Distributed algorithms play a central role in modern networked systems, where multiple agents must coordinate, reach agreement, or solve optimization problems using only local interactions. In this talk, I will present a robustness-oriented perspective on three representative classes of distributed algorithms: consensus algorithms, event-triggered consensus algorithms, and distributed optimization algorithms. The goal is to show that, despite their different formulations, these problems can be studied within a unified framework centered on robustness, performance, and network effects.

Sustainability has become a defining concern in modern engineering, driven by global challenges such as climate change, energy scarcity, and intensifying pollution. Efforts to reform engineering infrastructure to meet sustainability criteria require a re-evaluation of mechanical design and validation principles, which are increasingly insufficient in light of emerging fabrication technologies and unprecedented service conditions. In this talk, I will introduce our work on two problems central to sustainable engineering: the extreme mechanical behavior of advanced alloys in hydrogen environments—the backbone of the emerging hydrogen economy—and the mechanical validation of additively manufactured structures, which is essential for the industrial-scale fabrication of low-cost, durable metallic structures. Through a combination of multiscale modeling methods, we reveal the mechanisms of hydrogen–microstructure interactions and the porosity-governed deformation and fracture behavior of additively manufactured heterostructured materials. Our work advances the understanding of extreme mechanics in metallic materials and lays the foundation for mechanics-guided design of high-performance structural materials for sustainable engineering.

With the rapid development of artificial intelligence, robotics, and autonomous systems, vision-based learning is becoming a key technological pathway for enabling intelligent systems to perceive environments, understand tasks, and make autonomous decisions. The seminar will focus on frontier research such as machine vision, deep reinforcement learning, multi-agent systems, and embodied intelligence, and will analyze how visual information can be integrated with control, planning, and learning mechanisms to enable agents to achieve efficient perception, autonomous learning, and safe operation in complex environments.

Modular autonomous vehicles (MAVs) are transforming transit design through en-route (de)coupling and in-motion transfer. To evaluate this disruptive technology, we conduct a comprehensive cost-effectiveness analysis of an MAV-based system featuring a main corridor service integrated with local pick-up-and-delivery routes. We develop a continuum approximation (CA) model to quantify passenger and agency costs, comparing it against traditional fixed-route and on-demand services. Our analysis uncovers the complex interactions between system parameters and performance, identifying favorable operating regimes and providing managerial guidance for modular transit implementation.

3月25日，SYSH Hub Forum第七场学术论坛在香港科技大学（广州）W1 521S圆满举办。本次论坛主题为《材料科学与工程的职业发展之路》，由美国国家工程院院士Stephen Tsai担任主讲嘉宾。

With the rapid advancement of connectomics and large-scale neural activity recording technologies, neuroscience is entering a new era in which both high-resolution structural information and large-scale functional data can be simultaneously acquired. However, how to build computational models starting from the connectome that can quantitatively predict neural dynamics and brain function remains a central challenge in contemporary neuroscience. This talk focuses on the theme of connectome-driven differentiable digital twin brain modeling, presenting our recent series of research advances in the direction of differentiable brain simulation. We have developed a suite of core tools, including BrainPy, a multi-scale differentiable brain simulation framework; BrainUnit, a physical unit system; and BrainTrace, a large-scale online learning system, forming a systematic technology stack encompassing large-scale simulation, event-driven acceleration, physical unit constraints, and online learning optimization. Building upon this foundation, we have preliminarily constructed differentiable whole-brain digital twin models for Drosophila (fruit fly) and zebrafish, providing a novel computational pathway for quantitatively investigating the mapping relationships among connectome structure, neural dynamics, and functional activity.

SYSH Hub Forum是面向系统枢纽专业的高层次学术论坛，旨在邀请相关领域内的顶尖专家与具有影响力的学者分享研究、经验和成果，聚焦学术前沿研究，加强交流与合作。SYSH Hub Forum 每季度举办一次，于2024年春季首发。每次的主讲嘉宾由系统枢纽其中一个学域提名，被提名人必须是该领域内国际公认的专家学者，由院长和学域主任投票同意后进行邀请。

Compared to conventional AI systems, embodied intelligence requires a significantly richer and more diverse set of data modalities, including vision, language, motion dynamics, tactile feedback, depth perception, mesh-level object tracking, scene reconstruction, and audio signals. In this talk, we introduce ModalityNet, a large-scale multimodal data benchmark for embodied intelligence, covering diverse modalities such as vision, language, motion, touch, and depth. We propose a “data pyramid” framework that spans from high-precision teleoperation data to internet and synthetic data, characterizing the trade-off between data diversity and generalization. To bridge the gap between these layers, we introduce two intermediate data types: High-Precision Human-Environment Interaction (HiPHI) data and In-The-Wild (ITW) data. Specifically, HiPHI-Motion with Object and Vision (HiPHI-MOV) focuses on whole-body motion and object interaction, HiPHI-OmniModality (HiPHI-OM) emphasizes fine-grained manipulation with full-modality alignment, and ITW provides diverse behaviors in natural environments. Together, these datasets establish a comprehensive data foundation for embodied intelligence. This talk will also cover the design of these datasets and the corresponding model training paradigms.

周文浩教授现任广州医科大学附属妇女儿童医疗中心主任(院长)、国务院政府特殊津贴专家、二级主任医师、中华医学会儿科学分会副主任委员、中华医学会儿科学分会临床药理学组组长、中国医师协会医学遗传分会副会长、复旦大学特聘教授、博士生导师。他长期致力于新生儿危重症的临床诊疗与转化研究，重点聚焦新生儿脑病和罕见病。周教授建立了中国新生儿神经重症联盟和中国新生儿基因组计划，有力推动了我国新生儿医学的发展。他先后主持国家重点研发计划、国家自然科学基金重点项目等国家级课题16项，在Nature、Developmental Cell、JAMA Pediatrics、Brain、Neuron等国际权威期刊发表SCI论文200余篇，拥有发明专利7项、计算机软件著作权1项。主编《新生儿学》《实用新生儿学》等8部教材及专著。近五年连续入选全球前2%顶尖科学家榜单。曾荣获华夏医学科技奖一等奖、上海市科技进步奖一等奖、中华医学科技奖二等奖、第六届妇幼健康科学技术奖一等奖等多项重要奖项。

蔡教授将回顾其在材料科学与工程领域的长期学术生涯、在中美两国的求学与生活经历，以及工程科学的发展趋势。他将介绍其在复合材料失效模式与刚度研究中的重要成果，并分享其近年研发并推广的全新复合材料制备技术 —“双双层” 铺层工艺（Double‑double lamination）。该技术性能优于传统 “四层” 铺层（quad lamination），被美国国家航空航天局（NASA）的同事称为“真正具有革命性”。