.png){kind=icon}
.png){kind=icon}
Return
Game-Theoretic Fault-Tolerant Strategies for Multi-Agent Pursuit-Evasion With Actuator Faults
.png){kind=icon}
PRE
AI
H
L
W
DOI：10.1109/LCSYS.2026.3683403.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This letter studies game-theoretic fault-tolerant strategy design for multi-agent pursuit-evasion (MPE) systems subject to actuator faults. By explicitly embedding actuator fault effects into the strategy layer, the MPE interaction is formulated as a zero-sum dynamic game in which the pursuers aim to minimize capture errors in the presence of both evader actions and fault-induced uncertainties. To address the associated Hamilton-Jacobi-Isaacs (HJI) equation, a critic-only online adaptive dynamic programming algorithm is developed, where a single critic neural network approximates the optimal value function and corresponding feedback strategy via experience replay. The proposed approach yields an implementable fault-tolerant strategy with reduced computational complexity. Simulation results verify the effectiveness and robustness of the proposed method under actuator fault scenarios.
Keywords:
Circuits
Feedback
Circuits and systems
Product development
Communication systems
Computer networks
Electronic mail
Event detection
Graphical user interfaces
Wide area networks
Pursuit-evasion games
fault-tolerant strategy design
multi-agent systems
actuator faults
Journal
I
IF:
2
Papers:
94
Citations:
5.0K
Organization
Cited Papers