Academia.eduAcademia.edu

Figure from:Review on narrow phase collision detection system in virtual environment

See full PDFdownloadDownload figure
Table-2. Techniques for point of contact. penetrates into other object (the object completely inside the other object) (Escande, Miossec et al. 2007). Zhang et al., (Zhang, Kim et al. 2007) provided a generalized penetration depth computation which is able to compute object penetration depth faster than other previous method mentioned in their research paper but depends heavily on pre-processing configuration with a lot of predefined (Shengzheng and Jie, 2009; Choi, Ryu et al. 2010; Zhang and Du, 2011; Zhi, Shiqiu et al. 2011; Heng, KunChao et al. 2012). Later, Hausdorff distance in (Tang, Lee et al. 2009; Barton, Hanniel et al. 2010) was used to calculate the penetration depth for polygon but depends on closed models (where there is no hole for the object) and well- defined inside and outside vertices points (Je, Tang et al. 2012; Jia, Chitta et al. 2012; Zhang, Kim et al. 2014). Table-3 summarized all our findings for penetration depth issues. Considering the last component of narrow phase collision detection, Gilbert (Gilbert, Johnson et al. 1988) proposed a fast procedure to handle penetration of narrow phase collision detection but sometimes the technique reports false positive result for potential primitives (Cameron, 1997; Ponamgi, Manocha et al. 1997). Lin- Canny technique does not offer penetrating handling technique well but with a simple technique to cover penetration depth where it involves calculating the total distance travel from one point into another point that penetrate the other object (Bergen, 2001; Kim, Otaduy ef al. 2003; Stephane and Lin, 2006). The technique itself could be a repeated cycle forever if penetration happens (Kim, Otaduy et al. 2003; Stephane and Lin, 2006; Zhang, Kim et al. 2007). Mirtich (Mirtich, 1998), developed a V- Clip complex technique that is capable to handle penetration but it is inefficient when the object completely

Table 2 -2. Techniques for point of contact. penetrates into other object (the object completely inside the other object) (Escande, Miossec et al. 2007). Zhang et al., (Zhang, Kim et al. 2007) provided a generalized penetration depth computation which is able to compute object penetration depth faster than other previous method mentioned in their research paper but depends heavily on pre-processing configuration with a lot of predefined (Shengzheng and Jie, 2009; Choi, Ryu et al. 2010; Zhang and Du, 2011; Zhi, Shiqiu et al. 2011; Heng, KunChao et al. 2012). Later, Hausdorff distance in (Tang, Lee et al. 2009; Barton, Hanniel et al. 2010) was used to calculate the penetration depth for polygon but depends on closed models (where there is no hole for the object) and well- defined inside and outside vertices points (Je, Tang et al. 2012; Jia, Chitta et al. 2012; Zhang, Kim et al. 2014). Table-3 summarized all our findings for penetration depth issues. Considering the last component of narrow phase collision detection, Gilbert (Gilbert, Johnson et al. 1988) proposed a fast procedure to handle penetration of narrow phase collision detection but sometimes the technique reports false positive result for potential primitives (Cameron, 1997; Ponamgi, Manocha et al. 1997). Lin- Canny technique does not offer penetrating handling technique well but with a simple technique to cover penetration depth where it involves calculating the total distance travel from one point into another point that penetrate the other object (Bergen, 2001; Kim, Otaduy ef al. 2003; Stephane and Lin, 2006). The technique itself could be a repeated cycle forever if penetration happens (Kim, Otaduy et al. 2003; Stephane and Lin, 2006; Zhang, Kim et al. 2007). Mirtich (Mirtich, 1998), developed a V- Clip complex technique that is capable to handle penetration but it is inefficient when the object completely

Related Figures (4)

