Prof. Kiyoshi Kiyokawa
Nara Institute of Science and Technology, Japan
 

Speech Title: From Augmented Reality to Proactive Vision Care: Two Decades of Co-evolving XR and AI in Advanced Eyewear

Abstract: Over the past two decades, head-mounted displays (HMDs) have evolved from devices that overlay information onto the world into intelligent eyewear that senses, corrects, and cares for human vision itself. In this keynote, I trace this trajectory through my own research, organized around three movements. The first is the pursuit of the "ideal" display—a decades-long quest for HMDs that are compact and lightweight yet offer a wide field of view, high angular resolution, correct focus cues, faithful color reproduction, accurate calibration, low latency, and proper mutual occlusion—illustrated by wide field-of-view optical see-through designs based on retro-transmissive screens and a series of occlusion-capable displays. The second, and the heart of this talk, is sensing the eye: a progression from corneal-feedback AR, in which corneal imaging and deep learning enable calibration-free gaze and focus estimation, eye-contact detection, and even the diagnosis of abnormal eye movements, to wearable measurement of eye movements, electro-oculography, and pupillometry that infer refractive error, blink rate, and visual fatigue in daily life, and onward to a new generation of CMOS sensors that capture the ocular surface and even the axial length of the eye. The third is flexible problem-solving through AR and vision augmentation: enhancing, attenuating, or reorganizing what we see to assist people with strabismus, low vision, or visual hypersensitivity, and to support everyday tasks. I argue that the convergence of XR and AI is now turning continuous, multi-dimensional eye data into early prediction of ocular disease. This conviction has led us to launch a new large-scale research program on "proactive vision care"—advanced eyewear that watches the eyes to anticipate risk and intervene before symptoms appear, as a step toward preventive, personalized medicine. I close with reflections on what truly benefits the human behind the glasses.

Biography: Kiyoshi Kiyokawa is a Professor at the Nara Institute of Science and Technology (NAIST), where he leads the Cybernetics and Reality Engineering (CARE) Laboratory. He is a distinguished researcher in virtual reality (VR), augmented reality (AR), and human augmentation. Professor Kiyokawa received his M.S. and Ph.D. degrees from NAIST in 1996 and 1998, respectively. His career includes positions as an Associate Professor at Osaka University, a researcher at the Communications Research Laboratory (now NICT), and a visiting scholar at the University of Washington’s Human Interface Technology Laboratory. His significant contributions have been recognized with numerous accolades, including the 2022 IEEE VGTC Virtual Reality Technical Achievement Award, the inaugural 2022 IEEE VGTC Virtual Reality Service Award, and the title of Fellow from the Virtual Reality Society of Japan (VRSJ).

Professor Kiyokawa's research has resulted in several pioneering technical achievements. He is known for developing advanced head-mounted display (HMD) systems, including ELMO, the first occlusion-capable optical see-through HMD in 1999. His foundational work also includes VLEGO, one of the first collaborative immersive modelers, and SeamlessDesign, which featured the first transitional interface for switching between VR and AR. His research extends to vision augmentation and assistive interfaces, collaborative virtual and augmented reality, and innovative multimodal interfaces.

Beyond his research, Professor Kiyokawa has demonstrated a profound dedication to the academic community through extensive service and leadership. He has served on the Steering Committees for top-tier conferences, including IEEE VR, IEEE ISMAR, and IEEE 3DUI. His leadership roles are numerous, having served as General Co-Chair for IEEE VR 2019 in Osaka, which was the largest in-person conference in its history at the time. Additionally, he is on the Editorial Board of IEEE Transactions on Visualization and Computer Graphics (TVCG) and has frequently been a Board Member of the VRSJ.

 

 

 

Prof. Shin'ya Nishida
Kyoto University, Japan
 

Speech Title: Toward Digital Twins of Human Visual Perception

Abstract: Virtual and mixed reality technologies seek to reproduce the sensory experiences of the real world. Yet, faithfully simulating every aspect of human sensory input remains computationally infeasible. Practical systems must therefore simplify or omit information—but ideally in ways that users never notice. Achieving this goal requires not only advances in engineering but also a deep understanding of human perception. In other words, effective VR/MR systems should exploit the characteristics and limitations of the human visual system, allowing them to “fool the brain” without degrading the perceived experience.

