Prosiding Improving autonomous robot gripper position on lifting trash object based on Object Geometry Parameters and Centroid Modification

Naf'an, Emil and Sulaiman, Riza and Ali, Nazlena Mohammad (2023) Prosiding Improving autonomous robot gripper position on lifting trash object based on Object Geometry Parameters and Centroid Modification. International Visual Informatics Conference. (Submitted)

Text (Prosiding) Kelengkapan+Prosiding+2.pdf Download (7MB)

Abstract

Abstract. This study aims to improve the positioning accuracy of the gripper robot in lifting garbage objects. In general, the process of removing garbage objects only relies on the object's centroid. If the left and right sides of the centroid have the same parameters, then the lifting process usually goes well. If the object geometry parameters are different, then there is a possibility that the centroid point is in the 2D area, so that the robotic gripper cannot grip and lift the trash object. Likewise, if there is a difference in weight, the robot gripper will have difficulty lifting the object, because the object is one-sided. For this reason, the process of removing garbage objects needs to be improved by considering several parameters, not only the centeroid parameter. In this study, a method for lifting garbage objects is proposed based on several parameters, namely; geometry, centroid and garbage object type. The proposed method is called Object Geometry Parameters and Centroid Modification (OGP-CM). The test results show that the OGP-CM method is able to set the centroid position based on the geometric parameters and the type of trash. On the same object geometry, the improvement in accuracy is relatively low, ranging from 0.46% to 1.72%. A relatively high improvement in accuracy occurs for different object geometries, ranging from 11.54% to 13.09%. Thus the improvement of the position of the autonomous robot gripper in lifting objects using OGP-CM has been successfully carried out. Keywords: Improving, Gripper Position, OGP-CM, Robot.

Actions (login required)

View Item