A two-stage imitation learning framework for the multi-target search problem in swarm robotics

Jie Li, Ying Tan

Abstract: Highlights•For the multi-target search problem in swarm robotics, most existing work is about strategic design while this article focuses on strategy imitation.•To our knowledge, it is the first time that the combination of deep learning technologies and evolutionary algorithms is used for the problem.•The strategy obtained from the framework is close to the target strategy on multiple indicators.•The two-stage imitation learning framework can also be used for other swarm tasks.•The network design and evolutionary algorithm settings can be a good reference.