A multi-camera and multimodal dataset for posture and gait analysis.

Area of Science:

• Robotics
• Biomechanics
• Computer Vision

Background:

• Effective gait and posture monitoring is crucial for optimizing assistance from robotic devices and tracking user progress.
• Existing datasets may lack the multimodal, synchronized data needed for comprehensive analysis of human-robot interaction during gait rehabilitation.

Purpose of the Study:

• To present a novel, multi-camera, multimodal dataset for gait and posture analysis while using a wheeled robotic walker.
• To facilitate the development and evaluation of algorithms for pose estimation, human tracking, movement forecasting, and biomechanical analysis in the context of assistive robotics.

Main Methods:

• Collected synchronized, multimodal data (depth, inertial, kinematic) from 14 healthy participants using a wheeled robotic walker.
• Acquired depth data at 30 fps, synchronized with inertial (Xsens MTw Awinda) and kinematic data at 60 Hz.
• Recorded data across 3 gait speeds, 3 walking scenarios, and 3 locations, totaling approximately 92 minutes of synchronized recordings.

Main Results:

• Generated a detailed dataset comprising nearly 166,000 synchronized data samples.
• The dataset captures comprehensive gait and posture information during the use of a wheeled robotic walker.
• Data includes synchronized depth, inertial, and kinematic measurements under varied walking conditions.

Conclusions:

• The presented dataset provides a valuable resource for advancing research in assistive robotics and biomechanics.
• Enables the development and validation of sophisticated algorithms for gait analysis and human-robot interaction.
• Facilitates improved understanding of user progression and rehabilitation effectiveness through detailed movement analysis.