Defocus-Based Direct Visual Servoing

Abstract

Direct Visual Servoing (DVS) considers pixel brightness directly as input of robot control. Recent DVS variants consider image processing as smoothing or frequency domain transforms, resulting in large convergence domains.This paper proposes to consider defocus to optically smooth images without processing. The resulting Defocus-based DVS shows convergence domains competing with the state-of-theart, larger in some challenging cases, for lower complexity.

A Study on Learned Feature Maps Toward Direct Visual Servoing

“…Recent variants of DVS have explored various image representation techniques resulting in expanded convergence domains, such as in [5] where Photometric Gaussian Mixtures are introduced as visual features for DVS. In [6], a novel approach to DVS is proposed by considering defocus as a way to optically smooth images without additional image processing. It demonstrates competitive convergence domains compared to the state-of-the-art methods, with larger domains in various scenarios, for a lower computational complexity.…”

Direct visual servoing in the non-linear scale space of camera pose

“…Recent progresses have made both Direct VS (DVS) [8], [9], [12], [5] and image alignment [1], [21] achieve large convergence domains. One of the key ideas was to control the Gaussian smoothing of images together with camera pose Degrees of Freedom (DoF) [8], [1].…”

“…To the best of our knowledge, NLSS has only been developed in the image denoising field for preserving natural edges [17], [3] or used as a consequence of optical properties, e.g. defocus [5]. The new NLSS considers anisotropic Gaussian filtering depending on camera pose DoFs.…”