Aug 6 2026 (Thu) 11:00-12:00 | Seminar by Prof. Shengbo Li

Toward Physics-native Intelligence: A Path for World Model, Data Structure and Training Algorithm

Speaker:

Prof. Shengbo Li

School of Vehicle and Mobility & School of Artificial Intelligence

Tsinghua University

Date:    Aug 6 2026 (Thu)

Time:   11:00 am – 12:00 nn

Venue: HW612B, 6/F Haking Wong Building, The University of Hong Kong

Language: Mandarin

Abstract

As artificial intelligence evolves from vision and language to embodied systems such as autonomous driving and robotics, prevailing deep learning paradigms are facing significant challenges. This talk offers a systematic introduction of physics-native intelligence (Phi) and its design principles, as well as how to build its world model, data structure and training algorithms. Some representative physics-native algorithms will also be introduced, including (1) relativistic adaptive gradient descent (RAD) algorithm, which preserves symplecticity to ensure long-term training stability; (2) diffusion actor-critic with entropy regulator (DACER) algorithm, which endows the policy with the capability of multi-modal action distributions; (3) region-wise actor-critic-scenery (RACS) algorithm, which extends the standard binary iterative framework to a ternary one to handle safety constraints, simultaneously learning the feasible region and optimal policy with monotonicity and convergence guarantees; (4) bootstrap off-policy with world model (BOOM) algorithm, which enables bidirectional improvement between policy and world model, internalizing planning capabilities into the policy and thereby improving sample efficiency. Built upon physics-native principles, these algorithms form the core engine of embodied intelligence. The talk will be delivered using Mandarin.

About the speaker

Shengbo Li is a Professor at School of Vehicle and Mobility, and School of Artificial Intelligence, Tsinghua University. Before joining Tsinghua University, he has worked at Stanford University, University of Michigan, and UC Berkeley. His research has established a systematic framework for physics-native intelligence (Phi), and its world models, data structure, and training algorithms. On its basis, he and his team have contributed to absolute safety guarantees and ternary iterative mechanism for reinforcement learning, symplectic neural network optimizer, multimodal latent world model, geometry-aware Bayesian filter, as well as their applications in autonomous driving and robotics. His important awards include National Leading Talents in Sci. and Tech. Innovation in China, Youth Sci. & Tech Award of Ministry of Education, and Youth Sci. & Tech. Innovation Leader in Transportation Sector, National Sci. & Tech. Progress Award in China (Second Prize), National Award for Technological Invention in China (Second Prize), Grand Prize of Science and Technology Award of China in Automotive Industry, Natural Science Award of Chinese Association of Automation (First Prize). He also serves as the director of Technical Committee on AI of SAE-China, deputy director of Technical Committee on Vehicle Control and Intelligence of CAA, and the leader of AI working group in China Industry Innovation Alliance for ICVs.

-   ALL ARE WELCOME   -

Aug 7 2026 (Fri) 14:30-15:30 | Seminar by Prof. Bo Chen

Data-Driven Systems Design for Last-Mile Emergency Logistics: Drone-Based AED Delivery under Uncertainty

Speaker:

Prof. Bo Chen

Warwick Business School, University of Warwick

Fellow of the Academy of Social Sciences (UK)

Date:    Aug 7 2026 (Fri)

Time:   2:30 pm – 3:30 pm

Venue: Tam Wing Fan Innovation Wing Two, The University of Hong Kong

Abstract

Timely intervention is the critical determinant of survival for out-of-hospital cardiac arrest, yet many emergency medical service (EMS) systems fail to meet life-saving response-time thresholds under limited budgets and uncertain demand. Enabled by Internet-plus technologies, new emergency logistics paradigms are emerging, but their effective design remains an open challenge.

This talk presents an Internet-plus emergency logistics framework for drone-based delivery of disposable automated external defibrillators (AEDs). Shifting away from minimizing average or tail response times, the framework focuses on maximizing the number of patients reached within a critical therapeutic window, where intervention has the greatest impact. We formulate this problem as a modular capacitated maximum covering location model, jointly optimizing drone base locations and AED inventories under severe resource constraints and demand uncertainty.

To ensure real-world relevance, the framework integrates distributionally robust optimization, accounts for capacity congestion, and incorporates fairness considerations to mitigate geographic inequities. A large-scale case study using real OHCA data demonstrates substantial gains in timely coverage and equity compared with response-time-based benchmarks. Beyond emergency medicine, the keynote highlights broader lessons for Internet-plus logistics, manufacturing, and service operations, emphasizing threshold-based performance, robust capacity design, and socially responsible resource allocation in digitally enabled service networks.

About the speaker

Dr Bo Chen is a Full Professor at Warwick Business School, University of Warwick, and a Fellow of the Academy of Social Sciences (UK), the Operational Research Society (UK), and the Institute of Mathematics and its Applications (UK). He holds a BSc in Mathematics from Zhejiang University (1982), an MSc from the Institute of Applied Mathematics, Chinese Academy of Sciences (1987), and a PhD from Erasmus University Rotterdam (1994). Professor Chen has held visiting professorships at Stanford University (2003) and the University of Cambridge (2011), and Chair Professorships at Tsinghua University (2005–2011) and, since 2013, at Fudan University. In 2012, he was awarded a Higher Doctorate by the University of Warwick, recognising a lifetime of distinguished contribution to his field.

-   ALL ARE WELCOME   -

Student Seminar by Ms. Busra Bayrak on Jul 29, 2026, 3:00PM

Title: HFCV Adoption Under Externality: Low- and High-Travel-Distance Consumers

Speaker: Ms. Busra Bayrak (Management School, University of Liverpool)

Date: Jul 29, 2026 (Wednesday)

Time: 3:00 pm - 4:00 pm

Venue: Tam Wing Fan Innovation Wing Two, G/F, Run Run Shaw Building, HKU

Registration Link: https://forms.cloud.microsoft/r/sh8i2iympQ

ITS Student Committee will provide a beverage for registered participants.

Abstract: Hydrogen fuel cell vehicles (HFCVs) are increasingly viewed as a viable low-carbon option for transport decarbonization. Yet, HFCV adoption remains limited, mainly due to consumer anxiety associated with safety and convenience under an underdeveloped hydrogen refueling infrastructure. This paper studies HFCV adoption when consumer utility is shaped by network externalities. We develop an analytical adoption model with two heterogeneous customer segments, high-distance fleet users and low-distance private users, and derive equilibrium prices, demands, and profits. Our results show that (i) adoption can progress more rapidly among low-distance users, as segment-level usage intensity and perceived inconvenience shape adoption differently in the presence of externalities, (ii) the equilibrium can exceed the socially optimal adoption level depending on consumer patience and infrastructure maturity, (iii) the planner’s policy should combine infrastructure investment to resolve over-adoption with demand-side subsidies to close the residual under-adoption gap; neither instrument alone is sufficient. The analysis offers insights for manufacturers and policymakers seeking to accelerate HFCV adoption and address infrastructure related anxiety.

Bios: Busra Bayrak is a PhD student in the Department of Operations and Supply Chain Management at University of Liverpool Management School. Her research interest focuses on sustainable operations and supply chain management, particularly the interplay between the triple bottom line, government interventions, consumer behavior, and product adoption/diffusion processes.