| dc.description.abstract | Vision-based motion capture systems (MCSs) are often used as a way to create full-body virtual models of human beings, for applications ranging from movie Computer-Generated Imagery (CGI) to biomechanical analysis of human movements to medical purposes. However, vision-based MCS are often very expensive and require long and complicated preparation procedures. This study aimed to use inertial measurement units (IMUs), which are significantly more cost-effective and easier to use than visual-based MCSs, in order to create a motion capture system with accuracy comparable to that of visual motion capture systems. The IMUs used for the system include 3-axis gyroscopes, 3-axis accelerometers, and 3-axis magnetometers. A novel algorithm is introduced for orientation estimations which makes two position estimates—one using the gyroscope and one using a combination of the accelerometer and magnetometer—and an average is found between the two. Preliminary tests involving a subject performing shoulder abductions/adductions, elbow flexions/extensions, and hip flexions/extensions revealed low root-mean-squared error values and high correlation between joint angles calculated concurrently using the visual- and IMU-based motion capture systems. The ultimate goal of this application is to develop a graphical user interface (GUI) that can facilitate the accurate biomechanical analyses of the human and/or animal movement using kinematic data (e.g., 3D orientation) from low-cost and easy-to-use IMUs. Ultimately, the algorithm is expected to be made open-source, and this application will enable a more affordable, accessible, and portable biomechanics lab of human movement analysis for researchers and provide simple ways for clinicians to diagnose pathological movements of their patients. | en |
