Key point trajectory prediction method of human stochastic posture falls.

Area of Science:

• Biomechanics
• Human Motion Analysis
• Computer Vision

Background:

• Human falls involve complex posture changes affecting body posture, limb position, and movement.
• Accurate modeling of these changes is crucial for understanding fall dynamics and injury prevention.

Purpose of the Study:

• To develop a method for creating a 3D human model from key point data.
• To predict the motion trajectory of random human falls in 3D space.
• To accurately differentiate between falling and squatting motions.

Main Methods:

• Mapping 3D model key points to spatial coordinates.
• Utilizing construction decomposition to calculate rotation matrices and angular displacement.
• Applying curve fitting with a self-organizing mapping-based kernel function for trajectory prediction.

Main Results:

• The generated 3D model accurately represents human body structure.
• The method successfully predicts key point trajectories for random human falls.
• The system accurately distinguishes between random falls and squats.

Conclusions:

• The developed method provides a reliable approach for analyzing and predicting human fall trajectories.
• This technique enhances the understanding of human fall biomechanics.
• The findings have implications for fall detection systems and injury mitigation strategies.