Camera Placement Planning Avoiding Occlusion: Test Results Using a Robotic Hand/Eye System

Tarabanis, Kostantinos; Tsai, Roger Y.

Camera placement experiments are presented that demonstrate the effectiveness of a viewpoint planning algorithm that avoids occlusion of a visual target. A CCD camera mounted on a robot in a hand-eye configuration is placed at planned unobstructed viewpoints to observe a target on a real object. The validity of the method is tested by placing the camera inside the viewing region, that is constructed using the proposed new sensor placement planning algorithm and observing whether the target is truly visible. The accuracy of the boundary of the constructed viewing region is tested by placing the camera at the critical - locations of the viewing region boundary and confirming that the target is barely visible. The corresponding scenes from the candidate viewpoints are shown demonstrating that occlusions are properly avoided.



More About This Work

Academic Units
Computer Science
Department of Computer Science, Columbia University
Columbia University Computer Science Technical Reports, CUCS-501-89
Published Here
January 20, 2012