A Collision Detection Algorithm Based on Sphere and EBB Mixed Hierarchical Bounding Boxes

Effective collision detection is crucial for creating a realistic and immersive virtual reality (VR) experience, especially in large and complex VR environments. Collision detection research currently focuses on improving two key aspects: accuracy and real-time performance. This paper proposes a novel collision detection algorithm that uses a mixed layer bounding box approach. The algorithm uses Sphere and EBB (Ellipsoidal Bounding Box) methods to quickly eliminate disjoint objects in the virtual environment and reduce the number of triangular primitives that need to be detected. Tree structures are used to traverse nodes from top to bottom, and then the Devillers & Guigue method is used to calculate accurate triangle-to-triangle primitives, further improving the effectiveness and accuracy of the collision detection. Experimental results show that the proposed algorithm (Sphere-EBB) performs better than traditional hybrid bounding box algorithms like Sphere Axial Aligned Bounding Box (Sphere-AABB) and EBB in terms of mean detection time, detection rate, and frame frequency. This makes the proposed algorithm suitable for large-scale, complex model collision detection tasks in terms of collision detection time and accuracy.