Ryan Aylward, Joseph A. Paradiso

Abstract

In this paper, we present a wireless sensor platform designed for processing multipoint human motion with low latency and high resolution. One application considered here is interactive dance, in which a choreographer wishes to translate the movements of multiple dancers into real-time audio or video content to accompany the performance. This can only be accomplished using a distributed measurement system capable of responding quickly with enough information to describe the expressive movements of multiple people. Similar requirements exist for biomechanical analysis, especially in the context of athletic training, where high resolution is demanded, and instant feedback is also desirable. Our approach to addressing such aggressive requirements involves a high-speed wireless network of compact inertial measurement units (IMUs) that can be worn at various locations on the body. Each device is equipped with its own 1Mbps radio link and a full six-axis IMU, as well as a capacitive node-to-node proximity sensor. Currently, the system supports real-time data collection and processing for up to 25 nodes with 100Hz full state updates, thereby handling much higher data rates than its predecessors. With locally buffered data, sample rates of up to 1kHz have been achieved successfully. Early results discussed here demonstrate the feasibility of our design through testing with both dancers and professional athletes.