Dr. Lydia Tapia

Learning solutions are critical to solve hard problems in robotics. Contrary to traditional solutions, learning offers the promise of new solutions that generalize to previously unseen scenarios, enhanced robustness to noise, and/or performance that was previously unachievable.  However, despite these advancements, learning is often not directly integrated into robust solutions for robotics, often kept as a separate contained solution. 

In this talk, we will discuss the difficulties and successes in applying learning directly into full solutions for autonomous robots.   The solutions cover a wide variety of scenarios from coordinating multi-agents, directly using sensor information for decision making, and making long-range choices.  Across all these scenarios, the challenges are fundamental.  How is safety guaranteed or considered?  Do solutions generalize? Is the solution interpretable?

For this audience, I will also highlight our recent study on the involvement of women in robotics.  As WIE Chair for the Robotics and Automation Society (RAS), I was involved in an eighteen-year retrospective across all IEEE RAS supported conferences.  We saw that while women are historically underrepresented, recent efforts to diversify robotics conference leadership and invited speakers have made substantial gains.