Control Principles for Robot Learning

Explore the cutting-edge intersection of robotics and learning systems in this seminar from the Georgia Tech Research IRIM Series. Delve into the complexities of motion synthesis for efficient and flexible learning in continuous online deployment of robotic systems. Examine how learning goals create unique control-oriented challenges, including selecting appropriate measures as objectives, real-time control synthesis, imposing physics-oriented constraints on learning, and producing analyses that guarantee performance and safety with limited knowledge. Gain insights from Professor Todd D. Murphey of Northwestern University as he discusses various learning tasks encountered by robots, measures for information content of observations, and algorithms for generating action plans. Discover examples from both biology and robotics throughout the presentation, and contemplate future challenges in the field of robot learning and control principles.