Multimodal sensor fusion is the process of combining data or features from two or more heterogeneous sensing modalities — such as WiFi CSI, inertial measurement units (IMUs), or camera-based vision systems — into a unified representation for tasks like human activity recognition or indoor localization. It matters for the field because no single modality is universally reliable across all environments, activities, or occlusion conditions, and fusion approaches consistently improve robustness, accuracy, and generalizability beyond what any individual sensor achieves alone. Key variants include early fusion, where raw signals from multiple sensors are combined before feature extraction; feature-level fusion, where independently extracted features are concatenated or weighted prior to classification; and decision-level fusion, where independent model outputs are merged through voting or probabilistic combination.
Source Papers
- Human Activity Recognition via Wi-Fi and Inertial Sensors With Machine Learning ↗ — Human Activity Recognition via Wi-Fi and Inertial Sensors Wi
- OPERAnet, a multimodal activity recognition dataset acquired from radio frequency and vision-based sensors ↗ — OPERAnet, a multimodal activity recognition dataset acquired
- Occupancy Prediction in IoT-Enabled Smart Buildings: Technologies, Methods, and Future Directions ↗ — Occupancy Prediction in IoT-Enabled Smart Buildings: Technol