Research Problems: Collision Detection in Cloth Simulation Nur Saadah Mohd Shapri #1 , Abdullah Bade *2 , Daut Daman #3 #* Department of Graphics and Multimedia, Faculty of Computer Science and Information System, Universiti Teknologi Malaysia, 81310 Skudai, Johor Darul Takzim. Malaysia. 1 nursaadah5@gmail.com 2 abade08@yahoo.com 3 daut@utm.my Abstract— Collision detection can be segregated into two main divisions. The first division is known as inter-collision detection which refers as the detection process of two objects both for solid and deformable objects. The second division of collision detection is self-collision. Collision detection has three components namely collision detection, collision determination and collision response. Collision detection is a process to report Boolean information whether two objects come into contact. Likewise, collision determination also provides Boolean report with detailed information such as vertex, edge and surface. The last component is collision response that may involves some laws of physic. In order to apply proper response for collision detection, it needs to consider the object surface orientation. For an example, if there are the object surfaces that have no topological connections, the correct response is required to apply for suitable surface which is its very complicated and high computational cost. One of the major challenges is how to generate realistic and robust cloth animation with less computation time. In this paper, we provide an overview of the problems in collision detection between flexible objects such as cloth that most require further study in computer graphics. Keywords— Cloth simulation, Collision detection I. INTRODUCTION Cloth has a few unique features. The features consist of flexible, elastic and easy to pleat and shape. Therefore, it is no wondering if there are many researchers interest to study the structure and their characteristics. The characteristics are very complex to simulate and difficult to elaborate and making their prediction. According to Volino et al. [1], computation algorithms have been developed over many years and do not only simulate static clothes but also complex moving garments. There are two types collision in cloth: cloth with environment and cloth with cloth. Collision for complex cloth sometimes the researchers encountered a period of in order difficulty to construct stable cloth model, strong and realistic. Collision handling becomes the most important issue to avoid penetration between object animation. It is usually one of the bottlenecks in real-time animation. In addition, collision in virtual environment is one live natural nature in object. It is usually classified into three phases. First phase is identifying either existing collision between object. Second phase is determining point true position collision and third phase is generating reaction collision based on object qualities. Most of previous works, various of techniques were produced by researcher in their finding and solution which is related to collision. However, these techniques mainly suitable for certain environment only depend on the object and the physical property in certain applications such as robotics, animation, virtual reality, medical and Computer-Aided Design (CAD). Robotics technology usually involves study in solid object. Generally, collision between solid object can be changed the position, object direction and speed. This happened when the higher pressure imposed on object collides. Durupinar [2] strongly stated that self-collision detection is a special case of collision detection where both of the intersecting geometrical primitives are located side by side. In other words, these primitives are the collection of geometrical primitives that create the deformable object. Series of tasks are going to be faced such as detecting multiple collisions, collision consistency and adjacency in primitives and bounding boxes. In this situation, the propose algorithm and technique should robust and efficient enough to handle various type of collisions setting simultaneously. Without consideration of adjacency features, the collision handling is hard to be performed efficiently. It is hard to calculate the sub area where collision is going to occur. It is a very challenging task to create realistic, unteractive, stable and complex collision detection for cloth simulation. Choi and Ko [3] handle the collisions using voxel-based collision detection that inspired from [4]. The cloth particle and solid triangle were register to the corresponding voxel based on their spatial coordinates and perform the collision detection for each voxel. This technique checks the particle pairs and adds the repulsive proximity force between the colliding when the particles too close. In order to speed up the process of interference detection, hierarchically with bounding volume can be applied. There are several techniques that has been used by previous researchers to reduce the number of detecting collision in cloth simulation [5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16]. The latter will be describing in section III. See Figure 1 and Figure 2 for the simulation results from Choi and Ko’s and Lv et al.’s works.