## Collision Detection for Interactive Graphics Applications (1995)

### Download From

IEEE### Download Links

- [www.graphics.cornell.edu]
- [ftp.cs.brown.edu]
- DBLP

### Other Repositories/Bibliography

Venue: | IEEE Transactions on Visualization and Computer Graphics |

Citations: | 186 - 5 self |

### BibTeX

@ARTICLE{Hubbard95collisiondetection,

author = {Philip M. Hubbard and Philip Martyn Hubbard and Philip Martyn Hubbard},

title = {Collision Detection for Interactive Graphics Applications},

journal = {IEEE Transactions on Visualization and Computer Graphics},

year = {1995},

volume = {1},

pages = {218--230}

}

### Years of Citing Articles

### OpenURL

### Abstract

Solid objects in the real world do not pass through each other when they collide. Enforcing this property of "solidness" is important in many interactive graphics applications; for example, solidness makes virtual reality more believable, and solidness is essential for the correctness of vehicle simulators. These applications use a collision-detection algorithm to enforce the solidness of objects. Unfortunately, previous collision-detection algorithms do not adequately address the needs of interactive applications. To work in these applications, a collision-detection algorithm must run at real-time rates, even when many objects can collide, and it must tolerate objects whose motion is specified "on the fly" by a user. This dissertation describes a new collision-detection algorithm that meets these criteria through approximation and graceful degradation, elements of time-critical computing. The algorithm is not only fast but also interruptible, allowing an application to trade accuracy ...

### Citations

11403 |
Computers and Intractability: A Guide to the Theory of NP-Completeness
- Garey, Johnson
- 1979
(Show Context)
Citation Context ...y one node on that edge is in the subset. The postprocess should find as large an independent set as it can, corresponding to removing as much redundancy as possible. Unfortunately, Garey and Johnson =-=[GJ79]-=- show that finding the largest independent set of a graph is an NP-complete problem. A tractable alternative is to find an approximation to the largest independent set. The simplest approach is for th... |

1842 |
2] Computational Geometry: An Introduction
- Preparata, Shamos
- 1985
(Show Context)
Citation Context ...one or other cutting plane. The algorithm applies these ideas as it finds intersections between all the 2D segments. It finds these segment intersections by using the technique of Bentley and Ottmann =-=[21]-=-. This technique sweeps a line across the ff-t plane (from t = 0 to t = 1), reporting every intersection it finds. The broad phase needs the earliest (lowest t) segment intersection that corresponds t... |

1075 |
Computer Graphics Principles and Practice
- Foley, Fein, et al.
- 1996
(Show Context)
Citation Context ...ntact one another produce many special cases which are difficult to process efficiently. Other forms of surface representation also pose problems. Curved surfaces, such as parametric bicubic surfaces =-=[FvFH90]-=-, can be even more difficult to process, and pair-processing algorithms for these surfaces often resort to approximating the surfaces with polyhedra. A detection algorithm suffers from the pair-proces... |

588 | Decimation of triangle meshes
- SCHROEDER, ZARGE, et al.
- 1992
(Show Context)
Citation Context ...lark [Cla76] presents one of the earliest discussions of levels of detail, but gives no specific techniques for generating the levels. More recently, authors such as Turk [Tur92] and Schroeder et al. =-=[SZL92]-=- have described algorithms that produce these levels of detail automatically. Hoppe et al. [HDD + 93] describe what seems to be the most sophisticated approach to date. The algorithm generates each le... |

519 |
An analysis of timedependent planning
- Dean, Boddy
- 1998
(Show Context)
Citation Context ... introduced this idea in image rendering, calling it adaptive refinement. Cohen et al. [CCWG88] called this idea progressive refinement in the context of another approach to rendering. Dean and Boddy =-=[DB88]-=- first applied this idea to planning problems, calling the implementation an anytime algorithm. Van Dam [van93] uses the term negotiated graceful degradation for this class of techniques, and consider... |

