Gabriel Zachmann. Rapid collision detection by dynamically aligned DOP-trees. In Proc. of IEEE Virtual Reality Annual International Symposium; VRAIS '98. Atlanta, Georgia, March 1998.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....Hierarchy In complex dynamic scenes, bounding volumes have to be permanently readapted to the approximated geometry. Arbitrary object deformations cannot be expressed by rigid rotations and translations, thus orientable bounding volumes like OBBs [11] QuOSPOs [13] or Dynamically Aligned DOPs [24] are not suitable for cloth modelling. We therefore use common k DOPs [16] which provide better convergence than spheres [14] or ordinary AABBs [21] and can be adjusted to memory limits by the choice of k. Additionally, k DOP bounding volumes (BVk) inherit from AABBs the feature BVk(V) U ....

G. Zachmann. Rapid Collision Detection by Dy- namically Aligned DOP-Trees. Proc. of IEEE, VRAIS'98 Atlanta, 1998.

....Volumes In complex dynamic scenes, bounding volumes have to be permanently readapted to the approximated geometry. Arbitrary object deformations cannot be expressed by rigid rotations and translations, thus orientable bounding volumes like OBBs [13] QuOSPOs [16] or Dynamically Aligned DOPs [31] are not suitable for cloth modelling. We choose com mon k DOPs [19] as they show better convergence than spheres [17] or ordinary AABBs [26] The advantages of a bounding volume hierarchy of k DOPs over other hierarchies are exposed in section 4. 3.1 k DOPs A k DOP[19] discrete oriented ....

G. Zachmann. Rapid Collision Detection by Dynamically Aligned DOP-Trees. Proc. of IEEE, VRAIS'93 Atlanta, 1998.

....by applying methods of quadratic programming. The last but not least important approach presented here is replacing transformed FDH by its FDH. This method has been used in [He 96, Za 98] though the implementation is quiet different from what is presented here. The same method as presented in [Za 98] actually works only in 2D and for some very special cases also in higher dimensions, though the author does not explicitly mention that. The implementation in [He 96] is a general solution, however it requires memory in the order of magnitude larger than our approach. A special case of solution ....

....to some optimization criteria. 3.2.1 Choice of Axis The choice of axis often depends on particular bounding volume. The most simple approach is choosing one of coordinate axes as in [He 96] Authors of [Go 96] use longest side of oriented bounding box enclosing the set of primitives, while in [Za 98] a line passing through two most distant points is employed. The choice of the splitting axis in our implementation is closely related to the class of bounding volumes we use. The algorithm in our implementation uses combination of statistical method similar to the one in [Go 96] and fixed number ....

[Article contains additional citation context not shown here]

G. Zachmann. Rapid Collision Detection by Dynamically aligned DOP-Trees. Proc. of IEEE, VRAIS'98, Atlanta, Georgia, March 1998. 40

....planning problem is difficult to solve for a planning algorithm, the environments have simple geometry. CAD Applications. Some contributions to the field of CD are applied to industrial scenes. These include research aimed at the introduction of Virtual Reality techniques for car industry ( 27] [28]) and cooperation with Boeing ( 9] 15] There exist also some path planning applications in industrial environments ( 21] 25] In all these applications unimportant cavities are omitted, models are simplified, the number of calls to the CD minimized. 3 KCD for PRM in CAD Models. Previously ....

G. Zachmann. Rapid Collision Detection by Dynamically Aligned DOP-Trees. In IEEE Virtual Reality Annual International Symposium, 1998.

....models 1 1 Introduction Hierarchies of bounding volumes provide a fast way to perform exact collision detection between complex models. Examples of volume types that are used for this purpose are spheres [8, 5] oriented bounding boxes (OBBs) 4] and discreteorientation polytopes (DOPs) [6, 12]. In this paper, we present a collision detection scheme that relies on a hierarchical model representation using axis aligned bounding boxes (AABBs) In the AABB trees as we use them, the boxes are aligned to the axes of the model s local coordinate system, thus, all the boxes in a tree have the ....

