CS Colloquium: Aleksandra Faust - Towards Autonomy at Scale via Generalist Autonomous Agents

Speaker: Aleksandra Faust, --

Talk Title: Towards Autonomy at Scale via Generalist Autonomous Agents

Abstract: Autonomous agent in the real world is any system, from autonomous cars, service robots to digital assistants, that works to perform meaningful tasks. All these applications share common challenges.The agents need to generalize well, learn new tasks and adapt to new and changing environments. They need to be competent and safe. And it should be tractable to train. This talk will discuss scaling up autonomous agents in the real world, along generalization, quality, and cost dimensions, anchored on robotics, autonomous driving, and digital assistant applications. Traditionally, we have been focusing mostly on the edges of this constraint triangle, and in the first part of the talk we will explore examples of efficient generalization, efficient high quality agents, and high quality generalists through the lens of social navigation, locomotion, and web agents. However, recent advances in machine learning enable us to start seeing progress in the tracktable high performing general agents. The second part of the talk will discuss our current and emerging work on the generalist agents, including some the roles of model capacity and data quality. We will conclude with a look into what the challenges and opportunities that future might hold, specifically how the progress in Large Language and Generative models might benefit autonomy at scale, and how the field might be changing

Biography: Aleksandra Faust is a Senior Staff Research Scientist and Autonomous Agents research lead in Google Deepmind. Her research is centered around safe and scalable autonomous systems for social good, including reinforcement learning, planning, and control for robotics, autonomous driving, and digital assistants. Previously, Aleksandra founded and led Task and Motion Planning research in Robotics at Google, machine learning for self driving car planning and controls in Waymo, and was a senior researcher in Sandia National Laboratories. She earned a Ph.D. in Computer Science at the University of New Mexico with distinction, and a Masters in Computer Science from the University of Illinois at Urbana Champaign. Aleksandra won the IEEE RAS Early Career Award for Industry, the Tom L. Popejoy Award for the best doctoral dissertation at the University of New Mexico in the period between 2011 and 2014, and was named Distinguished Alumna by the University of New Mexico School of Engineering. Her work has been featured in the New York Times, PC Magazine, ZdNet, and was awarded Best Paper in Service Robotics at ICRA 2018, Best Paper in Reinforcement Learning for Real Life at ICML 2019, Best Paper of IEEE Computer Architecture Letters in 2020, and IEEE Micro Top Picks 2023 Honorable Mention

Host: Stefanos Nikolaidis

Location: Hughes Aircraft Electrical Engineering Center (EEB) - 248

Audiences: Everyone Is Invited

Contact: Melissa Ochoa