Movement-Driven Interaction with a Virtual Pet: A Rehabilitation Prototype for Bedridden Patients

초록

Post-surgery rehabilitation is often limited by patients' low motivation to engage in repetitive exercises. This study explores an interaction structure designed to encourage motivation in bedside contexts. Rather than relying on precise posture recognition, the system applies a forgiving-by-design approach, in which physical movements above a certain threshold are translated into actions performed by a virtual character. Instead of linking motion to a humanoid avatar, the prototype reflects user movements through a virtual animal character. This design is based on clinical feedback indicating that interaction with animals can enhance patient motivation. The system was developed using IMU and camera-based sensors to examine the feasibility of motion tracking in supine positions. It does not require precise avatar rigging, yet successfully generates meaningful responses within a virtual environment. This approach offers a practical interaction model for rehabilitation by connecting physical movements to actions within a virtual space. © 2025 IEEE.

키워드