A heterogeneous multi-agent cooperative pursuit method

By combining a trained collaborative target allocation network and a pursuit decision network with Transformer and LSTM models, the problem of heterogeneous agents collaboratively pursuing multiple intruding targets in complex environments is solved, improving the pursuit success rate and decision accuracy. This approach is suitable for UAV countermeasures and island defense.

CN122173915APending Publication Date: 2026-06-09BEIHANG UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
BEIHANG UNIV
Filing Date
2024-12-09
Publication Date
2026-06-09

AI Technical Summary

Technical Problem

Existing technologies struggle to effectively address the challenge of heterogeneous agents collaboratively pursuing multiple intruding targets in complex environments. This is especially true when obstacles and defensive targets are present, making it difficult to accurately determine the attack targets and the strategies of the intruding agents, thus hindering the pursuit.

Method used

A heterogeneous multi-agent cooperative pursuit method is adopted. By combining a trained cooperative target allocation network and a pursuit decision network with the evaluation module of the Transformer architecture and the LSTM model, the behavior of the target agent and the environmental state are predicted and the pursuit strategy is dynamically adjusted.

Benefits of technology

It improves the success rate and decision-making accuracy of heterogeneous intelligent agents, and is applicable to scenarios such as drone countermeasures and island and reef defense, achieving efficient pursuit of multiple intrusion targets.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122173915A_ABST
    Figure CN122173915A_ABST
Patent Text Reader

Abstract

This invention relates to a heterogeneous multi-agent cooperative pursuit method, belonging to the field of agent pursuit technology, and solves the problem of unknown intrusion intent of target agents and difficulty in pursuit in existing heterogeneous agent pursuit methods. Specific steps include: acquiring the state of the target agent set and the pursuit environment; combining the states of the target agents and the pursuit environment, and based on a trained cooperative target allocation network, obtaining the optimal pursuit target allocation result for each heterogeneous agent; after determining the pursuit target, each heterogeneous agent obtains the predicted target agent behavior based on the current state of the target agent and a trained target agent behavior prediction module; combining the predicted target agent behavior with the state of the pursuit environment, and based on a trained pursuit decision network, obtaining the optimal pursuit action; and the heterogeneous multi-agents perform cooperative pursuit based on the optimal pursuit action, achieving efficient pursuit of all intruding targets.
Need to check novelty before this filing date? Find Prior Art