Keynote Speakers-CRC 2020

Prof. Biao Huang, University of Alberta, Canada (IEEE Fellow)

Bio: Biao Huang obtained his PhD degree in Process Control from the University of Alberta, Canada, in 1997. He had MSc degree (1986) and BSc degree (1983) in Automatic Control from the Beijing University of Aeronautics and Astronautics. He joined the University of Alberta in 1997 as an Assistant Professor in the Department of Chemical and Materials Engineering, and is currently a Full Professor and NSERC Senior Industrial Research Chair. He is an IEEE Fellow, Fellow of the Canadian Academy of Engineering, Fellow of the Chemical Institute of Canada, and recipient of a number of awards including Germany’s Alexander von Humboldt Research Fellowship, APEGA Summit Award in Research Excellence, ASTech Outstanding Achievement in Science and Engineering Award and a best paper award from the Journal of Process Control. Biao Huang’s research interests include: process data analytics, machine learning, system identification, image processing, fault detection and isolation, and soft sensors. He has applied his expertise extensively in industrial practice. He has published 5 books and over 390 SCI journal papers. Biao Huang currently serves as the Editor-in-Chief for IFAC Journal Control Engineering Practice, Subject Editor for Journal of the Franklin Institute, and Associate Editor for Journal of Process Control.

Speech Title: Transfer Learning - Overview and Introduction

Abstract: With advance in computing power and multisensory technology, massive data accumulated can be processed by data-driven techniques to learn underlying driving forces and/or hidden patterns for better monitoring and control of industrial processes. In order to train a data-driven model with satisfactory performance, sizable amounts of labelled data are required. However, it is often expensive, time consuming and labor-intensive to gather well-labelled data although massive data are being collected. It becomes even more demanding to collect such labelled data for an industrial plant under situations such as, for example, when the plant has just started operation, the labels are the faulty events that are used to train a monitor, or the labels are lab data that are used to develop a predictive model. This is commonly referred to as the cold-start problem, where decisions have to be made while little or almost nothing has been known about the environment. In the context of the process industry, how to transfer the knowledge learned from some existing processes with well-labelled data into a related target process that has limited labels to establish a satisfactory model is the transfer learning. This presentation will give an overview and introduction to transfer learning.


Prof. Graziano Chesi, The University of Hong Kong, Hong Kong (IEEE Fellow)

Bio: Graziano Chesi is a Professor at the Department of Electrical and Electronic Engineering of the University of Hong Kong. He received the Laurea in Information Engineering from the University of Florence and the PhD in Systems Engineering from the University of Bologna. Before joining the University of Hong Kong, he was with the Department of Information Engineering of the University of Siena. He served as Associate Editor for various journals, including Automatica, the European Journal of Control, the IEEE Control Systems Letters, the IEEE Transactions on Automatic Control, the IEEE Transactions on Computational Biology and Bioinformatics, and Systems and Control Letters. He also served as chair of the Best Student Paper Award Committees for the IEEE Conference on Decision and Control and the IEEE Multi-Conference on Systems and Control. He is author of the books "Homogeneous Polynomial Forms for Robustness Analysis of Uncertain Systems" (Springer 2009) and "Domain of Attraction: Analysis and Control via SOS Programming" (Springer 2011). He was elevated to IEEE Fellow for contributions to control of nonlinear and multi-dimensional systems upon evaluation by the IEEE Control Systems Society.

Speech Title: Measuring the Entropy in Parametric Linear Systems

Abstract: This talk investigates the entropy measure in parametric linear systems, which is a key index in control with communications constraints. The problem consists of determining the largest entropy measure in linear systems depending polynomially on parameters constrained in a semialgebraic set. It is shown that upper bounds of the sought measure can be established via linear matrix inequality feasibility tests. Moreover, a priori and a posteriori conditions for establishing nonconservatism are proposed. Finally, two special cases of the proposed methodology are investigated: the first one concerns systems with a single parameter, and the second one concerns the determination of the largest spectral abscissa and radius.


