| Summary: | 碩士 === 國立高雄科技大學 === 資訊管理系 === 107 === In recent years, electronic sensing technology has developed vigorously, and the vehicle simulation training mode has become more and more extensive. From 1993, the driving simulation system and the driving simulator of the game industry are used for the convenience of the user driving training or the entertainment effect. At present, the domestic car driving training school adopts one-to-many coaching, and the number of coaches that each student can allocate is small and the common phenomenon is that the coach teaches once and the students practice independently. This system is proposed to increase training effectiveness.
This research is based on the requirements of the driving class. The virtual reality (VR) technology is used to construct a 3D simulation driving training scene, and the real vehicle is used as a cockpit detection user operation data to be transmitted to the PC. Developing VR system optimization is important for performance issues. In this paper, the Bezier Curve plug-in extrusion mesh Bezier curve is used instead of the default Unity Shuriken particle system, and the performance is significantly improved.
In addition, the study refers to the dynamics of the module object based on the quad tree algorithm. The Camera is used to manage the objects in distance from the scene to divide the scene into a number of 4x4 sub-images, and then load or unload depending on the center to avoid occupying unnecessary memory to improve the overall CPU efficacy.
The study found that the Bezier Curve plug-in replaces the Unity Shuriken particle system, which is three to ten times better than the latter. In addition, the results of the module object dynamic loading method using the quad tree algorithm in the proposed method are compared in one-time loading. It is found that the total memory of the former is significantly reduced by 60% and the FPS is doubled.
|
|---|
