Research and Literature Review on Developing Motion Capture System for Analyzing Athletes Action

Han Fang, Xuesong BoAdvances 2015 · DOI

Abstract

With the development of human-computer interaction technology, natural interaction will become the main way of interaction between human and computer. To support this technology, the computer must understand and capture human behavior characteristics, leading to the development of motion capture systems. The human body is a complex system composed of more than 200 rotary joints; to simulate vivid movement, one must provide all joint rotation angle values and positional information. This paper conducts a literature review and presents the development of a motion capture system for analyzing athletes' actions. The results demonstrate the effectiveness of the proposed system, with future plans focusing on robustness testing through further experimentation.

Introduction: The Need for Portable Motion Capture

Traditional motion capture systems are often constrained by their operating environments. Optical systems offer high accuracy but are prohibitively expensive and tethered to laboratory settings. Acoustic systems suffer from line-of-sight issues, while magnetic systems are sensitive to external interference. There is a clear demand for a "wear-resisting," affordable, and portable system capable of functioning in diverse environments, particularly for sports science and medicine. The authors propose the 3-d SUIT to address these limitations.

The 3-d SUIT System Architecture

The 3-d SUIT is a wearable inertial motion capture system built on a modular architecture. It utilizes 9-Degrees-of-Freedom (9DOF) Inertial Measurement Units (IMUs) to track movement across a 21-segment skeletal model.

Figure 1: Overview of the 3-d SUIT system architecture, including the 9DOF IMU components and the central processing unit.

As shown in Figure 1, the hardware integrates an STM32 32-bit MCU with specialized MEMS sensors, including 3-axis gyroscopes, accelerometers, and magnetometers. To combat the inherent "drift" of gyroscopes—where errors accumulate over time—the system employs sensor fusion. By using the Earth's gravitational and magnetic fields to provide an absolute reference frame (North-East-Down), the system maintains orientation accuracy throughout extended use.

Skeletal Modeling and Motion Reconstruction

The system models the human body as a hierarchy of 21 segments or "bones." The "Hips" serve as the static root node, and all other joint movements are calculated as rotations relative to their parent nodes. This hierarchical approach simplifies data processing, as only relative rotations are required to reconstruct complex poses. The software stack utilizes VC++ and OpenGL for real-time data processing, while 3ds Max and Motion-Builder are employed for character skinning and animation. Additionally, Bessel curve characteristics are utilized to smooth numerical data, aiming for more lifelike joint simulations.

Results and Simulation

The 3-d SUIT successfully simulated fundamental human movements, such as walking and running cycles. Unlike optical systems, the 3-d SUIT functions effectively without the need for calibrated camera arrays or specific lighting conditions.

Figure 4: 3D simulation results showing the skeletal reconstruction and the final skinned character model performing a walking motion.

As illustrated in Figure 4, the system effectively captures limb harmony for the arms and legs. While the current iteration shows some limitations in torso and head movement realism, the overall reconstruction of athletic maneuvers is promising. The authors claim that accuracy and efficiency are significantly enhanced compared to previous iterations.

Conclusion and Future Work

The 3-d SUIT provides a low-cost, portable alternative for high-fidelity human motion simulation. While the system demonstrates qualitative success in reconstructing movement, future work will focus on improving the "lifelike" quality of animations—specifically addressing the static nature of the torso—and optimizing the balance between battery life and sensor weight for extended athletic monitoring.

Summary updated

Abstract

With the development of human-computer interaction technology, pattern, natural interaction will become the main way of interaction between human and computer. Support technology, the computer must understand and capture the human behavior characteristics, so as to put forward the motion capture. The human body is a very complex system, composed of more than 200 rotary joint, if you need to simulate the human body movement of the more vivid, shall provide all joint rotation angle of the value and position of the human body and other relevant information. We conduct literature review on developing motion capture system for analyzing athletes' action in this paper. The result shows the effectiveness of the proposed system. In the future, we plan to do more related experiment to test the robustness of our system.
BibTeX
@inproceedings{fang2015research,
  author = {Han Fang and Xuesong Bo},
  title = {{Research and Literature Review on Developing Motion Capture System for Analyzing Athletes Action}},
  booktitle = {Advances},
  year = {2015},
  doi = {10.2991/etmhs-15.2015.103},
}
Figure