) techniques, to address the challenges of segmentation. By adapting classification neural networks for segmentation tasks, the proposed approach reduces dependency on manual segmentation. To demonstrate this concept, the Medical Segmentation Decathlon ‘Brain Tumours’ dataset was utilised. A ResNet classification neural network was trained, and XAI tools were applied to generate segmentation-like outputs. Our findings reveal that GuidedBackprop is among the most efficient and effective methods, producing heatmaps that closely resemble segmentation masks by accurately highlighting the entirety of the target object.
Indexed in Scopus / WoS · Scopus percentile: 81
- 2025
Virtual Reality and Gamification in Rehabilitation: Enhancing Patient Engagement in the Buzz-Wire Maze Task(opens in a new window)
Anete Savicka, Mārcis Kalniņš, Maksims Ivanovs
2025 IEEE 12th Workshop on Advances in Information, Electronic and Electrical Engineering, AIEEE 2025 - Proceedings, IEEE
Abstract
Virtual reality (VR) has recently emerged as a promising tool in rehabilitation, providing an interactive environment that can improve the motivation and engagement of patients. The potential of VR-based rehabilitation can be further expanded by means of gamification - the integration of game-like elements into non-gaming contexts. This study investigates the impact of VR and gamification on performance in a handheld buzz-wire maze task, which can be used in post-stroke rehabilitation. Participants (N=27) performed the task by guiding a tracking ring along a wire and trying to avoid contact, across three conditions: traditional mirror therapy, VR without gamification, and VR with gamification. Results showed significant improvements in performance in terms of decreased time and number of mistakes over repeated trials in all conditions. However, gamification did not produce consistent short-term gains in speed or accuracy. Although participants performed comparably in VR and the real-world condition, error rates (number of touches) remained higher in VR. Subjective feedback indicated mixed preferences, with many participants finding VR more engaging but the real environment easier to control. Overall, these findings support the potential of VR and gamification for motor training, while also highlighting areas for further refinement.
Indexed in Scopus / WoS
- 2025
A Distributed Time-of-Flight Sensor System for Autonomous Vehicles: Architecture, Sensor Fusion, and Spiking Neural Network Perception(opens in a new window)
Edgars Lielamurs, Ibrahim Sayed, Andrejs Cvetkovs, Rihards Novickis, Anatolijs Zencovs, Maksis Celitans, Andis Bizuns, George Dimitrakopoulos, Jochen Koszescha, Kaspars Ozols
Electronics, Vol. 14 (7), MDPI
Abstract
Mechanically scanning LiDAR imaging sensors are abundantly used in applications ranging from basic safety assistance to high-level automated driving, offering excellent spatial resolution and full surround-view coverage in most scenarios. However, their complex optomechanical structure introduces limitations, namely limited mounting options and blind zones, especially in elongated vehicles. To mitigate these challenges, we propose a distributed Time-of-Flight (ToF) sensor system with a flexible hardware–software architecture designed for multi-sensor synchronous triggering and fusion. We formalize the sensor triggering, interference mitigation scheme, data aggregation and fusion procedures and highlight challenges in achieving accurate global registration with current state-of-the-art methods. The resulting surround view visual information is then applied to Spiking Neural Network (SNN)-based object detection and probabilistic occupancy grid mapping (OGM) for enhanced environmental awareness. The proposed system is demonstrated on a test vehicle, achieving coverage of blind zones in a range of 0.5–6 m with a scalable and reconfigurable sensor mounting setup. Using seven ToF sensors, we can achieve a 10 Hz synchronized frame rate, with a 360° point cloud registration and fusion latency below 40 ms. We collected real-world driving data to evaluate the system, achieving 65% mean Average Precision (mAP) in object detection with our SNN. Overall, this work presents a replacement or addition to LiDAR in future high-level automation tasks, offering improved coverage and system integration.
Indexed in Scopus / WoS · Scopus percentile: 60
- 2025
Multi-Step Object Re-Identification on Edge Devices: A Pipeline for Vehicle Re-Identification
Tomass Zutis, Peteris Racinskis, Anzelika Bureka, Janis Judvaitis, Janis Arents, and Modris Greitans
Charting the Intelligence Frontiers – Edge AI Systems Nexus, p. 233-253, River Publishers
Abstract
In modern computer vision tasks, the ability to identify and track objects across different scenes and environments has become important for numerous applications, especially in transportation. Inspired by this need, we propose a method that leverages a multi-step process focused on extracting and using object features for object re-identification. The proposed pipeline includes the following steps: detecting an object, converting its features into a vector embedding, storing this embedding in a vector database, and then querying the database to find the same or similar objects based on their feature embeddings. This approach enables us to identify the same object across different images or cameras, even in varying locations. This is essential in scenarios like Vehicle Re-Identification. For such scenario, implementing this process on edge devices is crucial. Therefore, ways to tailor the pipeline and its outputs for edge devices are outlined. The paper details the pipeline’s structure along with the experimental setup demonstrating its application, particularly in vehicle re-identification. The pipeline achieves 70-80% re-identification precision when dealing with vehicle images from our network cameras and above 70% Rank-1 accuracy when dealing with a CityFlow video track scenario.
