DETECTION AND POSE ESTIMATED GRASPING OF AN INDUSTRIAL ROBOT IN BIN PICKING OPERATIONS

Avinash Sen

Abstract

The technique used by a robot to grab objects that are randomly placed inside a box or a pallet is called bin picking. Bin picking has evolved greatly over the years due to tremendous strides empowered by advanced computer vision technology, software development and gripping solutions. However,  the  creation  of  a  versatile  system,  capable  of  collecting  any  type  of  object  without deforming it, regardless of the disordered environment around it, remains a challenge. In this thesis a  solution  for  this  problem  that  is  based  on  learning  the  appearance  model  using  convolutional neural networks (CNN) is proposed. By synthetically combining object models and backgrounds of complex composition and high graphical quality, we are able to generate photo realistic images with accurate annotated 3D pose for all objects in our custom created dataset. Using this network, we can estimate the object poses with sufficient accuracy for real world semantic grasping in a cluttered bin by real robot.