Techniques for Selecting Pose Algorithms

Abstract

A technique for selecting one pose algorithm from m algorithms containing zero mean Gaussian errors is derived. The procedure consists of a two stage analysis. First, the joint entropy of each algorithm is found. The algorithm with minimum entropy is shown to possess the greatest possible lower bound reliability of meeting any quadratic specification of the pose error. In addition, the entropy of pose measurements is shown to be invariant with respect to homogeneous coordinate transformations, so the analysis is simplified for physically distributed robotic systems. A method of calculating the actual greatest lower bound reliability for a given quadratic specification is derived. The concepts are simulated using a visual pose measurement system developed by NASA