The 5th International S y mposium on Robotics in Construction June 6-8 ,1988 Tokyo, Ja p an Geometric Reasoning in Computer Integrated Building Construction Robert F. Woodbury ,' Steven J . Fenves," Nelson C. Baker ,"' Richard W. Quadrel , ° Clauss B. Strauch,° Carnegie Mellon University Pittsburgh , Pa. 15213 tel. (412) 268-8853 email rw@ cad.cs.cmu.edu ABSTRACT Geometric reasoning, the integration of geometric representation and inference in advanced computer sys- tems, is presented as an issue at the forefront of research in construction automation. The unique demands that construction automation poses on such reasoning are discussed. An architecture that provides a structure for geometric reasoning is presented and results from a prototype implementation are shown. A project to develop geometric reasoning in the construction domain of panelized building systems is introduced. Within this project, two exemplary applications, structural/architectural design and construction sequence planning, each supported by the same geometric reasoning facility, are being demonstrated. 1. INTRODUCTION A most challenging issue for current construction automation research is representation of and reasoning on geometric information. Geometry is ubiquitous in construction and reasoning upon it is an essential con- stituent of the major intellectual issues in the construction domain. With increasing automation of the con- struction industry, from design to fabrication, comes a need for capable geometric facilities to underly new computer based systems for representing, designing, communicating, planning, and acting. This need becomes all the more acute when the integration of many programs into a large-scale construction system is con- templated. The rich exchange of information in any integrated system appears to demand that the various components can communicate with common representations, most of which will be largely phrased in geometric terms. Future construction systems will need to distinguish themselves by several characteristics: • Multiple Levels of A bstraction. Effective problem-solving within the construction domain, be it physical design, construction planning, or robot motion planning, is usually hierarchical. Problems are initially solved at a high level of abstraction and then refined to consider increasing levels of detail. This top-down approach provides: a reduction of complexity, an ability to reason with incomplete information, and an ability to divide the problem among specialists. It is very useful if the representations employed by the various parts of the overall process reflect this hierarchical organization and provide special abstract views for each of the processes. • Evolutionary Development over Time. The models upon which problem-solving processes act are generally developed over time, usually by a group of agents, human or machine. For most of its existence the model is incomplete and undergoing evolutionary (incremental) change. It is useful to be able to perform tasks prior to the complete development of a model for two reasons: first, 'Assistant Professor, Department of Architecture "Sun Company University Professor, Department of Civil Engineering Graduate Research Assistant, Department of Civil Engineering `'Graduate Research Assistant, Department of Architecture `Researcher, Department of Architecture - 115 - i,