Abstract
We consider a robot that must sort objects transported by a conveyor belt into different classes. Multiple observations must be performed before taking a decision on the class of each object, because the imperfect sensing sometimes detects the incorrect object class. The objective is to sort the sequence of objects in a minimal number of observation and decision steps. We describe this task in the framework of partially observable Markov decision processes, and we propose a reward function that explicitly takes into account the information gain of the viewpoint selection actions applied. The DESPOT algorithm is applied to solve the problem, automatically obtaining a sequence of observation viewpoints and class decision actions. Observations are made either only for the object on the first position of the conveyor belt or for multiple adjacent positions at once. The performance of the single- and multiple-position variants is compared, and the impact of including the information gain is analyzed. Real-life experiments with a Baxter robot and an industrial conveyor belt are provided.
Highlights
Robots in open environments, such as those that arise in the Industry 4.0 paradigm, collaborative, or domestic robotics [1,2,3,4], are affected by significant uncertainty in perceiving their environment.A way to handle this is to use active perception [5,6,7,8,9], which closes the loop between the sensing and control modules of the robot: control actions are chosen to maximise the information acquired from the sensor and thereby reduce uncertainty
We provide some insight into the structure of the tree explored in the active perception problem, and we empirically determine the branching factor of DESPOT
In all the experiments of this section, we study the impact of the incorrect classification penalty, rmin, as it varies in absolute value from 5 to 1000. This is always done for the task of sorting a sequence of 10 light bulbs, and for observations of either the first object on the conveyor belt, or of the first two objects
Summary
A way to handle this is to use active perception [5,6,7,8,9], which closes the loop between the sensing and control modules of the robot: control actions are chosen to maximise the information acquired from the sensor and thereby reduce uncertainty. Two leading paradigms in active perception are passive and active detection. We consider here the following active perception problem, which uses passive detection and is relevant for Industry 4.0. A robot working in a factory has the task of sorting differently shaped objects that are transported on a conveyor belt. The conveyor belt is scanned from a set of poses (viewpoints), and the robot is able to move between the viewpoints
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