Traditionally, the development of immersive systems has relied heavily on user studies. While indispensable, human experiments are inherently limited by practical and ethical constraints, making it difficult to exhaustively explore the vast design space of VR/MR systems. This motivates a new paradigm: replacing the human component of the conventional framework with a digital twin of human perception. Such a perceptual digital twin would enable rapid evaluation and optimization of system designs while explicitly accounting for human perceptual characteristics.

Recent advances in computer vision have made computational models of vision far more powerful than ever before, in some cases surpassing human performance on visual recognition tasks. However, high performance alone does not guarantee that these models perceive the world as humans do. To serve as perceptual digital twins, computational models must reproduce not only human-level performance but also human perceptual behavior, including its strengths, limitations, and systematic biases. Achieving this goal requires both biologically and psychologically informed machine models and large-scale human perceptual datasets that allow direct comparisons between model predictions and human behavior.

In this keynote, I will discuss our recent efforts toward building digital twins of human visual motion perception, focusing on computational models and benchmark datasets that capture human characteristics.

Biography: Shin’ya Nishida is Professor at the Graduate School of Informatics, Kyoto University, and former Senior Distinguished Scientist at NTT Communication Science Laboratories, Japan.

His research focuses on human sensory information processing, including visual motion perception, time perception, material perception, tactile perception, and multisensory integration. Although originally educated in psychology at Kyoto University, he pursued a broad spectrum of research ranging from fundamental perceptual science to engineering-oriented studies during his long career at NTT laboratories. His work combines psychophysics, cognitive neuroscience, computational modeling, and engineering approaches to understand human perceptual intelligence. His recent interests include the use of machine vision systems to better understand human visual intelligence. He is widely recognized as one of Japan’s leading vision scientists and has served on the editorial boards of major journals in the field, including Journal of Vision, Vision Research, and Annual Review of Vision Science.

He has also played leading roles in large-scale interdisciplinary research initiatives, including the Japanese national projects “Innovative Shitsukan Science and Technology” (2015–2020) and “Deep Shitsukan” (2020–2025), both focusing on the science of material and sensory perception. He currently serves as Sub-program Director of JST Moonshot Goal 9.

He has received numerous honors, including the Japan Society for the Promotion of Science Prize (2006), the MEXT Prize for Science and Technology (2015), the Special Prize of the Japanese Psychological Association International Award (2023), and the Medal with Purple Ribbon from the Japanese government (2024).

 

 

 

Prof. Mayuri Mehta
Sarvajanik College of Engineering and Technology, India
 

 

Speech Title: Transforming Healthcare with AI: Emerging Trends, Applications, and Future Research Directions

Abstract: Artificial Intelligence (AI) is driving the transformation of next-generation healthcare by enabling intelligent, data-driven, and patient-centric solutions. The rapid growth of electronic health records, medical imaging, genomics, wearable sensors, and real-time monitoring systems has generated vast volumes of heterogeneous healthcare data. Advanced AI techniques are increasingly being leveraged to extract meaningful insights from this data, thereby improving clinical decision-making, diagnosis, and treatment planning.

Recent advancements in AI are reshaping healthcare applications such as disease diagnosis, medical image analysis, robot-assisted surgeries, biomedical wearables, personalized medicine, drug discovery, bioinformatics, telemedicine, and healthcare analytics. Despite these advancements, critical challenges remain, including data privacy, model interpretability, bias, regulatory compliance, and ethical deployment. Addressing these concerns is essential for building trustworthy and scalable healthcare systems.

This session provides a comprehensive overview of emerging AI trends and their applications in healthcare including use cases, associated challenges, and future research directions. It offers an interdisciplinary perspective, equipping participants from academia, industry, and healthcare domains with insights into the evolving AI-driven healthcare ecosystem.

Biography: Dr. Mayuri Mehta is a Professor of Computer Engineering at Sarvajanik College of Engineering and Technology, India, with over 25 years of academic, research and leadership experience. She acts as the institute’s International Relations and External Affairs Officer and leads the AI Task Force at Sarvajanik University.

Her research work focuses on Applied AI and Data Science, Medical Image Analysis, Health Informatics, and Computer Vision, with particular interest in AI for healthcare and societal impact.. She has delivered more than 150 invited talks, keynote lectures, and technical sessions at international conferences, universities, and professional forums across the world. Her talks have been hosted by institutions including Imperial College London, Coventry University & Ulster University in UK, University of Rhode Island in USA, and Pwani University in Kenya along with numerous IEEE International conferences & IEEE international sections.