518 |
A transformation for extracting new descriptors of shape
- Blum
- 1967
(Show Context)
Citation Context ...mulated-annealing algorithm from Section 5.3, such as the method for checking whether a set of spheres conservatively cover the object. 6.1 Medial-Axis Surfaces The medial axis was introduced by Blum =-=[Blu67] to descri-=-be the shapes of biological objects. He defines the medial axis of a two-dimensional (2D) shape by various physical analogies, including the "grassfire analogy": consider igniting simultaneo... |

417 | Adaptive Display Algorithm for Interactive Frame Rates During Visualization of Complex Virtual Environments
- Funkhouser, Séquin
(Show Context)
Citation Context ..., NY 14853-3801. Email address: pmh@graphics.cornell.edu. This approach is analogous, in essence, to rendering at multiple levels of quality, as explored by Bergman et al. [2], Funkhouser and S'equin =-=[8]-=-, and Maciel and Shirley [17]. The success of these rendering techniques suggests the value of this approach in collision detection. All these ideas are elements of time-critical computing, which stre... |

408 | Re-Tiling Polygonal Surfaces
- Turk
- 1992
(Show Context)
Citation Context ... of detail for each object. Clark [Cla76] presents one of the earliest discussions of levels of detail, but gives no specific techniques for generating the levels. More recently, authors such as Turk =-=[Tur92]-=- and Schroeder et al. [SZL92] have described algorithms that produce these levels of detail automatically. Hoppe et al. [HDD + 93] describe what seems to be the most sophisticated approach to date. Th... |

404 |
Simulated Annealing: Theory and Applications
- Laarhoven, Aarts
- 1987
(Show Context)
Citation Context ...� gradually decreases, the algorithm should arrive at a solution close to the global minimum for the problem. This general approach seems to work well for many practical problems; Laarhoven and Aart=-=s [LA87]-=-, for example, discuss several applications of simulated annealing to layout problems in VLSI design, and Press et al. [PTVF92] describe how simulated annealing generates an approximate solution to th... |

333 | On visible surface generation by a priori tree structures
- Fuchs, Kedem, et al.
- 1980
(Show Context)
Citation Context ... Turk's algorithm is likely to make it faster in many situations. Binary space partitioning (BSP) trees provide a different form of space subdivision. These structures were introduced by Fuchs et al. =-=[FKN80]-=- as a way to remove hidden surfaces during polygon rendering. Thibault and Naylor [TN87] discuss how to use BSP trees to perform Boolean set operations on two polyhedra. Their algorithm finds intersec... |

316 |
Hierarchical geometric models for visible surface algorithms
- Clark
- 1976
(Show Context)
Citation Context ...the benefits of rendering while meeting a cost goal, that is, keeping the frame rate high and nearly-constant. This algorithm assumes that it has access to the levels of detail for each object. Clark =-=[Cla76]-=- presents one of the earliest discussions of levels of detail, but gives no specific techniques for generating the levels. More recently, authors such as Turk [Tur92] and Schroeder et al. [SZL92] have... |

297 |
Iris performer: A high performance multiprocessing toolkit for real-time 3d graphics
- Rohlf, Helman
- 1994
(Show Context)
Citation Context ...m in a complete application. An application that uses multiple time-critical algorithms will be simpler to implement in a software framework that explicitly supports such algorithms. Rohlf and Helman =-=[RH94]-=- describe one of the first commercial software packages to strive for this goal. The system supports scene data structures that explicitly represent the geometry of objects at multiple levels of detai... |

288 | ICOLLIDE: An Interactive and Exact Collision Detection System for Large-Scale Environments
- Cohen, Lin, et al.
- 1995
(Show Context)
Citation Context ...e approach works for the bounding spheres or boxes of arbitrary objects, efficiently reducing the number of objects whose surfaces must be tested for intersection (by some other method). Cohen et al. =-=[4]-=- use simple spatial sorting to find intersections between bounding boxes; this sorting exploits interframe coherence to reduce its workload. Although these algorithms do not address the fixed-timestep... |

281 | Algorithms for reporting and counting geometric intersections - BENTLEY, OTTMANN - 1979 |

