A Reinforcement Learning Approach to View Planning for Automated Inspection Tasks
Manual inspection of workpieces in highly flexible production facilities with small lot sizes is costly and less reliable compared to automated inspection systems. Reinforcement Learning (RL) offers promising, intelligent solutions for robotic inspection and manufacturing tasks. This paper presents...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-03-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/21/6/2030 |
| _version_ | 1827697798165823488 |
|---|---|
| author | Christian Landgraf Bernd Meese Michael Pabst Georg Martius Marco F. Huber |
| author_facet | Christian Landgraf Bernd Meese Michael Pabst Georg Martius Marco F. Huber |
| author_sort | Christian Landgraf |
| collection | DOAJ |
| description | Manual inspection of workpieces in highly flexible production facilities with small lot sizes is costly and less reliable compared to automated inspection systems. Reinforcement Learning (RL) offers promising, intelligent solutions for robotic inspection and manufacturing tasks. This paper presents an RL-based approach to determine a high-quality set of sensor view poses for arbitrary workpieces based on their 3D computer-aided design (CAD). The framework extends available open-source libraries and provides an interface to the Robot Operating System (ROS) for deploying any supported robot and sensor. The integration into commonly used OpenAI Gym and Baselines leads to an expandable and comparable benchmark for RL algorithms. We give a comprehensive overview of related work in the field of view planning and RL. A comparison of different RL algorithms provides a proof of concept for the framework’s functionality in experimental scenarios. The obtained results exhibit a coverage ratio of up to <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>0.8</mn></mrow></semantics></math></inline-formula> illustrating its potential impact and expandability. The project will be made publicly available along with this article. |
| first_indexed | 2024-03-10T13:16:51Z |
| format | Article |
| id | doaj.art-d6f38d597a10483698a6737243ddb757 |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-03-10T13:16:51Z |
| publishDate | 2021-03-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-d6f38d597a10483698a6737243ddb7572023-11-21T10:20:07ZengMDPI AGSensors1424-82202021-03-01216203010.3390/s21062030A Reinforcement Learning Approach to View Planning for Automated Inspection TasksChristian Landgraf0Bernd Meese1Michael Pabst2Georg Martius3Marco F. Huber4Fraunhofer Institute for Manufacturing, Engineering and Automation IPA, Nobelstraße 12, 70569 Stuttgart, GermanyFraunhofer Institute for Manufacturing, Engineering and Automation IPA, Nobelstraße 12, 70569 Stuttgart, GermanyFraunhofer Institute for Manufacturing, Engineering and Automation IPA, Nobelstraße 12, 70569 Stuttgart, GermanyMax Planck Institute for Intelligent Systems, Max-Planck-Ring 4, 72076 Tübingen, GermanyFraunhofer Institute for Manufacturing, Engineering and Automation IPA, Nobelstraße 12, 70569 Stuttgart, GermanyManual inspection of workpieces in highly flexible production facilities with small lot sizes is costly and less reliable compared to automated inspection systems. Reinforcement Learning (RL) offers promising, intelligent solutions for robotic inspection and manufacturing tasks. This paper presents an RL-based approach to determine a high-quality set of sensor view poses for arbitrary workpieces based on their 3D computer-aided design (CAD). The framework extends available open-source libraries and provides an interface to the Robot Operating System (ROS) for deploying any supported robot and sensor. The integration into commonly used OpenAI Gym and Baselines leads to an expandable and comparable benchmark for RL algorithms. We give a comprehensive overview of related work in the field of view planning and RL. A comparison of different RL algorithms provides a proof of concept for the framework’s functionality in experimental scenarios. The obtained results exhibit a coverage ratio of up to <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>0.8</mn></mrow></semantics></math></inline-formula> illustrating its potential impact and expandability. The project will be made publicly available along with this article.https://www.mdpi.com/1424-8220/21/6/2030view planningreinforcement learningsimulationroboticssmart sensorsautomated inspection |
| spellingShingle | Christian Landgraf Bernd Meese Michael Pabst Georg Martius Marco F. Huber A Reinforcement Learning Approach to View Planning for Automated Inspection Tasks Sensors view planning reinforcement learning simulation robotics smart sensors automated inspection |
| title | A Reinforcement Learning Approach to View Planning for Automated Inspection Tasks |
| title_full | A Reinforcement Learning Approach to View Planning for Automated Inspection Tasks |
| title_fullStr | A Reinforcement Learning Approach to View Planning for Automated Inspection Tasks |
| title_full_unstemmed | A Reinforcement Learning Approach to View Planning for Automated Inspection Tasks |
| title_short | A Reinforcement Learning Approach to View Planning for Automated Inspection Tasks |
| title_sort | reinforcement learning approach to view planning for automated inspection tasks |
| topic | view planning reinforcement learning simulation robotics smart sensors automated inspection |
| url | https://www.mdpi.com/1424-8220/21/6/2030 |
| work_keys_str_mv | AT christianlandgraf areinforcementlearningapproachtoviewplanningforautomatedinspectiontasks AT berndmeese areinforcementlearningapproachtoviewplanningforautomatedinspectiontasks AT michaelpabst areinforcementlearningapproachtoviewplanningforautomatedinspectiontasks AT georgmartius areinforcementlearningapproachtoviewplanningforautomatedinspectiontasks AT marcofhuber areinforcementlearningapproachtoviewplanningforautomatedinspectiontasks AT christianlandgraf reinforcementlearningapproachtoviewplanningforautomatedinspectiontasks AT berndmeese reinforcementlearningapproachtoviewplanningforautomatedinspectiontasks AT michaelpabst reinforcementlearningapproachtoviewplanningforautomatedinspectiontasks AT georgmartius reinforcementlearningapproachtoviewplanningforautomatedinspectiontasks AT marcofhuber reinforcementlearningapproachtoviewplanningforautomatedinspectiontasks |