Large-Scale Swarm Control in Cluttered Environments
Author | Elsayed, Saber |
Author | Mabrok, Mohamed |
Available date | 2024-02-28T05:17:08Z |
Publication Date | 2024-01-01 |
Publication Name | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
Identifier | http://dx.doi.org/10.1007/978-981-99-8715-3_32 |
Citation | Elsayed, S., & Mabrok, M. (2023, December). Large-Scale Swarm Control in Cluttered Environments. In International Conference on Social Robotics (pp. 384-395). Singapore: Springer Nature Singapore. |
ISBN | 9789819987146 |
ISSN | 03029743 |
Abstract | In the evolving era of social robots, managing a swarm of autonomous agents to perform particular tasks has become essential for numerous industries. The task becomes more challenging for large-scale swarms and complex environments, which have not been fully explored yet. Therefore, this research introduces a methodology incorporating multiple coordinated robotic shepherds to effectively guide large-scale agent swarms in obstacle-laden terrains. The proposed framework commences with deploying an unsupervised machine-learning algorithm to categorise the swarm into clusters. Then, a shepherding algorithm with coordinated robotic shepherds drives the sub-swarms towards the goal. Also, a path planner based on an evolutionary algorithm is proposed to help robotic shepherds move in a way that minimises the dispersion of each sub-swarm and avoids potential hazards and obstructions. The proposed approach is tested on different scenarios, with the results showing a success rate of 100% in guiding swarms with sizes up to 3000 agents. |
Language | en |
Publisher | Springer Science and Business Media Deutschland GmbH |
Subject | Large-scale Path Planning Robotic Shepherding Swarm Control |
Type | Conference Paper |
Volume Number | 14453 LNAI |
Files in this item
Files | Size | Format | View |
---|---|---|---|
There are no files associated with this item. |
This item appears in the following Collection(s)
-
Mathematics, Statistics & Physics [726 items ]