Relative Position-Based Visual Servoing Control for Quadrotors

Abstract

This paper presents a position-based visual servoing control system that allows a quadrotor aerial robot with a single onboard camera to track a four degree of freedom relative pose trajectory with respect to a free-moving unknown target object composed of visual point features. The proposed vision system is able to estimate both the six degree of freedom relative motion and the target object model structure using only the images captured from the camera and recover scale using a minimal set of inter-feature point distances. An inner-loop attitude controller is used with an outer-loop position trajectory tracking control law to follow a relative trajectory with respect to the target object, while both bodies move in the inertial frame. The performance of the proposed algorithms are evaluated in simulation.