Achieving High Precision RFID Trajectory Tracing with Kinematic Features
Liang Zhang, Caidong Gu, Zhaobin Liu and Kan Lu
In applications such as gaming and gesture-based interactions, high precision tracking of an object’s trajectory can help rebuild the trajectory shape and to interpret the meaning of a human gesture. State-of-the-art systems can achieve an accuracy on the order of centimeters or millimeters. However, high accuracy schemes are usually more sensitive to noise. Furthermore, most trajectory tracing systems suffer data error or packet loss when an object is moving at relatively high speed. This work reports the achievement of high precision trajectory tracing via the adoption of kinematic features. The key idea of our approach is leveraging the kinematic equation to utilize the motion pattern generated by a continuous movement of the RFID tag for trajectory tracing. As such, the motion pattern can be applied to reconstructing the trajectory of a noisy data set while withstanding data error or packet loss. Linear geometry and linear complex are employed in a pattern recognition method to distinguish motion types and obtain geometric parameters. Analysis of the geometric parameters can help achieve efficient and high-precision reconstruction of the trajectory. A prototype was built with commercial off-the-shelf RFID readers and tags, and was used to obtain geometric characteristics for reconstructing the trajectory of moving objects. We evaluated the performance of our scheme with extensive experiments. In a lab environment, high precision geometric features of the trajectory shape of an RF source with the moving speed of up to 0.4m/s were obtained. In a field study, linear and arc-shaped gestures were used to simulate human-computer interaction commands. The results showed adequate recognition of gestures and only minor impact on the system by distortion or the integrity of the shape of human gestures.
Keywords: RFID; tracking; geometric feature; kinematic; linear complex; gesture
recognition