Pose Estimation of 5-DOF Manipulator using On-Body Markers

Abstract

Manipulators are helpful in performing various hazardous tasks like sanitization with chemicals in germs infected areas, spraying pesticides in fields, pick and place of heavy and hazardous materials where direct human intervention is difficult. For manipulators to perform its assigned task accurately, prior estimation of its pose needs to be pinpointed. End-effector grasping and arm manipulation require estimation of 3D object poses. Recently, a number of procedures and databases for vision-based estimation of object pose have been advised. However, it is not clear about the performance of the developed algorithms for visual pose estimation of robot manipulation. In this paper we present the pose estimation of a 5-dof PhantomX Reactor Arm using On-Body/Aruco Markers. Forward and inverse kinematics were used to estimate the pose from the position coordinates calculated using computer vision techniques. This paper implements an approach aimed at estimating the pose of a camera, affixed to a robotic manipulator, against a target object. We adopt a single camera single-shot technique that minimizes the reprojection error over all the rigid poses. The simulation and experimental results using industrial monocular camera for different sizes of On -body markers were presented.