The Research And Application Of Collision On Virtual Platform For Disassembly & Assembly

The technique of collision detection has been widely applied to many fields, such as robotic path planning, automatic assembly planning since 1970s. Many collision detection algorithms have been developed with different advantages. While applied to practical projects, it is worth to optimize and improve the collision detection algorithms to meet the actual requirements of special applications.In this thesis, we focus on the collision detection based application in Virtual Platform for Disassemble and Assembly. Current collision detection algorithms are analyzed in two aspects:real-time and accuracy. The accuracy of algorithms is increased in the order of BV (Bounding Volume), BVH (Bounding Volume of Hierarchical), Triangle-Mesh, Triangles; the real-time of algorithms is decreased in the order.A binary tree optimization based hierarchical framework of collision detection is proposed to solve the drawback of real-time occurred in current hierarchical collision detection algorithms. The selection of parts removed the impossible collision parts. According to the character of hybrid self-detection of collision, a top-to-down method is used to constrct the BVT(Bounding Volume Tree). And a strategy of binary tree optimization is proposed to reduce the number of basic elements attending the precise detection to improve the real-time performance. Finally, the detection of triangles is used to detect collisions. The results show that the optimization based hierarchical detection performs more efficient than the hierarchical detection, and has a faster detection speed.The effect of detection is good when applied to constrained parts, while the performance of real-time decreases when applied to un-constrained parties under complex scene in Virtual Platform for Disassembly&Assembly. In order to improve the accuracy and real-time performance, a triangle-mesh based collision detection algorithm is proposed, which includes edge collapse technology and memory optimization strategy. The edge collapse reduces the number of triangles attending the collision detection to improve the real-time performance. And memory optimization reduces the memory utilization rate with reasonable cost of accuracy. Experimental results show that the optimized triangle mesh algorithm meets the real-time requirement in virtual disassemble and assemble platform.