• Speaker: Dr. Jongyun Kim (The University of Manchester)

• Date: March 19, 2025, 16:30-17:30

• Venue: F407, Munji Campus

• 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 less exhibit the behaviours 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 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.

• Speaker: Dr. Hangju Cho

• Date:
  - January 12, 2025, 13:30-15:30
  - January 13, 2025, 13:30-16:30
  - January 14, 2025, 13:30-16:30

• Venue: L409, Munji Campus

• Outline:
  I. Some Bits of Math
  II. Parameter Optimization
  III. Introduction to Optimal Control
  IV. Linear Quadratic Optimal Control Problem
  V. Application to Optimal Guidance Problems

• Speaker: Prof. Inmo Jang (Korea Aerospace University)

• Date: January 10, 2025, 15:00-16:00

• Venue: L409, Munji Campus

• Abstract: This seminar explores the practical use of ChatGPT in research-focused software development, using the SPACE simulator project (http://space-simulator.rtfd.io/) as a case study. The focus is on three key areas: rapid prototyping, exploring new frameworks, and managing research codebases effectively. The discussion begins with how GPT facilitates the creation of functional code prototypes and their refinement, enabling researchers to iterate quickly and focus on experimental goals. Next, the seminar explores how GPT assists in understanding unfamiliar frameworks like ReadTheDocs by offering structured guidance, helping researchers grasp rules and best practices more efficiently. Additionally, GPT’s role in setting up CI pipelines through GitHub Workflows is highlighted, streamlining automated testing to ensure reproducibility and reliability of research software. Finally, tools for automating code reviews with GPT are introduced, showcasing how minimal setup effort can lead to significant time savings and improved code quality.

• Speaker: Prof. Namhoon Cho (Cranfield University)

• Date: January 8, 2025, 11:30-12:00

• Venue: F407, Munji Campus

• Abstract: Adaptive control is known as an approach that enables stable control under uncertainties without sacrificing nominal performance for enhanced robustness. However, it is not always straightforward to realise the attractive benefit in practice due to added complexity of adaptive control systems. A design automation framework that performs optimisation over the feasible space of controllers is believed to be potentially useful for alleviating the difficulties in the design processes. As a first step to formulate a controller optimisation problem, we desire to characterise the admissible design space to the most complete extent possible so that the search can be performed within the widest possible range of achievable performances. In this talk, we will discuss different facets of adaptive control systems focusing on how each element can be generalised to establish a flexible parameterisation of stable adaptive controllers that can be useful for tuning. Specifically, new ideas that extend model reference adaptive control beyond the classical architecture and adaptation law will be presented.

• Speaker: Prof. Mirko Kovac (Imperial College London & EPFL)

• Date: December 9, 2024, 10:30-11:30

• Venue: Haedong Lecture Hall, Mechanical Engineering Building (N7)

• Abstract: In future, robotics and AI will be an indispensable tool for data collection in complex environments, enabling the digitalisation of forests, lakes, off-shore energy systems, cities and the polar environment. However, such future robot solutions will need to operate more fiexibly, robustly and efficiently than they do today.

This talk will present how animal-inspired robot design methods can integrate adaptive morphologies, functional materials and energy-efficient locomotion principles to enable this new class of environmental robotics. The talk will also include application examples, such as flying robots that can place sensors in forests, aerial-aquatic drones for safe operations in underground and tunnel systems.