Towards Markerless Motion Estimation of Human Functional Upper Extremity Movement.

Area of Science:

• Biomechanics
• Robotics
• Computer Vision

Background:

• Markerless motion capture offers a portable, low-cost alternative to traditional optical and inertial systems for movement analysis.
• Existing markerless methods using RGB-D data lack accuracy for fine human motions, limiting their application to gross movements.
• Most current methods do not directly utilize depth images for motion estimation.

Purpose of the Study:

• To develop and evaluate a novel markerless motion capture method using depth images for accurate upper extremity movement analysis.
• To segment the upper extremity into kinematic components and estimate motion directly from depth data.
• To compare the proposed method's performance against a gold-standard motion capture system.

Main Methods:

• Utilized depth images from an RGB-D camera to compute upper extremity motion.
• Segmented the arm into rigid components (upper arm, lower arm) and estimated their motion using Iterative Closest Point (ICP) or Distance Transform.
• Calculated end-effector (wrist) motion relative to the torso.
• Compared results with data from Microsoft Azure Kinect and a 9-camera OptiTrack system.

Main Results:

• Point cloud methods demonstrated comparable performance to the OptiTrack motion capture system in tracking arm rotation and velocity.
• The proposed method shows potential as an affordable alternative for motion capture applications.
• Accuracy was validated for gross upper extremity movements.

Conclusions:

• Markerless motion capture using depth imaging is a viable and potentially cost-effective approach for analyzing human upper extremity movement.
• The developed method shows promise for applications in movement science and rehabilitation.
• Future research will focus on refining the method for capturing fine motor skills.