261 |
Introduction to artificial intelligence
- Charniak, McDermott
- 1985
(Show Context)
Citation Context ...n the currently-closest face. This approach is analogous to minimax q f p p v Figure 5.13: When p projects onto the vertex at v, d(p; v) ! d(p; q). search, which is popular in artificial intelligence =-=[CM86]-=-. The order in which the algorithm visits the cubes affects the number of faces the optimization prunes. The algorithm should start in the cube containing p, or the cube closest to p. It should then p... |

250 | A progressive refinement approach to fast radiosity image generation
- Cohen, Chen, et al.
- 1988
(Show Context)
Citation Context ...ality with time appears by various names in other areas of graphics and computer science. Bergman et al. [BFGS86] introduced this idea in image rendering, calling it adaptive refinement. Cohen et al. =-=[CCWG88]-=- called this idea progressive refinement in the context of another approach to rendering. Dean and Boddy [DB88] first applied this idea to planning problems, calling the implementation an anytime algo... |

242 | Collision detection and response for computer animation
- Moore, Wilhelms
- 1988
(Show Context)
Citation Context ...thm in practice. Space subdivision is a popular way to mitigate the all-pairs weakness (although\Omega\Gamma N 2 ) worst cases still exist for some object shapes). The algorithms of Moore and Wilhems =-=[18]-=-, Shaffer and Herb [25] and Smith et al. [26] exemplify the use of octrees by many researchers. Turk [29] presents a simpler technique that divides space uniformly; while Turk's application was molecu... |

240 | Generating Textures on Arbitrary Surfaces Using Reaction-Diffusion - Turk - 1991 |

238 |
Shape description using weighted symmetric axis features
- Blum, Nagel
- 1978
(Show Context)
Citation Context ...ints on the medial axis are equidistant from two sides of the boundary, so they represent a symmetrical axis or "skeleton." Figure 6.1 shows an example of a medial axis. In later work, Blum =-=and Nagel [BN78] state ano-=-ther definition of the medial axis, which they also call the "symmetric axis": the locus of the centers of maximal circles that fit inside the figure, with no circle containing another. They... |

236 | Towards image realism with interactive update rates in complex virtual building environments
- Airey, Rohlf, et al.
- 1990
(Show Context)
Citation Context ...pplication producing an animation, an important element of real-time performance is maintaining a high and nearly-constant frame rate. Opinions vary on the minimum acceptable frame rate. Airey et al. =-=[ARB90]-=- state that a rate of six frames per second (fps) is barely tolerable for building-walkthrough applications, while Funkhouser and S'equin [FS93] suggest 10 fps; in general, Brooks [Bro88] and Fuchs an... |

215 | ªVisual Navigation of Large Environments Using Textured Clusters,º
- Maciel, Shirley
- 1995
(Show Context)
Citation Context ...ss: pmh@graphics.cornell.edu. This approach is analogous, in essence, to rendering at multiple levels of quality, as explored by Bergman et al. [2], Funkhouser and S'equin [8], and Maciel and Shirley =-=[17]-=-. The success of these rendering techniques suggests the value of this approach in collision detection. All these ideas are elements of time-critical computing, which stresses that the time an algorit... |

202 | Collision detection between geometric models: a survey
- Lin, Gottschalk
- 1998
(Show Context)
Citation Context ... [7] use 4D structures somewhat similar to space-time bounds to solve the fixed-timestep weakness. Their approach also addresses the all-pairs weakness in some---but not all--- situations. Lin et al. =-=[16]-=- use similar 4D analysis in concert with an efficient technique for tracking the distance between two convex polyhedra. The algorithm incorporates extensions for nonconvex objects, but the authors do ... |

199 | Analytical methods for dynamic simulation of non-penetrating rigid bodies - Baraff - 1989 |

177 |
Computing Dirichlet Tesselations
- Bowyer
- 1981
(Show Context)
Citation Context ... and Accurate Voronoi Diagrams in Three Dimensions Given a distribution of points on the object's surface, the Goldak et al. algorithm [GYKD91] next builds a Voronoi diagram from these points. Bowyer =-=[Bow81]-=- describes a relatively straightforward algorithm for the Voronoi diagram of 3D points. This algorithm implicitly assumes infinite-precision arithmetic, however. Finite-precision arithmetic causes the... |

