3 June 2025 in IEEE Sensors Journal
Frequency-modulated Continuous-wave (FMCW) Radar is a promising device for hand gesture reconstruction in autonomous control systems. A few groundbreaking studies have achieved remarkable advancements in automated hand gesture recognition by combining FMCW radar systems with deep learning (DL) techniques. However, one limitation of these DL models is their inability to convey the uncertainty linked to the model’s predictions, which holds great significance in the context of autonomous control systems. This paper introduces the solution to address the challenge by developing an uncertainty-aware deep convolutional neural network. The network utilizes a CNN model that incorporates advanced techniques including Monte Carlo Dropout, Deep Ensemble Learning, and Spectral-normalized Neural Gaussian Process. Our novel network architecture aims to predict hand gestures efficiently while providing an estimation of the associated uncertainty in the model’s predictions. The proposed approach was evaluated on a dataset with ten gestures collected from ten volunteers. The model could predict gestures with an accuracy of over 99 % which is superior and noise-resistant to the existing deterministic models. The dataset was available at https://github.com/thetuantrinh/Hand-Gesture-Recognition.git.
27 April 2026 in IEEE Sensors Journal
Real-time hand pose estimation is essential for natural human-computer interaction (HCI), especially in situations where vision-based systems face challenges due to lighting, occlusion, or privacy issues. This paper introduces RadarPose, an innovative framework that estimates 2D hand keypoints for HCI applications using raw Frequency-Modulated Continuous Wave (FMCW) radar signals. To address the difficulty of manually annotating FMCW radar data, a cross-modal supervision strategy is implemented, utilizing synchronized RGB video frames processed by a pre-trained vision model to generate hand keypoint labels for training the radar signals. RadarNet, the core of this framework, is a lightweight convolutional neural network (L-CNN) designed to predict 21 hand keypoints, handedness, and hand presence directly from a raw data cube. By removing the Fast Fourier Transform (FFT) stages, this framework facilitates end-to-end learning in the radar complex data domain, resulting in significantly reduced latency. Once trained, the proposed framework relies solely on radar data for inference, ensuring both privacy protection and invariance to lighting conditions. The model comprises only 326k parameters and achieves real-time inference in just 1.19 milliseconds on an NVIDIA Jetson Nano. Experimental results indicate that RadarPose demonstrates competitive accuracy compared to vision-based models in well-lit environments and significantly outperforms them in low-light or occluded scenarios, highlighting its potential for robust, edge-compatible hand pose estimation. The code and dataset were available at https://github.com/thetuantrinh/UWB-Radar-Hand-Pose-Estimation.git.
1 March 2026 in Engineering Research Express
Nondestructive Testing (NDT) is an essential methodology for ensuring the structural integrity and safety of critical engineering structures such as nuclear power plans, railways, and aerospaces. Leveraging advanced signal analysis and Artificial Intelligence (AI) is necessary to enhance detection capabilities, but the development of bespoke AI models presents a significant bottleneck, requiring deep, cross-disciplinary expertise, and extensive manual effort. To address this challenge, we introduce the AI Co-Scientist, a novel framework designed for human-AI cooperation throughout the entire model development lifecycle. Within this human-in-the-loop system, AI agents and human experts work in partnership to perform data analysis, iterative AI model development, and rigorous model evaluation.The effectiveness of this collaborative framework was rigorously verified by applying it to a real-world Gaussian-Pulsed Eddy Current Testing (GPECT) system for the task of detecting corrosion in an aluminum specimen. The framework autonomously performs signal analysis and converges on a high-performance model. The proposed approach achieved a recall of 99.6%, an F1-score of 95.1%, and an AUC-ROC of 99.9% on a highly imbalanced real-world GPECT dataset, demonstrating its suitability for safety-critical corrosion detection. This result validates our approach, demonstrating that the AI Co-Scientist framework can significantly accelerate research and empower domain experts to focus on high-level strategic oversight.
1 November 2026 in IEEE Sensors Journal