G. Zachmann. Rapid collision detection by dynamically aligned DOP-trees. In Proc. IEEE Virtual Reality Annual International Symposium, pages 90-- 97, 1998. 14

....shell is a BV possessing even better tightness and smaller update cost[11] However, a common problem with these approaches is that they require considerable amount of memory storage. Discrete Orientation Polytope (DOP) is another good choice of BV that requires smaller memory storage than OBB[9][17]. A scaleable scheme utilizing bit addressing to detect possible collisions among thousand of objects with various resolutions has also been proposed in [5] In addition to these empirical proposals, theoretical results on a class of bounding volume hierarchies called BOXTREE have also been ....

G. Zachmann, "Rapid Collision Detection by Dynamically Aligned DOP-Trees," Proceedings of IEEE Virtual Reality Annual International Symposium (VRAIS '98), Atlanta, Georgia, March 1998.

....that in parallel close proximity situations, even 26 dops are not as tight as OBBs. 2. The individual node update cost Cu is high. Even with the k dop tumbling strategy, the average Cu is more than 10 times than Cv , as reported in [10] The dynamic node update method proposed by Zachmann [22] actually only works for some special cases in 3D, and the algorithm developed by Konecny [11] requires an expensive one time preprocessing (Co ) In all these solutions, Cu also linearly increases with k. 3. The implementation of k dops in [10] partially depends on the assumption that one ....

G. Zachmann. Rapid collision detection by dynamically aligned dop-trees. In Proc. of IEEE, VRAIS'98, Atlanta, Georgia, March 1998.

....can be excluded in many cases and a lower bound of their distance can be obtained. These envelopes, which are also called bounding volumes, can be spheres [Qui94, PG95, Hub96] spherical shells [KPLM98] axis aligned boxes [ZF95] oriented boxes [GLM96] or xed directions hulls [HKM 96, Zac98] For reasons of simplicity virtual objects are often modelled as rigid bodies, which are not subject to any defor mations when they collide. A popular method to calculate the forces acting during a collision consists in using springs. If two objects are going to interpenetrate, a spring force ....

....represents a part of the object that we want to approximate by a simpler shape, the so called bounding volume. In SiLVIA several types of bounding volumes are available: spheres [Hub95] arbitrarily) oriented bounding boxes (OBBs) GLM96] and xed directions hulls (FDHs or k DOPs) HKM 96] Zac98] Axisaligned bounding boxes (AABBs) are a special case of the latter. We build the tree in a top down approach. In doing so, we recursively divide the original face set by creating balanced partitions and compute tight bounding volumes on every tree node. The recursion terminates if no ....

G. Zachmann. Rapid collision detection by dynamically aligned DOP-trees. In Proc. of IEEE, VRAIS'98 Atlanta, March 1998.

No context found.

Gabriel Zachmann. Rapid collision detection by dynamically aligned DOP-trees. In Proc. of IEEE Virtual Reality Annual International Symposium; VRAIS '98. Atlanta, Georgia, March 1998.

No context found.

Gabriel Zachmann. Rapid collision detection by dynamically aligned DOP-trees. In Proc. of IEEE Virtual Reality Annual International Symposium; VRAIS '98, pages 90--97, Atlanta, Georgia, March 1998.

No context found.

G. Zachmann. Rapid collision detection by dynamically aligned DOP-trees. In Proc. VRAIS 1998.

No context found.

G. Zachmann. Rapid collision detection by dynamically aligned DOP-trees. In Proc. VRAIS 1998.

No context found.

Gabriel Zachmann. Rapid Collision Detection by Dynamically Aligned DOP-Trees. In Proc. of IEEE Virtual Reality Annual International Symposium; VRAIS '98, pages 90--97. Atlanta, Georgia, March 1998. 11

No context found.

G. Zachmann. Rapid collision detection by dynamically aligned DOP-trees. In Proc. VRAIS 1998.

No context found.

Gabriel Zachmann. Rapid collision detection by dynamically aligned DOP-trees. In Proc. of IEEE Virtual Reality Annual International Symposium; VRAIS '98, pages 90--97, Atlanta, Georgia, March 1998. 3, 5, 10

....like to know which algorithm to use at the back end. Maybe there is no single best algorithm, and di#erent algorithms perform best for di#erent complexities or types of objects. In this section, we will compare three algorithms: OBB trees [GLM96] BV trees of k DOPs [KHM 98] and DOP Trees [Zac98] The latter two use the same type of bounding volume, but di#erent algorithms for checking the overlap of two BVs. The former uses 3 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 10 20 30 40 50 60 70 complexity kilo polygons quickcd # # # # rapid dop tree # # # # # ....