176 |
Superquadrics and angle-preserving transformations
- Barr
- 1981
(Show Context)
Citation Context ...ields high performance. Sclaroff and Pentland [SP91] describe a novel solution to the pair-processing weakness. They suggest that polyhedral models be approximated by deformed superquadric ellipsoids =-=[Bar81]-=- whose surfaces have been modulated by a displacement map. These approximate models, which they call generalized implicit functions, have the advantage that the algorithm to detect collisions between ... |

169 | Realistic animation of rigid bodies
- Hahn
- 1988
(Show Context)
Citation Context ...thms, but these algorithms cannot meet the demands of many interactive applications. The main problem is that the algorithms are too slow, making real-time performance impossible. As an example, Hahn =-=[Hah88]-=- describes a simulation system that uses a relatively sophisticated detection algorithm, an algorithm that people would still consider using today. Hahn reports the performance of this system both wit... |

168 |
Automatic Creation of Object Hierarchies for Ray Tracing
- Goldsmith, Salmon
- 1987
(Show Context)
Citation Context ...ibing each occupied octant with a sphere. The resolution of the resulting sphere-tree increases with depth; even so, sphere-trees produced this way tend to fit an object loosely. Goldsmith and Salmon =-=[11]-=- present an algorithm that builds efficient bounding hierarchies for ray tracing. Optimizing a hierarchy for ray-object intersections does not directly promote the tightness needed for object-object c... |

168 | Improving static and dynamic registration in a optical see-through HMD
- Azuma, Bishop
- 1994
(Show Context)
Citation Context ...an compensate for latency in a tracking device is by predicting the values it will return in the immediate future. Liang et al. [LSG91], Friedmann et al. [FSP92], Deering [Dee92] and Azuma and Bishop =-=[AB94]-=- argue that techniques such as linear interpolation and Kalman filtering give acceptable prediction in some situations. Getting prediction to work in all situations is still a research problem, but fo... |

164 | J.F.Canny, "A fast algorithm for incremental distance calculation
- Lin
- 1991
(Show Context)
Citation Context ... nonconvex polyhedra given a decomposition of these polyhedra into convex pieces. The number of pairs of these pieces could be overwhelming for objects with complicated shapes, however. Lin and Canny =-=[LC91]-=- describe an efficient algorithm for finding the closest distance between two convex polyhedra. The algorithm is based on the following observation: when two convex polyhedra move only a small amount ... |

144 | Curved surfaces and coherence for non-penetrating rigid body simulation - Baraff - 1990 |

137 | Interval analysis for computer graphics
- Snyder
- 1992
(Show Context)
Citation Context ...e general motion, the authors suggest an approach based on interval analysis. Mudur and Koparkar [MK84] describe one of the first applications of interval analysis to graphics, but the work of Snyder =-=[Sny92]-=-, Duff [Duf92] and Snyder et al. [SWF + 93] is more relevant for collision detection. Interval analysis involves extending mathematical functions to take intervals as arguments and produce intervals a... |

136 |
High resolution virtual reality
- DEERING
- 1992
(Show Context)
Citation Context ...the only way an application can compensate for latency in a tracking device is by predicting the values it will return in the immediate future. Liang et al. [LSG91], Friedmann et al. [FSP92], Deering =-=[Dee92]-=- and Azuma and Bishop [AB94] argue that techniques such as linear interpolation and Kalman filtering give acceptable prediction in some situations. Getting prediction to work in all situations is stil... |

128 |
Set operations on polyhedra using binary space partitioning trees
- Thibault, Naylor
- 1987
(Show Context)
Citation Context ...ossible collision is here 1 2 3 4 5 Broad phase: next possible collision is here Broad phase: bounding spheres do not collide Figure 2: The broad phase and narrow phase in action. Thibault and Naylor =-=[28]-=- show how binary space partitioning (BSP) trees address the pair-processing weakness for polyhedral objects. An advantage of BSP trees is that for rigid objects, the structure of a tree can be optimiz... |

