Abstract: In the development of robot with vision capability, representation and shape rccognition of three dimensionalobjects are of crucial importance. 3-D model plays a very important role in 3-D object and scene understanding. 3-D model should beobject-centered, providing the most useful measure of scene. The size, shape, and orientation of objects can be derived from it. This paper proposes a 3-D GORC model, i.e.,a Generalized Object-oriented Runlength Coding model. Itconsists of a GORC physical data structure, detailed shape description and abstract description. 3-D GORC is anextended one of the 2-D object-oriented runlength coding. GORC 3-D model can be easily constructed fromdepthmap obtained directly by range finder with phase-modulated light. It has both advantages of volume repre-sentation and surface representation. Projection onto 2-D space, image algebraic operations and feature extractioncan be performed on GORC coded 3-D object efficiently.