Professor Martial Hebert
Director of The Robotics Institute at Carnegie Mellon University
The talk will cover all the building blocks of Robotics systems: Acting (physical design of robots and their control), Seeing (Computer vision, and more generally machine perception), Learning (Machine Learning for Robotics), Deciding (making autonomous decision such as planning for self-driving cars), Interacting with people, and integration (building safe and trustable integrated systems for real world applications). "Acting" includes research in dexterity, e.g., human-like robotic hands, agility, i.e., novel mechanical designs, and modularity, i.e., new technology for dynamic configuration of robots. "Seeing" includes progress in understanding the environment of a robot from images, videos, and point clouds using machine learning;
It will also cover recent advances in modeling people's behavior, something that is critical for human-robot interaction, as well as new 3D reconstruction and sensing techniques. Current directions for applying Machine Learning techniques to physical systems, such as learning from interacting with the environment, and research in autonomy for ground and aerial robots will be explored.