Ressearch
Distributed optimization and control:
Distributed cooperative control is the core and key to support cluster intelligent multiplication. Existing collaborative control methods generally assume that the communication network associated with MASs consists of only one connected component. However, the communication network in the real world is often divided into several sub-clusters due to its wide area scenarios or task requirements, resulting in the communication network sometimes containing several isolated sub-networks. Multi-intelligence systems with trunked communication networks are collectively referred to as trunked multi-agent systems (CMAS). How to realize the distributed and efficient cooperative control of cluster system in complex environment is one of the technical problems to be solved.
In the future, we plan to study the game antagonism of unmanned systems from the two directions of hardware platform and software algorithm. Firstly, the existing experimental platform will be upgraded on the platform, and a realistic environment for unmanned antagonism will be built for the study of algorithm deployment and application. Consult papers on software algorithms, study and study existing mainstream game adversarial algorithms, learn the simulation environment construction methods of mainstream teams in this direction, and try to build their own simulation environment on the server for adversarial algorithm verification.
In addition, we will try to participate in some cutting-edge competitions to promote learning through competitions.
Intelligent control based on Learning: the learning algorithm is organically integrated into the control of multi-agent system to improve the adaptability of the agent to the complex environment, so that the agent has the ability of learning optimization and self-adaptation.
