The ride-hailing service platforms have grown tremendously around the world and attracted a wide range of research interests. This talk will introduce some recent progress we have made on two key issues of ride-hailing service: demand forecasting and order matching. A key to ride-hailing service platforms is how to realize accurate and reliable demand prediction. However, most of the existing studies focus on region-level demand prediction while only a few attempts have been made to address the problem of origin-destination (OD) demand prediction. In our recent studies, we constructed dynamic OD graphs to describe the ride-hailing demand data from a graph perspective. We proposed two novel neural architectures named the Dynamic Node-Edge Attention Network (DNEAT) and the Dynamic Auto-structuring Graph Neural Network (DAGNN) to address the unique challenges of OD demand prediction from the demand generation and attraction perspectives. Online matching between idle drivers and waiting passengers is another key component in a ride-sourcing system. We proposed two ideas to improve the matching efficiency: early driver matching and delayed passenger matching. The first idea considers the vehicles whose destinations are close to the passenger’s origin so that the passenger’s waiting time may be shorter and the vehicle’s pick-up distance and fuel consumption can be saved. The second one assumes that a specific passenger request can benefit from a delayed matching since he/she may be matched with closer idle drivers after waiting for a few seconds. The problems are solved by reinforcement learning methods. Through extensive empirical experiments with well-designed simulators, we show that the proposed frameworks can remarkably improve system performance.

Proteogenomic identification of translated small open reading frames in human has revealed thousands of microproteins, or polypeptides of fewer than 100 amino acids, that were previously invisible to geneticists. Microproteins have now been linked to important cellular and physiological processes as well as disease in diverse organisms. We developed a high-throughput technology for global mapping of microproteins to specific subcellular organelles by proximity labeling techniques and mass spectrometry-based proteomics. We demonstrated that this technology enables detection of hundreds of microproteins within specific subnuclear regions in cultured human cell lines and in mouse model. Meanwhile, we characterized the biological function of an intrinsically disordered microprotein, NBDY, and showed that NBDY is a master regulator of the substrate specificity of the cytoplasmic RNA decapping complex. We identified that phosphorylation of NBDY from phosphoproteomics downstream of signaling pathways that control cell division and response to growth factors is sufficient to drive the membraneless organelle (Processing bodies) disappearance in human cells.

Metal additive manufacturing (AM) is a family of rising technologies that are well-known for their flexibility in fabricating geometrically and/or material-wisely complex metal parts that are difficult for traditional manufacturing processes. However, metal AM processes are often plagued with a lack of reproducibility and reliability. Computational modeling has become an appealing approach in understanding the Process-Structure-Property (PSP) relationship in AM processes, critical for process control and optimization.

Transportation has been and will continue to beunder rapidtransformation.Onesuchtransformationistransportation systems'increasing reliance on massive datasets, includingthose from vehicles (especially CAVs) and othertransportationusers, from the (intelligent)infrastructure,the data communicated betweenvehicles and the infrastructure, and other sources.However, the growing reliance on data poses criticalcybersecurity issues to transportation systems,among which is the so-called “data poisoning”attacks (DPA). DPAs aim to compromise a system’ sperformance (e.g., deviate traffic state estimation)by adding malicious noises or stealthy perturbationsto the dataset used by the system. Althoughpioneering and practically important, recent studieson DPAs against vehicular data and V2X data havesome limitations, especially in traffic stateestimation and prediction (TSEP) applications,which are often associated with physical constraintsand lack differentiability' in their mathematicalmodels. This work formulates DPAs against TSEPmodels as a general sensitivity analysis ofoptimization problems over data perturbations(attacks) and studies the Lipschitz continuity

Uncertainty widely exists in many engineering problems, such as measurement uncertainty, model uncertainty, parameter uncertainty, and manufacturing uncertainty. These uncertainties will affect the design and manufacture of engineering products. An important issue is how to properly deal with these uncertainties and get more reliable and robust engineering products. In this talk, I will discuss using Bayesian statistics to deal with uncertainty and help provide a better model prediction, especially when the physical model is not accurate (a low-fidelity model), combing experimental measurement data. In addition, I will also discuss using Bayesian statistics for reliability sensitivity analysis, i.e., measuring the effects of random parameters on the failure of structures, which will be helpful for failure control and reliability-based design. Some results will be shown with composite material coupon structures.

Global navigation satellite system (GNSS) plays an irreplaceable role in many areas such as unmanned autonomous systems. In this talk, the evolution of GNSS receiver architectures and the associated baseband signal processing algorithms will be discussed. In specific, the speaker will present his recent research works on vector tracking loop (VTL) and direct position estimation (DPE).

This work addresses a crowd-shippingproblem with transshipment (CSP-T) in thelast-mile delivery, where all requests can besatisfied by either using the own vehicle fleetor outsourcing with a small compensation tocrowd-shippersthrough transshipmentfacilities. The crowd-shippers show theirwillingness to deliver by submitting bids tothe e-commerce company. To minimize boththetravel cost of vehicles and thecompensation of crowd-shippers, the routesof vehicles and the selection of bids need tobe optimized simultaneously. We formulatethe CSP-T into an arc-based formulation anda route-based formulation, where the latter isstrengthened by the subset row inequalitiesBased on the route-based formulation, we

Self-organized microvascular networks (MVNs) have become key to thedevelopment of many microphysiological models. However, the selforganizing nature of this process, combined with variations between types orbatches of endothelial cells (ECs) often leads to inconsistency or failure toform functional MVNs Here, I will present two distinct approaches we havebeen using to tackle those problems.

High-accuracy modeling has integrated andaddressed error sources, improved system's safetymetrics and operational parameters such asaccuracy/integrity,and improved protectionlevel/alert limit. lt is the first time to consolidate allerror sources related to safety. The safe flapping-wing UAV has won the 1st prize @ AVIATIONINDUSTRYCORPORATION OF CHINA CupInternational Unmanned Aerial Vehicle Grand Prix(first author). The safe fixed-wing UAV has won the2nd prize @ AVIATION INDUSTRY CORPORATION OFCHINA Cup - International Unmanned Aerial VehicleGrand Prix Competition. The green aviation taxiingsystem has been developed to address the keytechnologies of. green, aviation , such as greenaviation drive systems. The developed ' green taxisystem has mitigated the

Human interaction with autonomous cyber-physical systems (CPS) is becoming ubiquitous in consumer products, transportation systems, manufacturing, and many other domains. My research aims to develop analytically computable tools to semantically understand and predict human-involved interactions using machine learning & AI techniques with support of data science toward eco-safe deployment of AI-based agents (e.g., autonomous cars) in the human-CPS such as smart cities.