Embedded and visual programming for SmartSuit motion capture system
Nguyen, Kim Doang
Date of Issue2009
School of Mechanical and Aerospace Engineering
Robotics Research Centre
This thesis presents a novel sensing technology using optical linear encoders (OLE) to capture human motion. A CAN bus architecture is proposed to connect the OLE-based sensor nodes, each of which consists of a tri-axis accelerometer and an OLE. The network of three SmartSuit sensing modules is able to capture full motion of human arm with a 7-DOF kinematic model. Firmware programs are developed and embedded into wearable sensor nodes to implement the architecture. In addition, the programming framework for motion capture and processing is introduced. Based on the framework, the motion capture software is developed with all necessary features. The software interfaces with the wearable sensor hardware via serial communication. The motion data are processed and stored in hierarchical models. The software’s graphics display unit regenerates the body motion using either OpenGL rendering method or modeling softwares. Moreover, the software can import and export standard motion data formats, facilitating our OLE-based motion capture system to communicate with various platforms. Experiments were carried out to compare the performance of the OLE sensing module with BIOPAC Goniometer. The results show that the OLE’s performance is comparable to that of those expensive systems, and also validate the sensor network architecture, firmware and the SmartSuit software. Furthermore, a statistical study is conducted to confirm the repeatability and reliability of the new OLE sensing module and the wearable sensor network. The results demonstrate that the new sensor system has strong potential to be used as a low-cost tool for motion capture, and arm function evaluation for short-term as well as long-term monitoring.