Online Planning-based Gene Regulatory Network for Swarm in Constrained Environment

Abstract

Swarm intelligence inspired by the all kinds of theory from nature has developed rapidly towards dealing with complicated problems. In realm of collective robots, a main stream is to urge robots more intelligent in performance of tasks, such as distributed system, complete self-organization and reliance only on local information. Gene regulatory networks (GRNs) construct the cell theory about the regulatory activities between genes and protein and succeeded in being exerted to multirobot system and achieving the collective entrapping and tracking function. The excellent self-organized and robust characteristic in GRNs guarantees collective tasks more fault-tolerant. Unfortunately, swarm robots are easily trapped into perplexed stuck and stagnate still when robots lose contact with targets because of obstacle blocking. To overcome the dilemma, we proposed online planning-based gene regulatory network (OP-GRN) to bring robots to normal orbit and supply guidance to reconstruct the contact with targets, which involves online grouping planning (OGP) and online path planning (OPP). The experiment results demonstrate the efficacy and superiority of our model in constrained environment.