Distributed and Scalable Cooperative Formation of Unmanned Ground Vehicles Using Deep Reinforcement Learning

Cooperative formation control of unmanned ground vehicles (UGVs) has become one the important research hotspots in application UGV and attracted more attention military civil fields. Compared with traditional algorithms, reinforcement-learning-based algorithms can provide a new solution lower complexity for real-time by equipping UGVs artificial intelligence. Therefore, this paper, distributed deep-reinforcement-learning-based cooperative algorithm is proposed to solve navigation, maintenance, obstacle avoidance tasks formations. More importantly, hierarchical triangular structure newly designed Markov decision process formations leader follower attributes make strategy learned reusable, so that arbitrarily increase number realize flexible expansion. The effectiveness scalability verified simulation experiments different scales.