3D INDOOR STATE ESTIMATION FOR RFID-BASED MOTION-CAPTURE SYSTEMS
MetadataShow full item record
The objective of this research is to realize 3D indoor state estimation for RFID-based motion-capture systems. The state estimation is based on sensor fusion by combining RF signal with IMU data together. 3D state-space model of sensor fusion and 3D nonlinear state estimation in NLE with both asynchronous and synchronous models to handle different sensor sampling rates were proposed. For 3D motion with indoor multipath, RMS error before estimation is 71.99 cm, in which 34.99 cm in xy- plane and 62.92 cm along z- axis. After NLE estimation using RF signal combined with IMU data, RMS error of 3D coordinates decreases to 31.90 cm, with 22.50 cm in xy- plane and 22.61 cm along z- axis, achieving a factor of 2 enhancement which is similar to the 2D estimation. In addition, using RF signal only obtains similar estimation results to using both RF and IMU, i.e., 3D RMS error of 31.90 cm, where 22.48 cm in xy- plane and 22.62 cm along z- axis. Hence, RF signal only is able to achieve fine-scale RFID-based motion capture in 3D motion, in consistency with the conclusion arrived at in 2D estimation. In this way, RFID-based motion capture systems can be simplified from embedding inertial sensors. EKF derives close results with 2 cm larger RMS error. In addition, ToF based position sensor in tracking achieves comparable and higher accuracy compared to RSS based position sensor based on the multipath simulation model, enabling ToF to be applied in fine-scale motion capture and tracking.