Mobile human pose estimation with wireless sensors

Abstract

Motion capturing is widely used across different applications such as medical analysis and creating animations. Inertial measurement units provide good motion capture data for a relatively low price and high mobility for the user. Recently, some completely wireless solutions have entered the market. However, these solutions rely on Wi-Fi or custom radio protocol networks, limiting the devices that can communicate with the suits. Utilising Bluetooth, a common standard for short range wireless communications, a motion capture suit can communicate directly withany device, reducing latency while increasing usability. However, the feasibility of Bluetooth for motion capture has yet to be analysed.

The necessary mathematics and theory for motion capture using Bluetooth communications is reviewed in order to design such a system, and the basic functional principalities of an inertial measurement unit is presented in order to understand the shortcomings of the sensor technology. The feasibility of Bluetooth for motion capturing is finally analysed by designing and testing an implementation of such a system, without making any changes to the default drivers of the devices or the operating systems. The designed system achieved similar or better performance when compared to the currently existing suits on the market, although, the receiver device was also identified as the bottleneck of the system. The amount of connected devices and the possible sampling rate are highly dependant on the Bluetooth radio of the receiver device.

It is proven that the technology is feasible for a motion capture suit that offers many integration possibilities beyond the commercial suits currently on the market, while having a much lower cost for the hardware. Furthermore, several methods for improving the already designed system are presented, providing important research questions in order to explore the true limits of a Bluetooth based motion capture system.