The increase in the number of industrial robots used in factories is modest compared to the current needs of industrial robots. It is partially because we do not have enough number of systems integrators, who design robot systems for specific applications by using general purpose industrial robots. Different from human workers, the industrial robots require a lot of peripheral systems specifically designed for given tasks, such as robot hand(s)/gripper(s), fixture(s), parts feeders, smart tools, etc. The current industrial robots, originally designed for universal automation, are far from the 'universal.'

In this presentation, we are going to introduce the concept of the universal manipulation. The goal of the universal manipulation is to create a robot, which executes different tasks using general-purpose tools without designing peripheral systems for each task, without using parts feeders, and without programing of the robots. We discuss the issues for realizing the universal manipulation and introduce some of the solutions for the issues, which include new design of multi-purpose robot hands, integrated control system of visual and impedance servo, teaching by demonstration and so on.

Dr. Kazuhiro Kosuge (Fellow, IEEE) received the B.S., M.S., and Ph.D. in control engineering from the Tokyo Institute of Technology, in 1978, 1980, and 1988 respectively. After having served as a R&D Staff of the Production Engineering Department, Nippon Denso Company, Ltd., a Research Associate at Tokyo Institute of Technology and an Associate Professor at Nagoya University, he joined Tohoku University as Professor in 1995 and served as Distinguished Professor from 2018 to March 2021. He is now serving as the Director of the Center for Transformative AI and Robotics, Specially Appointed Professor of Graduate School of Engineering, Tohoku University. From June 2021, he is also serving as Chair Professor in the Department of Electrical and Electronic Engineering, the University of Hong Kong.

He received Medal of Honor, Medal with Purple Ribbon, from the Government of Japan in 2018 - a national honor in recognition of his prominent contributions to academic and industrial advancements. He also received IEEE RAS George Saridis Leadership Award in Robotics and Automation in 2021 for his exceptional vision of innovative research and outstanding leadership in the robotics and automation community through technical activity management. He is an IEEE Fellow, JSME Fellow, SICE Fellow, RSJ Fellow, JSAE Fellow and a member of the Engineering Academy of Japan. He was the President of the IEEE Robotics and Automation Society, from 2010 to 2011, the IEEE Division X Director, from 2015 to 2016 and the IEEE Vice President for Technical Activities for 2020.

Deep learning models have been proven vulnerable to adversarial attacks. Making small-magnitude subtle changes to image and text data can result in significant testing errors on well-trained CNNs and DNNs. Whilst deep learning has become one of the most popular AI techniques being equipped into industry and business sectors, the safety and security concerns behind the vulnerability of deep learning models need to be studied and addressed. This talk will introduce several approaches that formulate the attacks and the defence from a game-theoretic perspective. The solution to the game-theoretic model is a Nash equilibrium with a pair of strategies representing the optimal states for the attackers and the defenders. This talk will also include key results of adversarial attacks and defence strategies on both image data and natural language text data.

Wei Liu is currently an Associate Professor of Machine Learning and the Director of Future Intelligence Lab at the School of Computer Science, Faculty of Engineering and Information Technology, the University of Technology Sydney (UTS). He also held honorary positions in CSIRO's Data61, the University of Melbourne, and SUSTech in China. He received his PhD degree in machine learning research from the University of Sydney (USyd) in 2011. Before joining UTS, he was a Machine Learning Researcher and Industry Project Manager at National ICT Australia (NICTA), and before that he was a Research Fellow at the University of Melbourne. His major research focuses are adversarial machine learning, game theory multimodal machine learning, tensor factorization, graph mining, causal inference, and anomaly detection. He had hold program chair and other chair positions in international conferences. He has published more than 90 papers in A*/A ranking conferences and journals, including AAAI, IJCAI, ICCV, KDD, ICDM, etc. One of his first-authored papers received the Most Influential Paper Award in PAKDD 2021. In addition, he has received three Best Paper Awards. He was nominated for the NSW Premier's Prizes for NSW Early Career Researcher award in 2017. Besides, he has received more than $2 million government competitive and industry research funding in the past six years.