Determination of rigid-body pose from imprecise point position measurements

The determination of a rigid-body position and orientation from the position of a number of its points is one of the fundamental problems in kinematics. This problem arises in robotics, biomechanics, automatic guided vehicles, real-time control of space structures, etc. Under ideal conditions, it is possible to apply classical methods to find a body´s position and orientation. However, in the presence of noise, these methods yield results that are unreliable and inconsistent. Two existing and a new method for determining position and orientation from noisy point coordinate data are presented. The theoretical analysis of the methods combined with an extensive simulation process led to conclusions about their behavior in different situations. The proposed method yields better orientation estimates than the other two methods, yielding reliable results both for absolute and relative position measurements and for low and high noise levels.