Passivity-based visual motion observer: From theory to distributed algorithms

Abstract

This paper gives an overview of a series of the authors' researches on the passivity-based visual motion observer and its applications to a variety of control and estimation problems. After providing notations and preliminaries, we introduce the visual motion observer which is a mechanism to estimate the target object's motion in 3D space from vision data. We next build a visual motion observer-based control system including the observer into the feedback loop and prove stability by using passivity. Then, we give three application scenarios of the observer. First scenario is dynamic control of a robot manipulator whose objective is to control a manipulator equipped with vision on its end-effector so that the camera tracks a moving target object. Second one is cooperative control where a network of rigid bodies with vision is controlled so as to achieve cooperation. Third one is cooperative estimation for visual sensor networks whose objective is to achieve tracking of estimates to a moving target and averaging them over the network simultaneously. In all of the scenarios, it is shown that the visual motion observer plays an important role in achieving the objectives.