| 摘要: | 隨著高齡化社會的加劇，老年人的身體功能退化問題日益受到重視，特別是在平衡能力方面，衰退所導致的跌倒風險更是影響其生活品質與安全的重大因素。傳統的復健訓練雖能有效強化肌力與穩定性，但常因缺乏趣味性與即時回饋而導致參與意願低落，影響訓練成效與持續性。有鑑於此，本研究旨在開發一套結合MediaPipe動作捕捉與Unity遊戲引擎的VR虛擬實境平衡訓練系統，作為提升平衡訓練參與度與效果的創新工具。系統設計採用市售HTC VIVE Pro作為VR操作平台，搭配網路攝影機與 MediaPipe Pose模型即時擷取使用者的33個人體關鍵點姿態資訊，並透過使用者資料包協定(User Datagram Protocol, UDP)將資料整合至使用Unity所開發的互動式遊戲場景中，建立即時回饋與沉浸式體驗。訓練內容共分為三個具情境敘事的關卡：第一關透過左右閃躲設計刺激重心轉移與前庭眼反射(Vestibulo-Ocular Reflex, VOR)功能；第二關要求玩家追蹤並擊落動態目標，以強化視覺與前庭整合反應；第三關則融合下蹲啟動機制，針對股四頭肌進行初階肌力訓練，強化姿勢控制與跌倒預防能力。
本研究採單組前後測實驗設計，透過單腳站立測試與 ABC 平衡信心量表（Activities-Specific Balance Confidence Scale）評估虛擬實境遊戲訓練系統之介入成效。考量本階段尚未進行倫理審查程序（Institutional Review Board, IRB），為避免高風險族群參與之倫理疑慮，實驗對象改以 18 至 25 歲之年輕成人為樣本，模擬實際應用場景。
實驗期程為期 7 至 10 天，結果顯示 5 名受測者於閉眼單腳站立測
試與遊戲內得分上，平均表現皆呈現一定幅度之提升。進一步以魏克生符
號等級檢定進行統計分析，發現部分指標達顯著水準，初步驗證本系統具
備促進平衡能力與操作熟練度之潛力。雖受限於樣本數限制，研究仍顯示
本訓練系統未來可望發展為一種新的低成本、高親和力之智慧型居家復健
模型，作為老年族群平衡訓練與個人化健康促進之可行應用基礎。;With the intensification of population aging, the decline in physical function among older adults has become an increasingly critical issue, particularly in balance ability. Impaired balance significantly increases the risk of falls, directly affecting quality of life and safety. Although traditional rehabilitation training can effectively enhance muscle strength and stability, it often lacks engagement and real-time feedback, leading to low motivation and poor adherence. To address this, the present study aims to develop a virtual reality (VR) balance training system that integrates MediaPipe motion tracking and the Unity game engine, providing an innovative tool to enhance participation and effectiveness in balance training.
The system utilizes the HTC VIVE Pro for VR interaction and a webcam with MediaPipe Pose to capture real-time skeletal data of 33 body landmarks. These motion data are transmitted via UDP protocol to control interactive scenes built in Unity, providing immersive feedback and real-time responsiveness.The training is structured into three narrative-rich levels: Level 1 involves lateral dodging to stimulate weight-shifting and vestibulo-ocular reflex (VOR); Level 2 requires visual tracking and striking moving targets, enhancing visual-vestibular integration; Level 3 introduces squat-based mechanics to activate the quadriceps for foundational strength training aimed at fall prevention.
This study adopts a single-group pretest-posttest experimental design to evaluate the effectiveness of a virtual reality (VR) game-based training system using the Single-Leg Stance Test (SLST) and the Activities-Specific Balance Confidence Scale (ABC Scale). Considering that the Institutional Review Board (IRB) approval was not obtained at this stage, and to avoid ethical concerns regarding high-risk populations, the participants were limited to young adults aged 18 to 25 as a simulated user group.
The experimental period spanned 7 to 10 days. Results from five participants showed a general trend of improvement in both the SLST under eyes-closed conditions and in-game performance scores. Further statistical analysis using the Wilcoxon signed-rank test revealed significant differences in selected indicators, supporting the system′s potential in enhancing balance abilities and operational proficiency. Although the sample size was limited, the findings suggest that the proposed system may serve as a low-cost, user-friendly smart rehabilitation model for home-based training, offering a promising approach for balance enhancement and personalized health promotion among older adults. |
