Swarm emergence control method based on group direction consistency and stability under a behavioral cloning framework

Abstract

The invention discloses a swarm emergence control method based on group direction consistency and stability under a behavioral cloning framework, and the method is used for controlling the swarm motion of an intelligent agent cluster, and comprises the steps: determining all neighborhood intelligent agents in the sensing range of the intelligent agents; constructing a virtual neighborhood center corresponding to the intelligent agent according to the state information of the intelligent agent and the state information of all neighborhood intelligent agents of the intelligent agent, and determining the state information of the virtual neighborhood center; using a preset deep neural network to extract recessive features of the state information of the intelligent agent at the current moment and the state information of the virtual neighborhood center, and obtaining the control output quantity of the intelligent agent at the next moment, each weight parameter of the preset deep neural network is trained and determined by using a cost function constructed based on group direction consistency and group system stability. According to the method, the intelligent agent cluster can be controlled to generate intelligent agent cluster swarming motion meeting the requirements of group direction consistency and group system stability based on local sensing information.

Description

technical field [0001] The invention relates to the technical field of intelligent body group motion control, in particular to a swarming control method based on group direction consistency and stability under a behavior cloning framework. Background technique [0002] The intelligent agent cluster is a kind of swarm robot designed inspired by the living habits of social animals in nature. The intelligent agent cluster does not have a centralized control structure. Through the local interaction between agents and the interaction between agents and the external environment, decision-making and execution of their own capabilities Actions within a certain range emerge from specific macro-group behaviors to perform specific tasks. Therefore, the research on the swarm control method of intelligent agents is of vital value and significance to further improve the energy efficiency of unmanned swarm systems and play the role of unmanned systems. [0003] The early researches on the...

AI Technical Summary

Problems solved by technology

However, the design of the clustering algorithm depends on the parameter setting, which is unstable, and the output of the behavior action is limited to discrete actions, which cannot be clearly mapped to the behavior of the agent in the continuous space.

Images