Nov. 7th(Thur.) AM9:30-10:30
Title: Improved Mobile Robot Path Planning
Task planning for mobile robots typically proceeds through a structured sequence of sub-functions. Once a target mission is assigned, the pipeline generally includes environment perception and obstacle recognition via onboard sensors (e.g., cameras, RGB-D, LiDAR), configuration-space (C-space) construction to encode robot geometry and kinematic limits, global and local path planning, and finally motion control for safe execution.
This talk presents representative research outcomes from the HRI Lab. at Dongguk University, organized along those sub-functions to highlight how advances at each stage compound into system-level gains. In obstacle recognition, we introduce work on detecting transparent obstacles such as glass or acrylic—cases where conventional camera or laser cues are unreliable — by leveraging alternative visual/reflectance cues and robust sensor processing to reduce false negatives. For path planning, some advancements in the cell decomposition method and the local minima problem in potential field techniques are discussed. Furthermore, a method is presented for improving the performance of RRT-based algorithms in sampling-based path planning by applying the triangle inequality.
Nov. 8th(Sat.) AM8:50-9:50
Artificial intelligence (AI) is increasingly applied in healthcare, but high accuracy alone is not sufficient for real-world clinical use. In medicine, AI systems must not only provide correct answers but also explain why those answers were reached. This is especially true in medical imaging, where building trust, ensuring safety, and maintaining accountability are essential. Explainable AI is therefore critical for AI systems to be accepted and relied upon by healthcare professionals.
This keynote will introduce key design principles for building explainable intelligent systems tailored for medical contexts. Unlike natural image domains, medical data is characterized by small sample sizes, high anatomical variability, and the need for transparent reasoning. Effective AI design in this field must consider domain knowledge, clinical workflows, and the cognitive processes of medical experts.
The talk will present two case studies to illustrate these concepts. The first involves distinguishing between two brain diseases with overlapping symptoms using MRI. The model combines localized expert-defined regions with broader structural information to improve both performance and interpretability. The second case explores pediatric brain development using head CT scans, where a region-based perturbation approach helps reveal which brain areas contribute to developmental age estimation.
These examples demonstrate how designing for explainability enables AI to support, rather than replace, human expertise. The goal is not just to build accurate systems, but to develop intelligent tools that align with the way clinicians think and make decisions. This talk encourages a human-centered approach to AI design and offers insights into how intelligent systems can meet the unique demands of medical environments.
Invited Talk
Speaker: Arunkumar Arulappan
Nov. 6th (Thr.) PM 14:20-15:20
Dr. Arunkumar Arulappan is currently employed with Vellore Institute of Technology (VIT), Vellore, India, as an Assistant Professor in the School of Computer Science Engineering and Information Systems (SCORE). He received his Ph.D. from the Faculty of Information and Communication Engineering, Anna University, Chennai, India, in 2023. He holds a B.Tech. in Information Technology from Anna University and an M.Tech. in Computer Science and Engineering from VIT. Dr. Arulappan’s research interests span a wide range of emerging technologies, including cloud-native deployment, Software-Defined Networking (SDN), Network Functions Virtualization (NFV), 5G/6G networks, AI/ML-based networking, Internet of Vehicles (IoV), and Unmanned Aerial Vehicles (UAV) communications. He is proficient in simulation tools such as MATLAB, ns-3, Mininet, OpenNet VM, and P4 programming. Additionally, he has hands-on experience with various open-source platforms and frameworks, including OpenStack, Cloudify, OPNFV, and tools under the Cloud Native Computing Foundation (CNCF) ecosystem.
Title: Cloud Computing in Smart Grids – Unlocking Efficiency and Scalability in Energy Management
Smart grids are modernized electrical grids that use information and communication technology to gather and act on information about the behavior of suppliers and consumers. It is projected that the global market for smart electricity meters will increase by over 45 percent between 2021 and 2027. While we know that the demand for IoT (Internet of Things) is set to accelerate through 2030, we can expect to see numerous advancements. This allows for real-time monitoring and management of energy distribution, making the grid more efficient, reliable, and sustainable. This talk will address the heart of integration of cloud computing, a technology that has been pivotal in enhancing the capabilities of smart grids. By leveraging the scalability, flexibility, and cost-effectiveness of cloud computing, utilities can improve grid management, enhance customer engagement, and promote energy efficiency. Combining IoT and smart grids promotes a more environmentally friendly, efficient, and flexible energy ecology.
In order to improve the operational and financial stability of electricity distribution companies nationwide, India launched the Ujwal DISCOM Assurance Yojana (UDAY) program, which will be covered in this lecture. Smart grid technologies and the Internet of Things have made it feasible to optimize load distribution, reduce distribution losses, and enable remote grid monitoring. This improved the supply of electricity and reduced financial losses considerably. The cloud’s role in the IoT’s smart grid architecture will only grow in the future, utilizing edge computing, 5G and 6G networks, enhanced security, and advanced analytics. The ongoing shift to a more intelligent, ecologically friendly, and networked grid marks the beginning of a thrilling new period in our quest for energy innovation.
Young Researchers’ Session
Nov. 8th (Sat.) PM 1:30-3:30
Speaker 1: Yoon Suk Kang (Assistant Professor, School of Computer Science, Chungbuk National University, Korea)
Title: Introduction to Hypergraph Mining
Group interactions are prevalent in complex systems such as research collaborations and online Q&A discussions, and they are effectively modeled as hypergraphs. In a hypergraph, hyperedges connect any number of nodes, naturally representing group interactions among entities. Recent studies have shown that hypergraphs are more effective than conventional pairwise graphs for modeling such complex relationships. In this presentation, we introduce the concept of hypergraph mining, which focuses on analyzing and leveraging these higher-order interactions. We also present a hypergraph expansion method—a key step in hypergraph mining—and an evaluation technique for assessing its effectiveness and robustness.
Speaker 2: Dae-Ho Kim (Assistant Professor, Department of Artificial Intelligence Engineering, Chosun University, Korea)
Title: AI-Driven Intelligent IoT Systems: LoRaWAN and UWB Approaches
With the advent of the Internet of Things (IoT) era, the fields of data science and artificial intelligence (AI) have experienced remarkable growth.
Following this trend, AI technologies for IoT systems have also advanced rapidly, and recent studies have focused on intelligent IoT systems based on data collected from the real world.
In this talk, I will introduce two representative examples AI-driven LoRaWAN and AI-applied UWB systems and discuss future directions for the development of intelligent IoT technologies.
Speaker 3: Do-Yup Kim (Assistant Professor, Department of Information and Telecommunication Engineering, Incheon National University, Korea)
Title: Temporal Information-Aware Decision-Making: A Unified Framework for Intelligent Network Optimization
This talk presents a unified framework for temporal information-aware decision-making in dynamic systems that evolve over time. By integrating duality theory with stochastic optimization, the framework systematically addresses long-term performance optimization and lower-bound constraints under time-varying conditions, without requiring prior knowledge of the underlying distribution of system dynamics. It provides a principled approach to balancing efficiency and fairness in decision-making processes. While the proposed methodology has been applied to communication and networked systems, its formulation is general and can be extended to a wide range of intelligent and sustainable optimization problems.
