Acquiring visual-motor models for precision manipulation with robot hands

Abstract

Abstract. Dextrous high degree of freedom (DOF) robotic hands provide versatile motions for fine manipulation of potentially very different objects. However, fine manipulation of an object grasped by a multifinger hand is much more complex than if the object is rigidly attached to a robot arm. Creating an accurate model is difficult if not impossible. We instead propose a combination of two techniques: the use of an approximate estimated motor model, based on the grasp tetrahedron acquired when grasping an object, and the use of visual feedback to achieve accurate fine manipulation. We present a novel active vision based algorithm for visual servoing, capable of learning the manipulator kinematics and camera calibration online while executing a manipulation task. The approach differs from previous work in that a full, coupled image Jacobian is estimated online without prior models, and that a trust region control method is used, improving stability and convergence. We present an extensive experimental evaluation of visual model acquisition and visual servoing in 3, 4 and 6 DOF.

Untitled

“…It can be seen that the quality of the manipulation is quite good, showing also that the learned skills can be transferred between similar objects. Although few quantitative results for other systems have been reported, the quality of the manipulations seems comparable to the one obtained by other systems where the manipulations are programmed by hand, such as (Speeter, 1991;Michelman & Allen, 1993;Jägersand et al, 1996;Fuentes & Nelson, 1996a). Figure 6 shows the value of the manipulation quality function f for a translation of 25 mm along the x-axis as a function of the generation.…”

Visual Servoing Using Projective Kinematics

“…Formally, the projective robot motion writes simply as H(q) = H −1 PE T(q)H PE (5), (7). Moreover, the fact that all revolutions are represented in camera space permits also to conjugate the inner factors of the poe (7).…”

“…Therefore, it allows for global visual seroving since it is valid over the entire configuration space and allows for arbitrary shapes of the tool to be recovered on-the-fly. In contrast to existing formulations of visual servoing, the Jacobian is neither an on-line estimated linear model [7], nor an apriori given approximation around the target [8], [9]. Above that, metric-representations, which would require a-priori calibration, were successfully eliminated in favor of projective camera space and projective kinematics.…”

Experimental evaluation of uncalibrated visual servoing for precision manipulation

“…Creating an accurate manipulation model is difficult if not impossible due to the many possible objects that could be grasped, and the inherent manipulator inaccuracies. In [16] we augment the visual model estimation and control with fine manipulation capabilities for the Utah/MIT hand. This is a review of the experiments in that paper.…”