Decentralized navigation method for swarm robots with individual loss due to failure
Ryuma Maeda, Takahiro Endo, Fumitoshi Matsuno · Journal of the Franklin Institute · 2023
This paper presents a decentralized navigation method with a leader robot guiding a robotic swarm and considers the possible occurrence of a failure. Multi-robot systems have attracted attention in part because they can maintain their performance as a swarm even when some robots fail. Such systems are also expected to work well in environments where a disaster has occurred, given that the chance of robot crashes is high in such situations. Therefore, it is important to design a navigation method that considers the possibility of failures so that the surviving robots in a swarm will not be impaired because of failed robots. In this paper, we design connectivity maintenance and a repair method for swarm robots in which each robot utilizes only local information. We then prove that the proposed method ensures the swarm follows the movement of a leader robot and reconstructs connectivity in the case one of the robots fails. Finally, we present experimental and numerical results confirming the effectiveness of the proposed method.