126 | Dynamic simulation of non-penetrating flexible bodies - Baraff, Witkin - 1992 |

109 |
Collision detection for moving polyhedra
- Canny
- 1986
(Show Context)
Citation Context ... objects, making these algorithms unsuitable for interactive applications. To eliminate the fixed-timestep weakness, some algorithms use 4D geometry, the fourth dimension being simulation time. Canny =-=[3]-=-, for example, derives functions of a time variable that express how convex polyhedra change over time; the roots of these functions denote collisions. Duff [6] uses interval analysis to quickly ident... |

106 | Multiresolution Modeling For Fast Rendering
- Heckbert, Garland
(Show Context)
Citation Context ...multiple hierarchy levels, each using different numbers of spheres to approximate an object at different resolutions. Multiresolution modeling is a difficult topic in general, as Heckbert and Garland =-=[12]-=- discuss in the context of rendering. For collision detection, the particular challenge is making each level fit the object tightly; tightness is necessary because the application could stop the detec... |

99 |
Interval arithmetic and recursive subdivision for implicit functions and constructive solid geometry
- Duff
- 1992
(Show Context)
Citation Context ... dimension being simulation time. Canny [3], for example, derives functions of a time variable that express how convex polyhedra change over time; the roots of these functions denote collisions. Duff =-=[6]-=- uses interval analysis to quickly identify and refine the regions in space and time that could contain collisions. This approach is one of the few to address all three weaknesses. Unfortunately, it a... |

88 | On Temporal-Spatial Realism in the Virtual Reality Environment
- Liang, Shaw, et al.
- 1991
(Show Context)
Citation Context ...three dimensions. With current technology, the only way an application can compensate for latency in a tracking device is by predicting the values it will return in the immediate future. Liang et al. =-=[LSG91]-=-, Friedmann et al. [FSP92], Deering [Dee92] and Azuma and Bishop [AB94] argue that techniques such as linear interpolation and Kalman filtering give acceptable prediction in some situations. Getting p... |

86 | Interactive collision detection
- Hubbard
- 1993
(Show Context)
Citation Context ...imations users find most appropriate for particular tasks. 11 Acknowledgements A preliminary, abbreviated version of this paper appeared at the IEEE Symposium on Research Frontiers in Virtual Reality =-=[13]. This res-=-earch was conducted as part of the author's doctoral studies at Brown University. The guidance of John "Spike" Hughes was essential to the completion of this work, and Andy van Dam and the r... |

85 | Coping with friction for non-penetrating rigid body simulation. Computer Graphics 25(4):31–40
- Baraff
- 1991
(Show Context)
Citation Context ...ould penetrate. Both parts of a collision-handling algorithm pose interesting problems, but this paper focuses on detection algorithms. For a discussion of response algorithms, see the work of Baraff =-=[1]-=-. Despite the wealth of literature on detection algorithms, no published algorithms adequately address the needs of interactive applications, such as vehicle simulators or virtual reality. To work in ... |

82 | Collision detection by four-dimensional intersection testing
- Cameron
- 1990
(Show Context)
Citation Context ...f all objects must be completely specified throughout simulation time. Interactive applications cannot meet this requirement, because a human user controls the motion of objects "on the fly."=-=; Cameron [Cam90]-=- presents an algorithm that handles more complicated shapes than Canny's approach, in particular, objects specified by the constructive solid geometry (CSG) paradigm. This algorithm is a 4D extension ... |

80 |
Image rendering by adaptive refinement
- Bergman, Fuchs, et al.
- 1986
(Show Context)
Citation Context ..., Cornell University, Ithaca, NY 14853-3801. Email address: pmh@graphics.cornell.edu. This approach is analogous, in essence, to rendering at multiple levels of quality, as explored by Bergman et al. =-=[2]-=-, Funkhouser and S'equin [8], and Maciel and Shirley [17]. The success of these rendering techniques suggests the value of this approach in collision detection. All these ideas are elements of time-cr... |

