This study introduces a simple and cost-effective localization system using a moving ultra-wideband (UWB) radar sensor and passive reflectors at fixed points. Our use of UWB radar ensures consistent performance across various lighting conditions and offers privacy protection. Our hybrid pipeline first predicts the ranges from the sensor to the reflectors, and then predicts the radar’s position from these ranges. Two key components of the hybrid pipeline are a neural network for range prediction and an optimization-based “association” step. The neural network solves the challenge of predicting individual ranges from the mixed radar signal, while the association step matches each predicted range to its corresponding reflector using a novel regularized trilateration formulation. Experiments validated our approach, yielding an average positional estimation error of approximately 0.20 m, making it suitable for human or robot tracking applications. Our contributions include a novel localization setup, an algorithm, and a real-world UWB dataset with annotated 3D ground-truth positions.

We present a deep learning-based approach called DipSAR for reconstructing millimeter-wave synthetic aperture radar (SAR) images from sparse samples. The primary challenge lies in the requirement of a large training dataset for deep learning schemes. To overcome this issue, we employ the deep image prior (DIP) technique, which eliminates the need for a large dataset and instead utilizes only the sparse sample itself. Our proposed DipSAR model recovers missing samples from sparse data and reconstructs the SAR image using a conventional method. In this study, we utilize an existing SAR dataset and create fourteen different patterns to generate additional sparse samples by removing certain data points. We then evaluate the performance of DipSAR in comparison to the conventional method. The results show that DipSAR outperforms the conventional method in terms of the intersection over union (IoU) score.