New Visual Servoing Control Strategies in Tracking Tasks Using a PKM

Citations Over TimeTop 14% of 2010 papers

Abstract

In this work the full architecture of the Robotenis system and a novel structure of visual control were shown in detail. In this article no results are shown but the more important elements to control and simulate the robot and visual controller were described. Two kinematic models were described in order to obtain two different jacobians were each jacobian is used in different tasks: the System simulator and the real time controller. By means of the condition index of the robot jacobians some singularities of the robot are obtained. In real time tasks the above solution and the condition index of the second jacobian are utilized to bound the work space and avoid singularities, in this work if some point forms part or is near of a singularity then the robot stop the end effector movement and waits to the next target point. Inverse dynamics of the robot is obtained by means of the Lagrange multipliers. The inverse dynamics is used in a non linear feed forward in order to improve the PD joint controller. Although improvement of the behaviour of the robot in notorious, in future works is important to measure how the behaviour is modified when the dynamics fed forward is added and when is not. The trajectory planner is added with two principal objectives: the trajectory planner assures that the robot capabilities are not exceeded and assures that the robot moves softly. The trajectory planner takes into account the movements of the end effector, this consideration has drawbacks: the principal is that the maximum end-effector capabilities are not necessarily the maximum joint capabilities, depends on the end effector position. Above drawbacks suggest redesigning the trajectory planner in order to apply to the joint space, this as another future work. Above elements are used in the visual controller and the robot controller has to satisfied the visual controller requirements. Thanks to the joint controller the robot is supposed stable and its response is considered faster than the visual system. Two cases are presented in this paper: the static case that is exposed in other works and some results and videos are shown, the another controller is called the dynamic case. An objective of the system is to play ping pong by itself and the controller of the dynamic case was specially designed in order to

Related Papers

• → Dexterous Workspace Optimization of a Tricept Parallel Manipulator(2011)49 cited
• → Optimization of a crossbar parallel machine tool based on workspace and dexterity(2015)12 cited
• Application of Pseudo-inverse Estimation of Image Jacobian Matrix to Visual Servoing(2006)
• → Pseudo-inverse Estimation of Image Jacobian Matrix in Uncalibrated Visual Servoing(2006)7 cited
• Comparing methods for estimating the Jacobian in visual servoing(2008)