An efficient wireless sensor network for real-time multiuser motion capture system

Abstract

A real-time multiuser motion capture system has been developed using an efficient network of compact wireless MARG (Magnetic, Angular Rate, and Gravity) sensor nodes worn by the users. The motion capture system is designed to provide a completely wireless 9 degrees of freedom (DOF) tracking for multiuser at a high-speed update rate. In order to meet the system requirements, this paper presents an innovative wireless sensor network, consisting of time synchronization mechanism, pipeline data aggregation method and local orientation estimation to guarantee the accuracy and timeliness of the multiple MARG sensor nodes data transmission. Adopting this wireless sensor network, the multiuser motion capture system could support 48 sensor nodes at 30Hz sampling frequency or 24 sensor nodes at 60/120Hz, which means the system was able to track the movement of 4 or 2 people in low or high frequency mode at the same time. Experimental results show that the wireless sensor network packet loss rate is under 1% and the local estimation of MARG sensor orientation using a gradient descent algorithm gets a good accuracy performance: <1.0° static RMS error, <2.0° dynamic RMS error.