| Augmented Reality could add the virtual objects created by PC into the real-world, and could achieve seamless integration with virtual reality and actual reality by means of some display devices, consequently, a real visual effect will be presented to the users. How to interact between virtual reality and actual reality is a new research direction in this technology. Collision Detection could detect whether there are contacts or penetrations among multiple objects and return the corresponding collision information, it provided theory support for interaction between virtual objects.Therefore, the Augmented Reality technology and Collision Detection technology were combined to achieve interactive function in Augmented Reality technology.With the use of augmented reality system ARToolKit as development environment, Open Scene Graph as developer's kit to make up for the trivial image description provided by lower-level OpenGL, Open Dynamic Engine as developer's kit to add dynamic behavior to the virtual object and so as to allow virtual object achieve mechanical motion like real object, in addition, the key technology and basic flowchart of how to combine the ARToolKit, Open Scene Graph and Open Dynamic Engine are discussed. Aligned Axis Bounding Box provided by OPCODE which built in ODE was improved with the use of Oriented Bounding Boxes, the improvement project and Basic flowcharts are explained. Last, the GPU cache was operated with the help of Stencil Buffer test and Depth Buffer test, the primary collision detection algorithm based on bounding boxes was accelerated by means of GPU, and an experiment was performed in the augmented reality scene.By means of experiment rendering and correlation table, the interactive augmented reality system realized simple interactive function, although the improvement measure in the collision detection algorithm increased the preprocessing time to build a tree, it shorted the real collision detection time observably, increased the collision detection efficiency. Speedup of the collision detection based on GPU was restricted by GPU speed and algorithmic limitation, the accelerated effect was not obvious, whereas the system still achieve the speed-up ratio, and promoted the efficiency. |
