AdverSAR: Adversarial Search and Rescue via Multi-Agent Reinforcement Learning

被引:4
作者
Rahman, Aowabin [1 ]
Bhattacharya, Arnab [1 ]
Ramachandran, Thiagarajan [1 ]
Mukherjee, Sayak [1 ]
Sharma, Himanshu [1 ]
Fujimoto, Ted [2 ]
Chatterjee, Samrat [3 ]
机构
[1] Pacific Northwest Natl Lab, Optimizat & Control Grp, Richland, WA USA
[2] Pacific Northwest Natl Lab, Data Analyt Grp, Richland, WA USA
[3] Pacific Northwest Natl Lab, Data Sci & Machine Intelligence Grp, Richland, WA USA
来源
2022 IEEE INTERNATIONAL SYMPOSIUM ON TECHNOLOGIES FOR HOMELAND SECURITY (HST) | 2022年
关键词
Search and Rescue; Multi-agent Reinforcement Learning; Adversarial Reinforcement Learning; Critical Infrastructure Security;
D O I
10.1109/HST56032.2022.10025434
中图分类号
TP39 [计算机的应用];
学科分类号
081203 ; 0835 ;
摘要
Search and Rescue (SAR) missions in remote environments often employ autonomous multi-robot systems that learn, plan, and execute a combination of local single-robot control actions, group primitives, and global mission-oriented coordination and collaboration. Often, SAR coordination strategies are manually designed by human experts who can remotely control the multi-robot system and enable semi-autonomous operations. However, in remote environments where connectivity is limited and human intervention is often not possible, decentralized collaboration strategies are needed for fully-autonomous operations. Nevertheless, decentralized coordination may be ineffective in adversarial environments due to sensor noise, actuation faults, or manipulation of inter-agent communication data. In this paper, we propose an algorithmic approach based on adversarial multi-agent reinforcement learning (MARL) that allows robots to efficiently coordinate their strategies in the presence of adversarial inter-agent communications. In our setup, the objective of the multi-robot team is to discover targets strategically in an obstacle-strewn geographical area by minimizing the average time needed to find the targets. It is assumed that the robots have no prior knowledge of the target locations, and they can interact with only a subset of neighboring robots at any time. Based on the centralized training with decentralized execution (CTDE) paradigm in MARL, we utilize a hierarchical meta-learning framework to learn dynamic team-coordination modalities and discover emergent team behavior under complex cooperative-competitive scenarios. The effectiveness of our approach is demonstrated on a collection of prototype grid-world environments with different specifications of benign and adversarial agents, target locations, and agent rewards.
引用
收藏
页数:7
相关论文
共 12 条
  • [1] Learning Robot Swarm Tactics over Complex Adversarial Environments
    Behjat, Amir
    Manjunatha, Hemanth
    Kumar, Prajit Krisshna
    Jani, Apurv
    Collins, Leighton
    Ghassemi, Payam
    Distefano, Joseph
    Doermann, David
    Dantu, Karthik
    Esfahani, Ehsan
    Chowdhury, Souma
    [J]. 2021 INTERNATIONAL SYMPOSIUM ON MULTI-ROBOT AND MULTI-AGENT SYSTEMS (MRS), 2021, : 83 - 91
  • [2] Dongzi Wang, 2020, 2020 IEEE 3rd International Conference on Information Systems and Computer Aided Education (ICISCAE), P1, DOI 10.1109/ICISCAE51034.2020.9236869
  • [3] Iqbal S, 2021, Arxiv, DOI arXiv:1905.12127
  • [4] Lowe R, 2017, ADV NEUR IN, V30
  • [5] Lyu X, 2021, Arxiv, DOI arXiv:2102.04402
  • [6] Omidshafiei S, 2015, IEEE INT CONF ROBOT, P5962, DOI 10.1109/ICRA.2015.7140035
  • [7] Papoudakis G, 2019, Arxiv, DOI arXiv:1906.04737
  • [8] Pinto L, 2017, PR MACH LEARN RES, V70
  • [9] Collaborative Multi-Robot Search and Rescue: Planning, Coordination, Perception, and Active Vision
    Queralta, Jorge Pena
    Taipalmaa, Jussi
    Pullinen, Bilge Can
    Sarker, Victor Kathan
    Tuan Nguyen Gia
    Tenhunen, Hannu
    Gabbouj, Moncef
    Raitoharju, Jenni
    Westerlund, Tomi
    [J]. IEEE ACCESS, 2020, 8 : 191617 - 191643
  • [10] Silver D., 2015, Lecture 2: Markov decision processes