....# # # # Figure 3: Results for the suite of benchmark object as shown in Figure 1 (headlight, door lock, car body, hose, sphere, hyperboloid, torus) All times have been obtained on a R10000 194MHz, averaging over distance, orientation, and object frame. oriented bounding boxes. In contrast to [Zac98] we have determined 24 as the optimal number of orientations. For the comparison, we used the implementations Rapid and QuickCD [Got97, KHM99] We have found, that it is extremely di#cult to compare collision detection algorithms, because in general they are very sensitive to conditions and ....

Gabriel Zachmann. Rapid collision detection by dynamically aligned DOP-trees. In Proc. of IEEE Virtual Reality Annual International Symposium; VRAIS '98, pages 90--97. Atlanta, Georgia, March 1998.

....so that the object can glide along other parts; in our earlier implementation, there was no gliding, which caused the object on the lead to get stuck in tight environments. So, at any time it can assume only valid positions. Of course, exact and fast collision detection is a prerequisite (Zachmann, 1998). This is only a first step. A completely reliable verification will check the virtual hand for collisions as well. Also, the hand and or part should slide along smooth rigid objects to make assembly easier for the user. Feedback to the user. Any VR system should be as responsive as possible, ....

Zachmann, G. (1998). Rapid collision detection by dynamically aligned dop-trees. In Proc. of ieee virtual reality annual international symposium; vrais '98. Atlanta, Georgia.

....on a SGI R10000 194 MHz. All times are (unless otherwise noted) in millseconds. 2 Related work Research so far has mainly focused on algorithms for the collision detection problem, given a pair of objects. Hierarchical algorithms for polygonal objects have been presented by [GLM96, KHM 98, Zac98, Hub95, OD99, NAT90] Non hierarchical approaches have been presented by [MPT99, Gei00] Incremental convex algorithms have been presented by [CLMP95, vdB99, Chu96] Algorithms for reducing the number of pairs to be tested have been presented by [CLMP95, BF79, PS90, YW93, Hub93] A system solving ....

....like to know which algorithm to use at the back end. Maybe there is no single best algorithm, and different algorithms perform best for different complexities or types of objects. In this section, we will compare three algorithms: OBB trees [GLM96] BV trees of k dops [KHM 98] and DOP Trees [Zac98] The latter two use the same type of bounding volume, but different algorithms for checking the overlap of two BVs. The former uses oriented bounding boxes. In contrast to [Zac98] we have determined 24 as the optimal number of orientations. For the comparison, we used the implementations Rapid ....

[Article contains additional citation context not shown here]

Zachmann, G. Rapid collision detection by dynamically aligned DOP-trees. In Proc. of IEEE Virtual Reality Annual International Symposium; VRAIS '98. Atlanta, Georgia, March 1998.

....16. Annotations can be put into the scene by voice commands. Fig. 17. Violations of security distance are highlighted by yellow, collisions are red. stuck in tight environments. So, at any time it can assume only valid positions. Of course, exact and fast collision detection is a prerequisite [18]. This is only a first step. A completely reliable verification will check the virtual hand for collisions as well. Also, the hand and or part should slide along smooth rigid objects to make assembly easier for the user. Documentation. If a certain assembly task cannot be done, then the result ....

G. Zachmann. Rapid collision detection by dynamically aligned dop-trees. In Proc. of IEEE Virtual Reality Annual International Symposium; VRAIS '98, Atlanta, Georgia, Mar. 1998.

No context found.

ZACHMANN G.: Rapid collision detection by dynamically aligned DOP-trees. In Proc. of IEEE Virtual Reality Annual International Symposium; VRAIS '98 (Atlanta, Georgia, Mar. 1998), pp. 90--97. 121

No context found.

G. Zachmann. Rapid collision detection by dynamically aligned DOP-trees, Proceedings of IEEE Virtual Reality Annual International Symposium, 90--97, 1998 1, 2

No context found.

G. Zachmann. Rapid collision detection by dynamically aligned dop-trees. In Proc. of IEEE Virtual Reality Annual International Symposium; VRAIS 98, 1998.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC