Skill Fusion in Hybrid Robotic Framework for Visual Object Goal Navigation
In recent years, Embodied AI has become one of the main topics in robotics. For the agent to operate in human-centric environments, it needs the ability to explore previously unseen areas and to navigate to objects that humans want the agent to interact with. This task, which can be formulated as Ob...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-07-01
|
| Series: | Robotics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2218-6581/12/4/104 |
| _version_ | 1797583285233123328 |
|---|---|
| author | Aleksei Staroverov Kirill Muravyev Konstantin Yakovlev Aleksandr I. Panov |
| author_facet | Aleksei Staroverov Kirill Muravyev Konstantin Yakovlev Aleksandr I. Panov |
| author_sort | Aleksei Staroverov |
| collection | DOAJ |
| description | In recent years, Embodied AI has become one of the main topics in robotics. For the agent to operate in human-centric environments, it needs the ability to explore previously unseen areas and to navigate to objects that humans want the agent to interact with. This task, which can be formulated as ObjectGoal Navigation (ObjectNav), is the main focus of this work. To solve this challenging problem, we suggest a hybrid framework consisting of both not-learnable and learnable modules and a switcher between them—SkillFusion. The former are more accurate, while the latter are more robust to sensors’ noise. To mitigate the sim-to-real gap, which often arises with learnable methods, we suggest training them in such a way that they are less environment-dependent. As a result, our method showed top results in both the Habitat simulator and during the evaluations on a real robot. |
| first_indexed | 2024-03-10T23:36:44Z |
| format | Article |
| id | doaj.art-840a53f627ed4cc29a18f530a35e49fd |
| institution | Directory Open Access Journal |
| issn | 2218-6581 |
| language | English |
| last_indexed | 2024-03-10T23:36:44Z |
| publishDate | 2023-07-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Robotics |
| spelling | doaj.art-840a53f627ed4cc29a18f530a35e49fd2023-11-19T02:55:18ZengMDPI AGRobotics2218-65812023-07-0112410410.3390/robotics12040104Skill Fusion in Hybrid Robotic Framework for Visual Object Goal NavigationAleksei Staroverov0Kirill Muravyev1Konstantin Yakovlev2Aleksandr I. Panov3AIRI, 105064 Moscow, RussiaFederal Research Center for Computer Science and Control of Russian Academy of Sciences, 119333 Moscow, RussiaFederal Research Center for Computer Science and Control of Russian Academy of Sciences, 119333 Moscow, RussiaAIRI, 105064 Moscow, RussiaIn recent years, Embodied AI has become one of the main topics in robotics. For the agent to operate in human-centric environments, it needs the ability to explore previously unseen areas and to navigate to objects that humans want the agent to interact with. This task, which can be formulated as ObjectGoal Navigation (ObjectNav), is the main focus of this work. To solve this challenging problem, we suggest a hybrid framework consisting of both not-learnable and learnable modules and a switcher between them—SkillFusion. The former are more accurate, while the latter are more robust to sensors’ noise. To mitigate the sim-to-real gap, which often arises with learnable methods, we suggest training them in such a way that they are less environment-dependent. As a result, our method showed top results in both the Habitat simulator and during the evaluations on a real robot.https://www.mdpi.com/2218-6581/12/4/104navigationroboticsreinforcement learningfrontier-based exploration |
| spellingShingle | Aleksei Staroverov Kirill Muravyev Konstantin Yakovlev Aleksandr I. Panov Skill Fusion in Hybrid Robotic Framework for Visual Object Goal Navigation Robotics navigation robotics reinforcement learning frontier-based exploration |
| title | Skill Fusion in Hybrid Robotic Framework for Visual Object Goal Navigation |
| title_full | Skill Fusion in Hybrid Robotic Framework for Visual Object Goal Navigation |
| title_fullStr | Skill Fusion in Hybrid Robotic Framework for Visual Object Goal Navigation |
| title_full_unstemmed | Skill Fusion in Hybrid Robotic Framework for Visual Object Goal Navigation |
| title_short | Skill Fusion in Hybrid Robotic Framework for Visual Object Goal Navigation |
| title_sort | skill fusion in hybrid robotic framework for visual object goal navigation |
| topic | navigation robotics reinforcement learning frontier-based exploration |
| url | https://www.mdpi.com/2218-6581/12/4/104 |
| work_keys_str_mv | AT alekseistaroverov skillfusioninhybridroboticframeworkforvisualobjectgoalnavigation AT kirillmuravyev skillfusioninhybridroboticframeworkforvisualobjectgoalnavigation AT konstantinyakovlev skillfusioninhybridroboticframeworkforvisualobjectgoalnavigation AT aleksandripanov skillfusioninhybridroboticframeworkforvisualobjectgoalnavigation |