Prof. Li Youfu, City University of Hong Kong, Hong Kong

Bio: You-Fu Li received the PhD degree in robotics from the Department of Engineering Science, University of Oxford in 1993. From 1993 to 1995 he was a research staff in the Department of Computer Science at the University of Wales, Aberystwyth, UK. He joined City University of Hong Kong in 1995 and is currently professor in the Department of Mechanical Engineering. His research interests include robot sensing, robot vision, and visual tracking. In these areas, he has published over 300 papers including over 150 SCI listed journal papers. He has served as an Associate Editor for IEEE Transactions on Automation Science and Engineering (T-ASE), Associate Editor and Guest Editor for IEEE Robotics and Automation Magazine(RAM), and Editor for CEB, IEEE International Conference on Robotics and Automation (ICRA).

Speech Title: Towards 3D Gaze tracking and its applications for human-robot cooperation

Abstract: Gaze tracking can be useful for communicating human intentions to the robot as an interface. However, tracking an object via estimating the 3D points of gaze (POG) in space is very difficult to achieve. In this talk, I present a gaze estimation framework for head-mounted gaze tracking systems (HMGT). The primary challenges here include the tedious calibration with conventional approach, the extrapolation and parallax errors, and large visual axe errors. I will report our work in addressing these challenges and developing an HMGT system for flexible implementation. A new data acquisition method is proposed to collect the training gaze data. Instead of successively fixating at a grid of calibration points, the user looks at a point while rotating his head, resulting in densely distributed calibration points over a wider field of view. An outlier removal method is proposed based on smooth-pursuit to identify the user’s distractions during the calibration process thereby improving the reliability of the training data. A regression model is constructed to effectively approximate the gaze mapping functions. A region-wise regression model for 2-D POG estimation is extended for back-projecting the pupil centers. With the visual axes of both eyes estimated, the 3-D gaze point can be inferred as the point closest to two visual axes. To solve the problem of the inflexible calibration procedures and significant errors in depth estimation for head-mounted eye tracking systems, we also propose to adopt an RGBD camera as the scene camera of our system, and present a 3D gaze estimation approach with an auto-calibration method. When the calculated 3-D POG in the scene camera coordinates is transformed to the world coordinates, it can be used in human-robot cooperation for different manipulation tasks.

Prof. Xuebo Zhang, Nankai University, China

Bio: Dr. Xuebo Zhang received the B.Eng. degree in Automation from Tianjin University in 2006, China, and the Ph. D. degree in Control Theory and Control Engineering from Nankai University in 2011, China. From July 2011, he joined the Institute of Robotics and Automatic Information Systems (IRAIS), Nankai University, China. He is currently a full Professor, and he serves the deputy head of IRAIS and also the deputy head of Tianjin Key Laboratory of Intelligent Robotics (TJKLIR), Nankai University, China. His research interests include planning and control of autonomous robotic and mechatronic systems with focus on time-optimal planning and visual servo control; intelligent perception including robot vision, visual sensor networks, SLAM, etc. He is a Technical Editor of IEEE/ASME Transactions on Mechatronics and an Associate Editor for ASME Journal of Dynamic Systems, Measurement and Control.

Speech Title: Real-time optimal motion planning for autonomous robotic systems

Abstract: As one of the most fundamental research topics in robotics, high-performance motion planning is still challenging in many applications because it needs to consider many constraints as well as many performance indexes simultaneously, such as real-time requirements, motion efficiency, safety issues, smoothness, uncertain environmental constraints, robot kinematic and dynamic constraints, and so on. In this talk, I will discuss the recent developments in both global and local planning. For global planning, I will talk about a new heuristic-guided approach to increase the computational efficiency with lower memory consumption. For local planning, a decoupling framework consisting of local path planning and velocity planning, will be discussed. A sparse optimization approach to generate a safe and smooth local path will be presented, and then a real-time, complete and time-optimal velocity planning approach that was proposed to generate the most efficient trajectory along the preplanned path with rigorous mathematical proofs. Such a hierarchical motion planning framework has been experimentally verified, and comparative results showed its remarkable improvements in terms of computational efficiency, motion efficiency, and the trajectory flexibility to help robot navigation in challenging environments.