A robot vision VLSI processor for the rectangular solid representation of three-dimensional objects

Abstract

A compact rectangular solid representation of three-dimensional (3-D) objects is proposed for motion planning of intelligent robots. A VLSI-oriented algorithm that transforms the discrete point representation of a 3-D object to a rectangular solid representation is considered. A new architecture is proposed to perform matching operations between a rectangular solid and discrete points in parallel. Performance evaluation shows that the VLSI processor that implements the algorithm is 25,000 times faster than a RAM-based architecture. © 1997 Scripta Technica, Inc. Syst Comp Jpn, 28(3): 54–61, 1997