A Research on Improvement of Accuracy of Collision Detection for Rotaion Rigid Bodies In Real-Time Graphics

Abstract

Today, collision detection between rigid bodies in computer games often uses Bounding Volumes that allow for fast decision making. However, because rigid bodies move discretely in the computer, they are tunneling each other without colliding if their motion velocity is very large. Various CCD (Continuous Collision Detection) methods have been proposed to address this problem and have been used in FPS (First Person Shooting) games that handle rigid bodies with large velocities such as bullets. One example is the Sweep-based CCD method and the Speculative CCD method used in Unity. However, the Sweep-based CCD method ignores angular motion and does not support rotational motion, while the Speculative CCD method has low collision detection accuracy for rotational motion. To solve this problem, this study interpolates the motion trajectory by using a fan shape, which is an approximate shape of the rotational motion trajectory of a rectangular body, to improve the accuracy of collision detection for rotating rigid bodies. This paper proposes a collision detection method between a fan shape and a primitive shape, and presents the verification results of whether the accuracy of collision detection is improved by interpolating the rotational motion trajectory using the fan shape, as well as the execution speed of the proposed method.