RIAIoT Lunch Seminar Series: “Learning to Understand and Interact with the 3D World” & “LLM-based Multi-Agent System for Language Learning”
摘要
We are excited to announce the RIAIoT Lunch Seminar Series, featuring two engaging talks from our members: “Learning to Understand and Interact with the 3D World” by Prof. Bo Yang (1st speaker) and “LLM-based Multi-Agent System for Language Learning” by Dr Zhiyuan Wen (2nd speaker). This event will be held on January 27, 2026, at The Hong Kong Polytechnic University.
Abstract
- Learning to Understand and Interact with the 3D World
In this talk, I will present our recent work on addressing the core challenges of machine perception and interaction in complex environments, structured around four essential layers: 3D geometry learning for spatial representation, 3D semantics learning for meaningful interpretation, 3D physics learning for understanding dynamic interactions, and robot skill learning for embodied action. By unifying these layers, our research enables the development of robust world models that empower embodied AI systems to perceive, reason, and act effectively in real-world scenarios.
- LLM-based Multi-Agent System for Language Learning
In the era of globalization, English proficiency is essential for academic, professional, and cross-cultural success. However, traditional English education often struggles to meet diverse learner needs due to rigid curricula, limited authentic practice, and insufficient personalized feedback. Existing AI tools also tend to focus on isolated skills or generic content, leading to plateaus in learners’ language development. To address these challenges, we are developing a multi-agent platform powered by large language models (LLMs) that integrates immersive scenario simulation, adaptive tutoring, and longitudinal progress tracking. The platform offers two core learning modes. The “one-on-one” personalized tutoring mode enables AI tutors to assess learners’ proficiency, design tailored study plans, and deliver real-time feedback. The “many-to-one” scenario-based simulation mode immerses learners in authentic contexts, such as job interviews and academic debates, where they interact with multiple AI agents playing various roles. By seamlessly combining structured guidance with contextualized practice, our platform aims to help learners confidently transfer foundational language skills into real-world communication situations.
Details
- Title: “Learning to Understand and Interact with the 3D World” & “LLM-based Multi-Agent System for Language Learning”
- Date: 27 January 2026 (Tuesday)
- Time: 12:00 n.n. – 1:30 p.m.
- Venue: CD620, 6/F, Hong Kong Chinese Manufacturers' Association Building (Block C), The Hong Kong Polytechnic University (Capacity: 30, first-registered, first-served)
- Lunch Arrangement: A light lunch set (Food & drink from the PolyU gourmet shop) will be provided. You can choose your preferred option in the registration link.
- Register Methods: click https://polyu.hk/fTDXT or scan the QR code on the poster above
The seminar is primarily open to RIAIoT members, but please feel free to forward this email to invite Postdocs/PhD students in your team to register and attend. Kindly reply by January 23 (Friday).
