Multimodal Motion Tracking: A Review

Motion tracking underpins applications in healthcare, sports, surveillance, and smart environments, yet most existing surveys cover only two or three sensing modalities at a time. This paper reviews multimodal motion tracking across all four major modalities (radio frequency, optical, inertial, and magnetic), together with their preprocessing pipelines and fusion strategies. Drawing on more than 300 publications, we analyze fusion at the data, feature, and decision levels and link each choice to deployment constraints such as latency, privacy, and energy use. Three findings stand out. First, a preprocessing impact analysis built from published ablation studies shows that removing calibration or geometric alignment increases tracking error by 103% to 1900%, depending on modality and task. Second, benchmark datasets exhibit consistent gaps in demographic representation, environmental diversity, and synchronization documentation. Third, preprocessing quality and fusion strategy should be co-designed rather than chosen separately, since errors in one propagate into the other. We close with recommendations for next-generation dataset curation and outline open problems in privacy preservation, real-time processing, and cross-domain generalization.