Keynote: Structured Optimization for Dexterous Robots
By Aida Rahmattalabi
Russ Tedrake from MIT gave a keynote talk on “Structured Optimization for Dexterous Robots.” As the talk started, the crowd gradually gathered, and shortly after, the room was packed with enthusiastic listeners. Russ Tedrake introduced the shortcomings in the present robotics research and how he proposed to address them by “embracing learning and optimization.” In particular, he highlighted the importance of robustness and he gave numerous examples of robotic systems that still fail to act reliably in real-world settings. To overcome these challenges, he is employing tools from optimization to model the dynamics of the system as well as the uncertainties (e.g., environmental disturbances) in order to design reliable control for the robotic systems. He shared fascinating examples of robots succeeding at challenging tasks, such as walking on rough terrains. Thanks to his research efforts, we are now able to see robotic arms capable of sorting out the dishes and placing them in the right tray in a washing machine when no one is home!