Ratiu Vlad, Ratiu Ovidiu, Smeyers Olivier Raphael, Dadarlat Vasile Teodor, Vos Stefan, Rednic Ana
Computer Science Department, Technical University of Cluj-Napoca, 400114 Cluj-Napoca, Romania.
Control Data Systems SRL, 400267 Cluj-Napoca, Romania.
Sensors (Basel). 2025 Jul 25;25(15):4620. doi: 10.3390/s25154620.
Real-time positioning is a technological field with a multitude of applications, which expand across many scopes: from positioning within a large area to localization within smaller spaces; from locating people to locating equipment; from large-scale industrial or military applications to commercially available solutions. There are at least as many implementations of real-time positioning as there are applications and challenges. Within the domain of Radio Frequency (RF) systems, positioning has been approached from multiple angles. Some of the more common solutions involve using Time of Flight (ToF) and time difference of arrival (TDoA) technologies. Within TDoA-based systems, one common limitation stems from the computational power necessary to run the multi-lateration algorithms at a high enough speed to provide high-frequency refresh rates on the tag positions. The system presented in this study implements a complete hardware and software TDoA-based real-time positioning system, using wireless Ultra-Wideband (UWB) technology. This system demonstrates improvements in the state of the art by addressing the above limitations through the use of a hybrid Machine Learning solution combined with algorithmic fine tuning in order to reduce computational power while achieving the desired positioning accuracy. This study presents the design, implementation, verification and validation of the aforementioned system, as well as an overview of similar solutions.
实时定位是一个具有众多应用的技术领域,其应用范围广泛:从大面积内的定位到较小空间内的定位;从人员定位到设备定位;从大规模工业或军事应用到商用解决方案。实时定位的实现方式至少与应用和挑战的数量一样多。在射频(RF)系统领域,定位已经从多个角度进行了探讨。一些更常见的解决方案涉及使用飞行时间(ToF)和到达时间差(TDoA)技术。在基于TDoA的系统中,一个常见的限制源于以足够高的速度运行多边测量算法以在标签位置上提供高频刷新率所需的计算能力。本研究中提出的系统使用无线超宽带(UWB)技术实现了一个完整的基于硬件和软件的TDoA实时定位系统。该系统通过使用混合机器学习解决方案并结合算法微调来解决上述限制,从而在降低计算能力的同时实现所需的定位精度,展示了在现有技术水平上的改进。本研究介绍了上述系统的设计、实现、验证和确认,以及类似解决方案的概述。