Figure-1. Collision detection pipeline (Zachmann, 2001). object has thin surfaces or fast affecting objects that eventually missed the collision. It is important for hide- fidelity simulation even though it does not require for fast response simulation such as in computer games or animation. Thus, CCD hardly missed any potential collision between those configurations. As shown in Figure-1, the pipeline shows two general phases of collision detection pipeline which are broad phase collision detection (starting from object handler until interesting pairs block) and narrow phase collision detection (starting from grid until colliding pairs block). Generally, the main challenges of narrow phase collision detection stage when precision becomes the important element to determine the colliding pairs.
Figure-1. Collision detection pipeline (Zachmann, 2001). object has thin surfaces or fast affecting objects that eventually missed the collision. It is important for hide- fidelity simulation even though it does not require for fast response simulation such as in computer games or animation. Thus, CCD hardly missed any potential collision between those configurations. As shown in Figure-1, the pipeline shows two general phases of collision detection pipeline which are broad phase collision detection (starting from object handler until interesting pairs block) and narrow phase collision detection (starting from grid until colliding pairs block). Generally, the main challenges of narrow phase collision detection stage when precision becomes the important element to determine the colliding pairs.
Table-1. Techniques for distance computation. Meanwhile, point of contact technique for narrow phase collision detection was first introduced by (Gilbert, Johnson et al. 1988). The technique used simplex based implementation based on GJK but failed to satisfy the termination criteria (Gilbert and Foo 1990) that had been used to calculate the possible nearest points. Lin Canny technique was introduced then by implementing reducing resolution technique for detecting point of contact (Lin and Canny 1991). However, the technique missed certain cases where it may fall in infinite loop for complex polygons (Cameron, 1997; Chong-Jin and Gilbert, 2001; Sundaraj and Retnasamy, 2007; Shen and Li, 2010). Mirtich (1998) in (Chakraborty, Jufeng et al. 2008) proposed a V-Clip that provided solution for the Lin- Canny but not suitable when the coherence level is high. Later Ma et al (2012) proposed a distance computation technique for canal surfaces to detect point of contact but dedicated only for canal surfaces and involved complex calculations with only a few iterations and not suitable for massive models (Je, Tang et al. 2012; Yu and Yamane, 2013; Wang, Shi et al. 2014; Zhang, Kim et al., 2014). Trajectory based technique was implemented for detecting point of contact has been proposed in Schulman but according to the author itself, it involves expensive iteration and the trajectory could be missed sometimes (Schulman, Lee et al. 2013). Table-2 shows a summarized version of point of contact techniques.
Table-1. Techniques for distance computation. Meanwhile, point of contact technique for narrow phase collision detection was first introduced by (Gilbert, Johnson et al. 1988). The technique used simplex based implementation based on GJK but failed to satisfy the termination criteria (Gilbert and Foo 1990) that had been used to calculate the possible nearest points. Lin Canny technique was introduced then by implementing reducing resolution technique for detecting point of contact (Lin and Canny 1991). However, the technique missed certain cases where it may fall in infinite loop for complex polygons (Cameron, 1997; Chong-Jin and Gilbert, 2001; Sundaraj and Retnasamy, 2007; Shen and Li, 2010). Mirtich (1998) in (Chakraborty, Jufeng et al. 2008) proposed a V-Clip that provided solution for the Lin- Canny but not suitable when the coherence level is high. Later Ma et al (2012) proposed a distance computation technique for canal surfaces to detect point of contact but dedicated only for canal surfaces and involved complex calculations with only a few iterations and not suitable for massive models (Je, Tang et al. 2012; Yu and Yamane, 2013; Wang, Shi et al. 2014; Zhang, Kim et al., 2014). Trajectory based technique was implemented for detecting point of contact has been proposed in Schulman but according to the author itself, it involves expensive iteration and the trajectory could be missed sometimes (Schulman, Lee et al. 2013). Table-2 shows a summarized version of point of contact techniques.
of manipulating limited type of polygon, the proposed technique should indulge a better solution for any type of object that consist of vertices, edges and surfaces. The generalized technique should be able to handle distance computation, point of contact, and penetration depth with a reasonable speed and nearly accurate computation. Discrete collision detection (DCD) works has been studied a decade ago and the pioneers of this work are (Goldsmith and Salmon, 1987; Gilbert, Johnson et al. 1988; James, 1988; Baraff, 1989; Omohundro, 1989; Gilbert and Foo, 1990; Lin and Canny, 1991) where one of the remarkable researches conducted by (Gilbert, Johnson et al. 1988). The author created an efficient technique for computing the Euclidean distance between triangles using mathematical programming method. It used the position and orientation of the object and detect any potential collision along a continuous path (refer to Figure-1). An improvement has been made later on by using a simple
of manipulating limited type of polygon, the proposed technique should indulge a better solution for any type of object that consist of vertices, edges and surfaces. The generalized technique should be able to handle distance computation, point of contact, and penetration depth with a reasonable speed and nearly accurate computation. Discrete collision detection (DCD) works has been studied a decade ago and the pioneers of this work are (Goldsmith and Salmon, 1987; Gilbert, Johnson et al. 1988; James, 1988; Baraff, 1989; Omohundro, 1989; Gilbert and Foo, 1990; Lin and Canny, 1991) where one of the remarkable researches conducted by (Gilbert, Johnson et al. 1988). The author created an efficient technique for computing the Euclidean distance between triangles using mathematical programming method. It used the position and orientation of the object and detect any potential collision along a continuous path (refer to Figure-1). An improvement has been made later on by using a simple
Figure-3. Differences between normal oct-tree and medial axis surface introduced by Hubbard (Hubbard, 1996). Hubbard introduced a sphere tree construction by using oct-tree based type based on previous research by Quinlan (Hubbard 1996). By using space-time bound technique, Hubbard technique allowed the collision detection technique to work fast by approximation and graceful deprivation on the objects. In his dissertation, he concluded that the implementation of the technique works faster than the previous one. REFERENCES
Figure-3. Differences between normal oct-tree and medial axis surface introduced by Hubbard (Hubbard, 1996). Hubbard introduced a sphere tree construction by using oct-tree based type based on previous research by Quinlan (Hubbard 1996). By using space-time bound technique, Hubbard technique allowed the collision detection technique to work fast by approximation and graceful deprivation on the objects. In his dissertation, he concluded that the implementation of the technique works faster than the previous one. REFERENCES