80 |
Fast detection of polyhedral intersection
- Dobkin, Kirkpatrick
- 1983
(Show Context)
Citation Context ...algorithm based on BSP trees, but I chose to use BSP trees in Chapter 7 for simplicity of implementation. 2.3.3 Other Methods To detect collisions between two convex polyhedra, Dobkin and Kirkpatrick =-=[DK83]-=- present an algorithm that requires O(log 2 n) operations, with n being the sum of the number of vertices in the two polyhedra. This algorithm uses the decomposition of each polyhedron into drums, whi... |

64 | Generalized implicit functions for computer graphics
- Sclaroff, Pentland
- 1991
(Show Context)
Citation Context ...r pair-processing algorithm, Garcia-Alonso et al. [9] associate a precomputed regular grid with each object; as objects rotate, however, the grids lose some of their efficiency. Sclaroff and Pentland =-=[24]-=- present a pair-processing algorithm that approximates surfaces with deformed superquadrics. The authors do not demonstrate that their approach works for a wide variety of surfaces, but they are the f... |

64 | Interval methods for multi-point collisions between time dependent curved surfaces
- Snyder
- 1993
(Show Context)
Citation Context ...assumes that objects' motions over time are fully known in advance. Thus, an interactive application cannot use this algorithm. Recent advances in interval analysis, such as the work of Snyder et al. =-=[27]-=-, still retain this disadvantage. Variations on these algorithms would apply to 4D geometry that estimates where objects could be in the future without assuming fully pre-specified motion; this idea i... |

62 |
Solving the collision detection problem
- Garica-Alonso, Flaquer
- 1994
(Show Context)
Citation Context ...id objects, the structure of a tree can be optimized to improve performance; Naylor [19] describes several optimization techniques. To accelerate their pair-processing algorithm, Garcia-Alonso et al. =-=[9]-=- associate a precomputed regular grid with each object; as objects rotate, however, the grids lose some of their efficiency. Sclaroff and Pentland [24] present a pair-processing algorithm that approxi... |

53 |
Hierarchical Data Structures and Algorithms for Computer Graphics II: Applications
- Samet, Webber
- 1988
(Show Context)
Citation Context ...niques for building tight sphere-trees; more details appear elsewhere [14]. The literature contains few references on tight hierarchies. An algorithm that builds an octree representation of an object =-=[23]-=- can also build a sphere-tree, by circumscribing each occupied octant with a sphere. The resolution of the resulting sphere-tree increases with depth; even so, sphere-trees produced this way tend to f... |

50 |
Three-dimensional shape description using the symmetric axis transform I:theory
- Nackman, Pizer
- 1985
(Show Context)
Citation Context ...al circles that fit inside the figure, with no circle containing another. They also describe how quantitative properties of a medial axis relate to properties of the figure's shape. Nackman and Pizer =-=[NP85]-=- discuss similar properties for medial-axis surfaces of three-dimensional (3D) objects. The definition of a 3D medial-axis surface is simple extension of the 2D definition: the locus of centers of max... |

46 |
Interference detection among solids and surfaces
- Boyse
- 1979
(Show Context)
Citation Context ...ed-timestep weakness, many authors use four-dimensional (4D) analysis, where the fourth dimension explicitly represents simulation time. One of the earliest examples of this idea is the work of Boyse =-=[Boy79]-=-. He derives simple equations for the motion of polyhedral edges and faces over time; a root of any such equation corresponds to the time and location of a collision. By assuming only one stationary a... |

46 |
Geometric collisions for time-dependent parametric surfaces
- Herzen, Barr, et al.
- 1990
(Show Context)
Citation Context ...s the fixed-timestep weakness, but only for piecewise-linear motion that is fully specified throughout time. To detect collisions between arbitrary time-varying parametric surfaces, Von Herzen et al. =-=[VBZ90]-=- use the mathematical technique of the Lipshitz condition. A Lipshitz condition maps a change in the parameters of a surface to an upper bound on the change in the surface itself. By applying this map... |