Research On Interference Detection In Virtual Assembly

Vrtual assembly is an important part of virtual prototyping.The authenticity of assembly effect has a direct impact on the immersive user experience.Furthermore,interference detection is a key element of virtual assembly.Although algorithms to solve the interference detection problem on CPUs have already been well studied,as the models in virtual assembly become more and more complex,these serial algorithms are no longer competent for fast simulation.With the rapid rising of general purpose parallel computing on GPU,many attempts to parallelize interference detection result in massive speed up.But the real-time performance is still a challenge,with a low degree of parallelism.This thesis gives a brief review on interference detection and heterogeneous parallel computing,and provides an acceleration approach to interference detection based on heterogeneous parallel computing on GPU.The approach has four stage,all of which are executed in each frame.First,sort the objects parallelly by their Morton code using bitonic sort.Second,build the bounding volume hierarchy by parallelly generate all internal nodes in a single run,utilizing the good characteristic of Morton code.Third,generate the bounding volumes layer by layer using bottom up method.Finally,traverse the hierarchy parallelly performing interference detection by maintaining a traverse stack.At last,interference detection is performed on test obj ects,which shows the validity of the proposed approach.