Behavior analysis of catching using 3D pose estimation
No Thumbnail Available
Date
2025
Authors
Mazrouei, AmirHossein
University of Lethbridge. Faculty of Arts and Science
Journal Title
Journal ISSN
Volume Title
Publisher
Lethbridge, Alta. : University of Lethbridge, Dept. of Neuroscience
Abstract
Catching, a complex and fundamental prehension task, is crucial for daily life yet remains understudied despite its implications for robotics, rehabilitation, and neuroprosthetics. This thesis investigates the intricate sensorimotor coordination involved in human catching, building upon theories like the Dual Visuomotor Channel (DVC) and Multiple Motor Channel (MMC) to understand how the brain orchestrates dynamic hand movements.
Ten right-handed participants engaged in externally thrown, self-thrown, and visually guided "pretend" catches using four ball sizes. Behavior was recorded with three GoPro cameras, and 3D pose estimation was performed via FreeMocap (Matthis & Cherian, 2022), leveraging MediaPipe (Zhang et al., 2020) for 2D analysis and triangulation for 3D reconstruction. Three primary kinematic metrics were quantified: 1) Euclidean distance between the thumb tip and other fingertips (opposable distance); 2) Perpendicular distance from each fingertip to the palm plane (prehensile distance); and 3) The hand's rotation angle in the X-Z plane, derived from the palm's normal vector.
Results revealed Maximum Pregrasp Aperture (MPA) scaled linearly with ball diameter, indicating anticipatory hand shaping. Distinct grasping strategies emerged for different ball sizes: larger balls elicited "precision catches" characterized by significant finger splay and thumb-pinky opposition, while smaller balls often resulted in "power catches" with minimal thumb involvement and greater finger flexion into the palm. Self-catches further highlighted the interplay of anticipatory and feedback control. These findings enhance understanding of human prehension, providing quantitative data valuable for advancing motor control models, developing adaptive robotic systems, and improving human-machine interfaces.
Description
Keywords
Catching , Hand movements , Grasping strategies , Motor control