Full-day Workshop

Confirmed Speakers (in alphabetical order):

Dana Nau

Department of Computer Science at The University of Maryland

Machine-Learning Challenges in Planning for Hierarchically Organized Systems


To plan and carry out the activities of a hierarchically organized autonomous system, generally the planning and acting functions need to reflect the system's hierarchical organization. One approach for this is refinement planning and acting, in which tasks at each level of the hierarchy are refined into collections of tasks at the next level down, with the bottom levels consisting of "primitive" tasks (e.g., motion planning) to be sent to the system's execution platform(s). The speaker will give an overview of refinement planning and some of the challenges that it poses for machine learning.


Dana Nau is an AI researcher whose research includes both automated planning and game theory. Some of his best-known work includes the discovery of game trees that are "pathological" in the sense that deeper lookahead produces worse decisions; the strategic planning algorithm used to win the 1997 world championship of computer bridge; the SHOP and SHOP2 planning algorithms; two graduate-level textbooks on automated planning and acting; and evolutionary game-theoretic studies of the evolution of human behavioral norms. Dr. Nau has more than 300 refereed publications. He is an AAAI Fellow and an ACM Fellow.

David Hsu

Department of Computer Science at National University of Singapore

Robot Planning and Learning under Uncertainty: Data, Models, and Actions


Planning and learning are two primary means towards intelligent robots. Planning provides a powerful way to predict the consequences of immediate actions far into the future, but it requires accurate world models, which are often difficult to acquire in practice. Alternatively, direct policy learning circumvents the need for models and learns a mapping from robot perceptual inputs to actions directly. However, without models, it is much more difficult to generalize and adapt learned policies to new contexts. In this talk, I will present our recent work on robust robot decision-making under uncertainty through planning, through learning, most importantly by integrating planning and learning. I will give several examples, including autonomous vehicle navigation among many pedestrians and human-robot interaction.


David Hsu is a professor of computer science at the National University of Singapore (NUS) and a member of NUS Graduate School for Integrative Sciences & Engineering. He is an IEEE Fellow. His research spans robotics and AI. In recent years, he has been working on robot planning and learning under uncertainty for human-centered robots. He has chaired or co-chaired several major international robotics conferences, including WAFR 2004 and 2010, RSS 2015, and ICRA 2016.

Oliver Brock

Department of Computer Engineering and Microelectronic at Technische Universität Berlin

The Right Tool for the Job: About Swiss Army Knives, Hammers, Motion Planning, Control, and Machine Learning


Why is it so important to use the right tool for the job? Because it makes everything much easier! But why does the right tool make the job easier? The right tool reflects a deep understanding of a good solution to the problem. It makes that good solution easily actionable (think can opener, bike pump, car jack, …). This implies that the right tool cannot be good for all problems. A universal tool cannot be a good tool. There are, of course, situations when a Swiss Army knife (aka machine learning?) is super-handy---because you are happy to have a tool that works at all and you don’t need the best tool). [Sorry for the long tool analogy.] The same holds true motion planning and control. There are many different subproblems, each potentially deserving its own best tool. For some problems we already have very good tools, for some we are using the wrong tools, and for yet others we don’t know yet. In this talk, I will attempt to provide an overview of different problems, their inherent problem structure, and the most natural solution tool. While doing that I will talk about trajectory optimization, combining planning and control, planning motion in contact, and combining motion and task planning.


Oliver Brock is the Alexander-von-Humboldt Professor of Robotics in the School of Electrical Engineering and Computer Science at the Technische Universität Berlin in Germany. He received his Diploma in Computer Science in 1993 from the Technische Universität Berlin and his Master's and Ph.D. in Computer Science from Stanford University in 1994 and 2000, respectively. He also held post-doctoral positions at Rice University and Stanford University. Starting in 2002, he was an Assistant Professor and Associate Professor in the Department of Computer Science at the University of Massachusetts Amherst, before to moving back to the Technische Universität Berlin in 2009. The research of Brock's lab, the Robotics and Biology Laboratory, focuses on mobile manipulation, interactive perception, grasping, manipulation, soft material robotics, interactive machine learning, deep learning, motion generation, and the application of algorithms and concepts from robotics to computational problems in structural molecular biology. He is the president of the Robotics: Science and Systems foundation.