Greetings!
We are pleased to invite you to an expert seminar. Please refer to the details below and we look forward to your participation.
= Details =
1. Venue: L407
2.Date & Time: March 19, 2025 (Wednesday), 16:30-17:30
3. Purpose: Invited Expert Seminar (Dr. Jongyun Kim)
4. Seminar Title: A Bio-Inspired UAV Flocking Method Using Reinforcement Learning
5. Abstract:
Many animals in nature exist within collective groups. Ants, bees, herring, geese, and even humans can be observed living in various forms of collectives. Within certain animal groups, researchers have identified inactive individuals, also termed as ‘lazy workers’, exhibiting reduced activity patterns. While these individuals appear to exhibit fewer behaviors necessary for the current task of the collective, research on these inactive individuals suggests they can provide benefits to the collective task performance. This talk presents an approach that applies the concept of ‘lazy workers’ to UAV flocking. The approach operates at a higher level of an existing flocking controller, specifically by introducing the concept of ‘laziness’ to the inputs or outputs of the controller. Numerical simulation results will be presented to discuss the potential advantages of this strategy regarding resource utilization and task performance within UAV flocking systems.
6. Biography:
Dr. Jongyun Kim has been a Research Associate in the Department of Mechanical, Aerospace and Civil Engineering at the University of Manchester since July 2024. He received his BSc and MSc degrees in Mechanical and Aerospace Engineering from the Ulsan National Institute of Science and Technology, South Korea, in February 2018 and February 2020, respectively. In January 2025, he completed his PhD at Cranfield University, United Kingdom, with a thesis titled “Flocking Control Methods for a Multi-Agent System Using Deep Reinforcement Learning and Lazy Workers.” His research interests include multi-agent systems, sparse rewards in reinforcement learning, autonomous drone landing, and communication relays.
