Design of a Deep Post Gripping Perception Framework for Industrial Robots

Abstract

The use of flexible and autonomous robotic systems is a possible solution for automation in dynamic and unstructured industrial environments. Pick and place robotic applications are becoming common for the automation of manipulation tasks in an industrial context. This context requires the robot to be aware of its surroundings throughout the whole manipulation task, even after accomplishing the gripping action. This work introduces the deep post gripping perception framework, which includes post gripping perception abilities realized with the help of deep learning techniques, mainly unsupervised learning methods. These abilities help robots to execute a stable and precise placing of the gripped items while respecting the process quality requirements. The framework development is described based on the results of a literature review on post gripping perception functions and frameworks. This results in a modular design using three building components to realize planning, monitoring and verifying modules. Experimental evaluation of the framework shows its advantages in terms of process quality and stability in pick and place applications.