Collaborative Hunting For Distributed Unmanned Swarm Validation System Design And Implementation

With the rapid development of unmanned systems and artificial intelligence technology,unmanned platforms represented by unmanned ground vehicles have been able to replace humans to complete some complex and dangerous tasks in real life.The use of unmanned swarms for collaborative hunting has been widely used in civil and military fields such as disaster rescue,field investigation,natural resource exploration,and pirate strikes.At present,most of the research on the collaborative Hunting of unmanned swarms focuses on the research of Hunting algorithms,some of which only stay in the theoretical stage and are difficult to implement.To realize the implementation of the hunting intelligent algorithm from theory to practical application,the effectiveness of the hunting algorithm must be verified.However,the current verification of the hunting algorithm has the following difficulties.First,the verification of the hunting algorithm is currently too idealized,and many elements are ignored in the verification process,such as the problem of collaborative communication between unmanned platforms,which leads to the deviation in the performance of the hunting algorithm between the verification process and application in the actual hunting or hunting algorithm cannot be deployed into practical application scenarios at all.Second,some of the current hunting algorithm verifications are tightly coupled with specific algorithms,so the hunting verification system can only verify certain specific algorithms.Third,the functions of the current Hunting verification system are not perfect.Some do not support the simulation of physical effects,some do not support dynamic access to algorithms,and some do not support switching hunting verification scenarios and configuring various capability parameters.It is difficult to find a fully functional hunting verification system.To sum up,thesis designs and implements a collaborative hunting verification system for distributed unmanned swarms to solve these problems,help hunting algorithm designers to quickly complete the verification and iteration of Hunting algorithms,and realize the practical application of hunting algorithms from theory to practice.Thesis first introduces distributed data middleware to provide information-level support for collaborative decision-making between unmanned platforms.Each unmanned platform completes collaborative decision-making through active message sharing and the subscription notification mechanism provided by the information system.Second,a componentized hunting semi-physical simulation platform with physical effects is decoupled from the hunting algorithm to improve the verification efficiency and accuracy of the hunting verification system.Finally,a multi-agent cooperative hunting decision-making algorithm is proposed based on the MDP decision model.In order to verify the verification system and the hunting algorithm designed in thesis,a specific example of hunting targets for a group of unmanned vehicles is designed,and the target escape algorithm and target-following-based hunting algorithm are designed as a comparison algorithm.Each algorithm module runs on an embedded board,thus forming a set of demonstration verification system to verify the hunting verification system proposed in thesis.