Game Based Approach to Learn Martial Arts for Beginners

摘要

Long hours of training are required to achieve perfect motion in certain sports such as martial arts and dance. In this paper, we are trying to develop a game-based learning system for martial arts that required emphasis of gesture movement with the aid of Kinect sensor. Emergence of controller-less sensors in gaming industry was said to exhibit the potential of bringing joy and fun into teaching and learning experience. We had seen many commercialized controller-less sensors that enhanced the game experience, brought us one step closer to the virtual reality. No doubts, games have unique opportunity to create experiences that unfold over long span of time. Why not we have them kill two birds with one stone? Besides the pleasure brought by the game, we can have the player indirectly learn the real sport through the game play and subsequently train them within the long span of time during game play. Our main idea is to unobtrusively teach player about martial arts during the game play. We have chosen Kinect sensor for its affordability and ease of programming. We have evaluated accuracy of Kinect sensor for motions capture by developing a simple program in Xcode with OpenNI framework and NITE middleware to estimate the player's height by retrieving joint points coordinates information. We have obtained the results of 100% user detection rate, high accuracy but low precision for height estimation with Kinect. This analysis convinced that Kinect is suitable to replace wearable sensor and to fit in our system with proper adjustment and arrangement.