She owns 18 patents, 6 published books and 60+ research papers, and has secured multiple research grants. Her contributions to engineering education and research have been recognized through multiple honors, including the ‘Best Paper Awards’, the ‘Nation Builder Award (Rotary District 3060)’, ‘Best Teacher Award’ by 112 years old philanthropic Sarvajanik Education Society, and ‘Researcher of the Year Award (Engineering – Female)’. She was also featured in the “Women in AI” initiative by INDIAai (INDIAai.gov.in), recognizing her contributions to Artificial Intelligence research and education (Women in AI on INDIAai).

 Dr. Mehta is a Senior Member of IEEE and an active member of IEEE societies including Women in Engineering, EMBS, and SPS. She is also a Lifetime Member of professional bodies such as ISTE and CSI.

 

 

 

Prof. Wei-Chang Yeh
ASPEED, NTHU, and CYCU Chair Professor, National TsingHua University, Taiwan
 

 

Speech Title: Agentic AI for Multimodal Intelligence, Virtual Environments, and Academic Innovation

Abstract: This keynote presents an agentic AI framework for building reliable, multimodal, and human-centered intelligent systems. In the context of image and signal processing, artificial intelligence, virtual reality, and immersive environments, the talk explores how AI agents can move beyond passive assistance toward closed-loop perception, reasoning, action, evaluation, and self-correction.

The proposed framework organizes multiple cooperating agents to support complex knowledge workflows. These agents can process heterogeneous information such as text, images, signals, simulation results, presentation materials, review comments, and user feedback. Through structured coordination, they transform fragmented academic and engineering tasks into traceable, reusable, and improvable workflows.

Five representative agents are introduced. The NarratorAgent supports multimodal presentation generation by connecting slides, scripts, voice narration, and visual explanation. The ReviewerAgent assists with manuscript review, response analysis, and technical consistency checking. The ThesisAgent provides longitudinal support for graduate research by tracking research progress, argument quality, and revision history. The TeachingAgent enables adaptive instruction by combining student feedback, learning materials, and interactive explanation. The OrchestratorAgent coordinates task routing, model selection, verification, and human-in-the-loop decision control.

For ICISPC and AIVR audiences, the keynote highlights how agentic AI can be connected with multimodal signal interpretation, visual reasoning, virtual simulation, digital twins, and immersive learning environments. In such settings, agents must not only generate content but also interpret visual and temporal information, communicate across tools and platforms, and maintain reliability under uncertainty. This raises important design questions: how should multimodal evidence be fused, how should errors be detected, how should agents verify one another’s outputs, and when should human judgment override automated decisions?

Drawing on practical deployment experience in AI robotics, digital twins, AMR navigation reliability, and academic AI workflows, the talk discusses transferable lessons on system architecture, reliability control, privacy-aware operation, and human-AI collaboration. The central message is that agentic AI should augment, not replace, human experts. By combining multimodal perception, virtual environments, and structured agent collaboration, future AI systems can reduce repetitive workload, improve decision consistency, and support more creative, reliable, and human-centered innovation.

Biography: Dr. Wei-Chang Yeh is the ASPEED Chair Professor and Chair Professor in the Department of Industrial Engineering and Engineering Management at National Tsing Hua University (NTHU), Taiwan. He also serves as Chair Professor at Chung Yuan Christian University. He received his M.S. and Ph.D. degrees in Industrial Engineering from the University of Texas at Arlington.

Dr. Yeh’s research focuses on algorithm design, exact solution methods, soft computing, network reliability, AI-enabled decision systems, and NP-hard optimization problems. He has published more than 300 SCI-indexed journal papers and holds more than 70 patents. Since 2020, he has been listed among Stanford/Elsevier’s Top 2% Scientists worldwide for both career-long and single-year impact. His major honors include two Outstanding Research Awards, one Distinguished Scholars Research Project Award, and two Overseas Research Fellowships from Taiwan’s MOST/NSTC.

He currently serves as an Associate Editor for IEEE Transactions on Reliability, IEEE Access, and Reliability Engineering & System Safety. He is the proposer of Simplified Swarm Optimization (SSO) and the Binary-Addition-Tree (BAT) framework. Dr. Yeh is also an NVIDIA University Ambassador for the Deep Learning Institute and has received NVIDIA research grant support. His recent work connects AI robotics, digital twins, reliability-aware navigation, and intelligent decision support for advanced manufacturing and autonomous systems.

 

 

•