Łukasik Szymon, Szott Szymon, Leszczuk Mikołaj
Systems Research Institute Polish Academy of Sciences, 01-447 Warszawa, Poland.
AGH University of Krakow, 30-059 Krakow, Poland.
Sensors (Basel). 2024 Sep 19;24(18):6051. doi: 10.3390/s24186051.
Outdoor positioning has become a ubiquitous technology, leading to the proliferation of many location-based services such as automotive navigation and asset tracking. Meanwhile, indoor positioning is an emerging technology with many potential applications. Researchers are continuously working towards improving its accuracy, and one general approach to achieve this goal includes using machine learning to combine input data from multiple available sources, such as camera imagery. For this active research area, we conduct a systematic literature review and identify around 40 relevant research papers. We analyze contributions describing indoor positioning methods based on multimodal data, which involves combinations of images with motion sensors, radio interfaces, and LiDARs. The conducted survey allows us to draw conclusions regarding the open research areas and outline the potential future evolution of multimodal indoor positioning.
室外定位已成为一项无处不在的技术，催生了许多基于位置的服务，如汽车导航和资产追踪。与此同时，室内定位是一项具有众多潜在应用的新兴技术。研究人员不断致力于提高其精度，实现这一目标的一种通用方法包括使用机器学习来整合来自多个可用数据源（如摄像头图像）的输入数据。针对这个活跃的研究领域，我们进行了系统的文献综述，并识别出约40篇相关研究论文。我们分析了基于多模态数据描述室内定位方法的贡献，多模态数据涉及图像与运动传感器、无线电接口和激光雷达的组合。所进行的调查使我们能够就开放的研究领域得出结论，并概述多模态室内定位的潜在未来发展。
