Data-Driven Rendering of Motion Effects for Walking Sensations in Different Gaits.

Area of Science:

• Human-Computer Interaction
• Virtual Reality
• Robotics

Background:

• Motion effects enhance immersion in 4D interactive applications like films and VR games.
• Human locomotion scenes are common, necessitating realistic motion feedback.

Purpose of the Study:

• To propose a data-driven framework for automatic generation of walking sensations.
• To develop an algorithm for synthesizing motion effects that mimic different gaits.

Main Methods:

• Collected locomotion data using body-attached motion sensors during various gaits (walking, running, stumping).
• Processed sensor data into multi-degree-of-freedom commands for a motion platform.
• Utilized Principal Component Analysis (PCA) to reduce the dimensionality of motion commands.

Main Results:

• Demonstrated that motion commands can be effectively represented in a lower-dimensional space using PCA.
• Developed a synthesis algorithm for new motion commands to elicit specific gait sensations.
• Validated the method's perceptual performance through two user studies.

Conclusions:

• Confirmed the feasibility of mimicking walking sensations using motion platforms and human locomotion data.
• Established an automatic generation algorithm for motion effects conveying distinct gait impressions.
• Advanced the development of immersive experiences in 4D interactive applications.