DIODEM - A Diverse Inertial and Optical Dataset of kinEmatic chain Motion

Area of Science:

• Biomechanics
• Robotics
• Sensor Fusion

Background:

• Inertial Motion Tracking (IMT) faces significant challenges like magnetometer-free sensing and motion artifacts.
• Existing algorithms need rigorous evaluation across these combined challenges using accurate ground truth data.

Purpose of the Study:

• To present DIODEM, a novel, comprehensive dataset designed for systematic evaluation of IMT algorithms.
• To facilitate the development of robust IMT solutions by addressing key technical challenges.

Main Methods:

• Collected 46 minutes of synchronized optical and inertial data from five-segment Kinematic Chains (KCs).
• Utilized 20 markers and ten Inertial Measurement Units (IMUs), including foam-attached sensors to simulate motion artifacts.
• Incorporated diverse joint types (1D, 2D, 3D) and kinematic configurations (arm, gait).
• Included varied motions: random movements, pick-and-place tasks, and gait patterns.

Main Results:

• The DIODEM dataset offers a mechanically controlled setup with known kinematics.
• It systematically includes motion artifacts and supports sparse sensing scenarios.
• The dataset covers diverse joint types and comprehensive motion variety.

Conclusions:

• DIODEM enables systematic study of individual and combined IMT challenges.
• The dataset will advance algorithm development for biomechanics, human-computer interaction, and autonomous systems.