Heuristic reasoning strategy for automated sensor placement

A heuristic reasoning strategy for automating the placement of sensors in the design of multi-station convergent closerange photogrammetric networks is presented. Optical inspection of three-dimensional (30) objects requires acquisition of a configuration of multiple, spatially separated images about the object. The placement of these sensor stations is subject to a complex and interrelated set of constraints and measurement considerations and, therefore, heuristic solutions are necessary. This strategy exploits the idea of generic multi-station sensor configurations which comprise an ideal number and geometry of sensor stations for a class of measurement problems. A representation, termed the "constraint sphere," is introduced which accommodates a reasoning scheme in which these ideal station poses can be efficiently modified to suit the current object geometry, workspace restrictions, and measurement criteria. Examples demonstrate the strategy's potential. This approach is the first to deal with the mensuration considerations in multi-station sensor placement. vision applications is often conducted in a trial-and-error