Juan Cervino, Saurav Agarwal, Vijay Kumar, Alejandro Ribeiro
{"title":"分散式多目标覆盖控制的受限学习","authors":"Juan Cervino, Saurav Agarwal, Vijay Kumar, Alejandro Ribeiro","doi":"arxiv-2409.11311","DOIUrl":null,"url":null,"abstract":"The multi-objective coverage control problem requires a robot swarm to\ncollaboratively provide sensor coverage to multiple heterogeneous importance\ndensity fields (IDFs) simultaneously. We pose this as an optimization problem\nwith constraints and study two different formulations: (1) Fair coverage, where\nwe minimize the maximum coverage cost for any field, promoting equitable\nresource distribution among all fields; and (2) Constrained coverage, where\neach field must be covered below a certain cost threshold, ensuring that\ncritical areas receive adequate coverage according to predefined importance\nlevels. We study the decentralized setting where robots have limited\ncommunication and local sensing capabilities, making the system more realistic,\nscalable, and robust. Given the complexity, we propose a novel decentralized\nconstrained learning approach that combines primal-dual optimization with a\nLearnable Perception-Action-Communication (LPAC) neural network architecture.\nWe show that the Lagrangian of the dual problem can be reformulated as a linear\ncombination of the IDFs, enabling the LPAC policy to serve as a primal solver.\nWe empirically demonstrate that the proposed method (i) significantly\noutperforms existing state-of-the-art decentralized controllers by 30% on\naverage in terms of coverage cost, (ii) transfers well to larger environments\nwith more robots and (iii) is scalable in the number of fields and robots in\nthe swarm.","PeriodicalId":501175,"journal":{"name":"arXiv - EE - Systems and Control","volume":"32 1","pages":""},"PeriodicalIF":0.0000,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Constrained Learning for Decentralized Multi-Objective Coverage Control\",\"authors\":\"Juan Cervino, Saurav Agarwal, Vijay Kumar, Alejandro Ribeiro\",\"doi\":\"arxiv-2409.11311\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"The multi-objective coverage control problem requires a robot swarm to\\ncollaboratively provide sensor coverage to multiple heterogeneous importance\\ndensity fields (IDFs) simultaneously. We pose this as an optimization problem\\nwith constraints and study two different formulations: (1) Fair coverage, where\\nwe minimize the maximum coverage cost for any field, promoting equitable\\nresource distribution among all fields; and (2) Constrained coverage, where\\neach field must be covered below a certain cost threshold, ensuring that\\ncritical areas receive adequate coverage according to predefined importance\\nlevels. We study the decentralized setting where robots have limited\\ncommunication and local sensing capabilities, making the system more realistic,\\nscalable, and robust. Given the complexity, we propose a novel decentralized\\nconstrained learning approach that combines primal-dual optimization with a\\nLearnable Perception-Action-Communication (LPAC) neural network architecture.\\nWe show that the Lagrangian of the dual problem can be reformulated as a linear\\ncombination of the IDFs, enabling the LPAC policy to serve as a primal solver.\\nWe empirically demonstrate that the proposed method (i) significantly\\noutperforms existing state-of-the-art decentralized controllers by 30% on\\naverage in terms of coverage cost, (ii) transfers well to larger environments\\nwith more robots and (iii) is scalable in the number of fields and robots in\\nthe swarm.\",\"PeriodicalId\":501175,\"journal\":{\"name\":\"arXiv - EE - Systems and Control\",\"volume\":\"32 1\",\"pages\":\"\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-09-17\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"arXiv - EE - Systems and Control\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/arxiv-2409.11311\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"arXiv - EE - Systems and Control","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2409.11311","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Constrained Learning for Decentralized Multi-Objective Coverage Control
The multi-objective coverage control problem requires a robot swarm to
collaboratively provide sensor coverage to multiple heterogeneous importance
density fields (IDFs) simultaneously. We pose this as an optimization problem
with constraints and study two different formulations: (1) Fair coverage, where
we minimize the maximum coverage cost for any field, promoting equitable
resource distribution among all fields; and (2) Constrained coverage, where
each field must be covered below a certain cost threshold, ensuring that
critical areas receive adequate coverage according to predefined importance
levels. We study the decentralized setting where robots have limited
communication and local sensing capabilities, making the system more realistic,
scalable, and robust. Given the complexity, we propose a novel decentralized
constrained learning approach that combines primal-dual optimization with a
Learnable Perception-Action-Communication (LPAC) neural network architecture.
We show that the Lagrangian of the dual problem can be reformulated as a linear
combination of the IDFs, enabling the LPAC policy to serve as a primal solver.
We empirically demonstrate that the proposed method (i) significantly
outperforms existing state-of-the-art decentralized controllers by 30% on
average in terms of coverage cost, (ii) transfers well to larger environments
with more robots and (iii) is scalable in the number of fields and robots in
the swarm.