Contact Modeling and Collision Detection in Human Joints - 3D Multiscale Physiological Human - page 183

Fig. 8.1 Sliding and striking collision during rotation. The light green object is rotating about its
center and can collide with the dark blue object in two different ways: 1-sliding, 2-striking [ 13 , 14 ]
(Reprinted from Journal of Biomechanics, Vol. 42, Arbabi E, Boulic R, Thalmann D, Fast collision
detection methods for joint surfaces, pp. 91-99, Copyright (2009), with permission from Elsevier)
the style of movement (rotation or sliding), and consequently improve the efficiency
of the collision detection for the rotational cases. The methods are tested in different
scenarios and compared with some of the previous methods (including general and
specific ones), where the comparisons show the efficiency of the proposed ones.
When an object is rotating, two kinds of collision may happen: 1-sliding or
2-striking. In the sliding case the colliding area is almost parallel to the rotational
trajectory and the penetration is usually considered in the radial direction. On the
other hand, in the striking case, the objects' colliding area is almost perpendicular
to the rotational trajectory and the penetration is usually considered in the angular
direction (Fig. 8.1 ).
8.2.2 Cylindrical Segmenting Collision Detection
Arbabi et al. proposed a novel fast method suitable for rotational strike, based on
discretizing the space in the cylindrical orientation [ 14 ]. This kind of segmentation
helps the method to be more adjusted to the style of movement (rotation), and con-
sequently improve the efficiency of the collision detection for the rotational cases.
The cylindrical segmentations are done in the same orientation of rotation which
does not only increase the speed of rigid collision detection but also increases the
speed of updating for deformable collision detection (in angular direction). The cor-
responding ring-shaped segments (see Fig. 8.2 ) are optimal for conducting collision
detection for the associated axis of rotation vs. cube shaped cells (e.g. [ 17 ]).
After the cylindrical segmentation of the space (around the fixed object) is done, a
table for the fixed object, based on the cylindrical segmentation, is created. Then, the
list of the fixed polygons is stored in the corresponding table cells. By comparing the
position of each mobile vertex with the fixed polygons stored in the corresponding