Real-Time Short-Term Pedestrian Trajectory Prediction Based on Gait Biomechanics.

Area of Science:

• Robotics
• Biomechanics
• Human-Robot Interaction

Background:

• Short-term trajectory prediction is crucial for human-robot interaction in shared environments.
• Traditional physics-based models struggle with human walking's complex dynamics.
• Existing kinematic models face performance limitations due to gait variability.

Purpose of the Study:

• To develop a computationally efficient, short-term human walking trajectory prediction method.
• To leverage gait biomechanics for improved prediction accuracy in real-time applications.
• To enable feasible implementation on low-cost, portable devices for human-robot interaction.

Main Methods:

• A novel prediction method based on a single biomechanical variable derived from gait analysis.
• Real-time computational approach with a low burden, suitable for embedded systems.
• Experimental validation using a benchmark with multiple subjects on diverse paths.

Main Results:

• The proposed method demonstrates sufficient accuracy for practical applications.
• The low computational cost makes it suitable for real-time, resource-constrained environments.
• Successful evaluation across straight and curved walking paths.

Conclusions:

• The biomechanics-based approach offers a viable solution for short-term human trajectory prediction.
• This method enhances the safety and efficiency of human-robot collaboration.
• The approach is adaptable for various human-robot interaction scenarios requiring predictive capabilities.