- 2025
Privacy Auditing of Lithium-Ion Battery Ageing Model by Recovering Time-Series Data Using Gradient Inversion Attack in Federated Learning(opens in a new window)
Sudars, K., Namatevs, I., Nikulins, A., & Ozols, K.
Applied Sciences, Vol. 15 (10), p. 5704, MDPI
Abstract
The exchange of gradients is a widely used method in modelling systems for machine learning (e.g., distributed training, federated learning) in privacy-sensitive domains. Unfortunately, there are still privacy risks in federated learning, as servers can reconstruct clients‘ training data through gradient inversion attacks. To protect against such attacks, we need to know when and how privacy is being undermined, largely due to the black box nature of deep neural networks. Although gradient inversion has been used to classify images and text data, its use in time-series is still largely unexplored. In this paper, we empirically evaluate the practicality of recovering lithium-ion battery time-series data from the gradients of a transformer-based model, both without and with differential privacy, in a time-series federated learning framework. It is especially significant in the case of electric vehicles (EVs). As shown in this paper, additional protection by differential privacy leads to the saturation of gradient inversion attacks, i.e., the reconstructed signal maintains a certain error level, depending on the applied privacy budget level. With this empirical evaluation, we provide insights into effective gradient perturbation directions, unfairness with respect to privacy, and privacy-preferred model initialisation.
Indexed in Scopus / WoS · Scopus percentile: 79
- 2025
Motor Imagery EEG Classification with Real-World and EEG-GAN-Generated Data(opens in a new window)
Duplevska Diana, Savicka Anete, Ivanovs Maksims
2025 IEEE 12th Workshop on Advances in Information, Electronic and Electrical Engineering, AIEEE 2025 - Proceedings, IEEE
Abstract
Accurate classification of EEG signals is crucial for applications such as brain-computer interfaces (BCI). In this study, we examined the effectiveness of long short-term memory (LSTM) networks for classifying EEG signals recorded during six different motor imagery tasks. The dataset included recordings from 16 electrodes and consisted of actions such as forearm supination, elbow flexion, forearm pronation, hand closing, hand opening, and elbow extension. Our model achieved an average F1 score of 0.87 on the real-world EEG data. To evaluate the feasibility of training the model on similar synthetic data, we generated synthetic EEG signals for hand opening and closing tasks using EEG-GAN. When tested on synthetic data, the LSTM model trained solely on real data achieved the average F1 score of 0.76. These results highlight the potential of generative models for augmenting EEG datasets and the robustness of LSTM-based classifiers.
Indexed in Scopus / WoS
- 2025
LPWAN Technologies for IoT: Real-World Deployment Performance and Practical Comparison(opens in a new window)
Dmitrijs Orlovs, Artis Rušiņš, Valters Skrastiņš, Janis Judvaitis
IoT, Vol. 6 (4), MDPI
Abstract
Low Power Wide Area Networks (LPWAN) have emerged as essential connectivity solutions for the Internet of Things (IoT), addressing requirements for long range, energy efficient communication that traditional wireless technologies cannot meet. With LPWAN connections projected to grow at 26% compound annual growth rate until 2027, understanding real-world performance is crucial for technology selection. This review examines four leading LPWAN technologies—LoRaWAN, Sigfox, Narrowband IoT (NB-IoT), and LTE-M. This review analyzes 20 peer reviewed studies from 2015–2025 reporting real-world deployment metrics across power consumption, range, data rate, scalability, availability, and security. Across these studies, practical performance diverges from vendor specifications. In the cited rural and urban LoRaWAN deployments LoRaWAN achieves 2+ year battery life and 11 km rural range but suffers collision limitations above 1000 devices per gateway. Sigfox demonstrates exceptional range (280 km record) with minimal power consumption but remains constrained by 12 byte payloads and security vulnerabilities. NB-IoT provides robust performance with 96–100% packet delivery ratios at −127 dBm on the tested commercial networks, and supports tens of thousands devices per cell, though mobility increases energy consumption. In the cited trials LTE-M offers highest throughput and sub 200 ms latency but fails beyond −113 dBm where NB-IoT maintains connectivity. NB-IoT emerges optimal for large scale stationary deployments, while LTE-M suits high throughput mobile applications.
Indexed in Scopus / WoS
- 2025
Short-Term Wind Energy Yield Forecasting: A Comparative Analysis Using Multiple Data Sources
Dmitrijevs, N.; Komasilovs, V.; Orlova, S.; Kamolins
Energies 2025