Добавил:
Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз: Предмет: Файл:
(EOD).Mechatronics.pdf
Скачиваний:
82
Добавлен:
23.08.2013
Размер:
5.07 Mб
Скачать

page 729

prisms. This method is suggested as a possible precursor for the use of video information about the workcell, from a high level vision system. This has two points of interest; the objects which are suspended from the side will be grossly misrepresented, but this form of encription suits video cameras well. The assumption that the manipulator may be treated as a set of 4 d.o.f. is based on the limitation of the problem to only pick and place operations and insertion (or fitting) operations. This unfortunately means that the objects may only be rotated about the vertical axis when in motion.

The use of Freeways between obstacles allows a choice between alternate paths.

Figure B.5 Freeway Between Blocks

The freeways are basis for maps of joint configurations which are acceptable for motion through these freeways. The methods then find the path using link constraints. This method is described in algorithm form, and the algorithms are quite substantial.

40.10.5 SPATIAL PLANNING : OCT-TREE

In a paper by T.Soetadji [1986] a method for 3D path planning by a mobile robot is suggested. The method is based upon use of an Oct-tree representation of 3D space. In the paper, the development of the Oct-Tree routines is discussed, and a robotic system for implementation is suggested. Once the Oct-tree is set up, an A* or a Breadth First search is used to find the best path for the mobile robot. This method finds the minimum distance (with collision avoidance) on a VAX 750. The search time for the path is on the order of 1 second. The tree structure also proves very efficient, because it had only occupied 1.7 MBytes of memory for a very complicated environment.

page 730

40.10.6 SPATIAL PLANNING : VORONOI DIAGRAMS

A newer approach to representing space has been done with Voronoi Diagrams. O.Takahashi and R.J.Schilling [1989] have suggested such a method. Using an environment of polygons, the pathways may be represented with Voronoi diagrams, then represented in a graph.

Figure B.6 Voronoi Diagram of Simple Work Space

Mid-Points Between Obstacles

After the Voronoi diagram has been set up in graph form, a path may be found. This is simplified by the use of some heuristic rules for wide paths, tight bends, narrow gaps, and reversing, which identify a number of orientations. This procedure produces short smooth paths (which avoid obstacles) for 2D objects on an IBM compatible computer with no co-processor in 10 seconds to 1 minute. This method has potential for use with vision systems. Algorithms suggested by A.C.Meng [1988] allow for fast udate of Voronoi diagrams, in a changing environment. This makes the operation much faster, by avoiding the complete reconstruction of the diagram, and makes real time trajectory correction feasible.

40.10.7 SPATIAL PLANNING : GENERAL INTEREST

E.G.Gilbert and D.W.Johnson [1985] created an optimization approach to path planning (with the piano movers problem), with rotations and collision avoidance. This method runs 10 to 20 minutes on a Harris-800 computer. This method was based on work which was later published by E.G.Gilbert, D.W.Johnson, and S.S.Keerthi [1987] which provides algorithms for calculating distances between convex (and concave) hulls in a very short time (order of milliseconds).

40.10.8 SPATIAL PLANNING - VGRAPHS

A method proposed by K.Kant and S.Zucker [1988] involves collision avoidance of rectangles based upon the search of a VGRAPH. This method also suggests the use of a Low Level controller Collision Avoidance Module. This controller would use low level information about the

Соседние файлы в предмете Электротехника