著者

Kazuho Kobayashi Takehiro Higuchi Seiya Ueno

出版者

International Society of Artificial Life and Robotics

雑誌

Proceedings of the Joint Symposium of AROB-ISBC-SWARM2023

巻号頁・発行日

pp.1491-1496, 2023-01-25

---

This paper proposes a hierarchical and distributed strategy for patrolling missions by robotic swarms, including a fixed base station. One of the essential requirements for autonomous robotic swarms is predictability from human operators. As a clue to satisfy this requirement in patrolling missions, the strategy employs hierarchized algorithms to maintain continuous connectivity to the base station by (i)global patrol and (ii)local patrol. Each robot selects the location to patrol by one of the two algorithms, according to the robot’s role. The paper also introduces a performance metric for the base station’s situational awareness, which may indicate the swarm behaviors’ predictability. The simulation study tested the proposed strategy and compared it to an existing strategy. The proposed strategy demonstrated successful patrol behavior with continuous connectivity to the base station. Though the existing strategy performed better in some aspects, the proposed strategy effectively covered the whole mission area and provided the base station with higher situational awareness.

著者

Kobayashi Kazuho Higuchi Takehiro Ueno Seiya

出版者

International Society of Artificial Life and Robotics

雑誌

Proceedings of the Joint Symposium of AROB-ISBC-SWARM2022

巻号頁・発行日

pp.897-902, 2022-01

---

Swarm robotics requires a practical scheme to maintain supervision by human operators or managers, especially in complicated or life-threatening situations. For this purpose, this paper proposes an algorithm to maintain connectivity between robot swarm and fixed base station during missions. The main idea of the algorithm is maintaining connectivity by role allocation and switching among robots without centralized control by the base station. Our simulation studies have shown no significant inequality of computational cost among robots over the emulated patrol missions. Furthermore, as the total number of robots in the swarm increase, computational cost per robot does not increase significantly. These results have shown the distributed nature and scalability of the proposed algorithms.