Visual servoing of a robotic manipulator based on fuzzy logic control

Abstract

In this paper a solution to a manipulation control problem (target identification and grasping) using a red platform in combination with a vision system is proposed. The task for the end-effector is to approach a randomly placed spherical object of known size. The control law is approximated by using an approach based on fuzzy logic. The controller determines the parameters of the target (a spherical object acquired from the vision system mounted on the manipulator) using a vision algorithm. The fuzzy rules are built in a supervised way, after studying the behavior of the system. Experimental results obtained using an industrial manipulator (PUMA 560 with 6 DOF) are presented.

Visual motor control of a 7 DOF robot manipulator using a fuzzy SOM network

“…(21). During the coarse movement, joint angles are moved by θ out 0 and its new end-effector position is recorded by the cameras as v 0 .…”

“…Moreover, while neural networks have been used extensively for visual motor coordination problem, application of fuzzy logic to this problem has been limited at least in the context of manipulators. In the context of visual servoing of manipulators, fuzzy logic has been used for identifying targets and grasping tasks [21][22][23]. Kumaresan et al [24] suggested a fuzzy logic-based scheme where the need for re-calibrating cameras was eliminated in case of any inadvertent camera displacements.…”

A hierarchical behavior-based approach to manipulation tasks

“…All tasks re-use the same set of basic behaviors, resulting in a small overall set of rules. For instance, the full PickUp behavior consists of 19 fuzzy rules, compared with, e.g., the over 120 used in [5].…”

“…However, the block starts to move away from its position, and the arm tracks it and comes closer to it, until it can finally grasp it. Notice that this situation would be problematic if the behaviors were organized, say, in a finite state machine [10,5]. The last example, shown in Fig.…”

“…For instance, Pettinaro [10] mostly uses sequential activations of generic arm behaviors to perform assembly tasks. De Giuseppe [5] uses sequential activation of fuzzy behaviors to perform a pick-up task. All the above approaches are based on a winner-takeall strategy for behavior combination: only one behavior is in the control of the manipulator at any given time.…”