Postdoctoral Research Scholar, Stanford University
Towards Trustworthy Autonomy: Exploring and Exploiting Structure in Human-Robot Systems
While one can easily imagine autonomous vehicles being released in the near future, it can be assumed that this transition will not be instantaneous. This suggests two key points: (1) levels of autonomy will be introduced incrementally (e.g. active safety systems as currently released), and (2) autonomous vehicles will have to be capable of driving in a mixed environment, with both humans and autonomous vehicles on the road. For such intelligent systems to safely operate, human behaviors must be reliably modeled in an accurate and precise manner that is easily integrated into control frameworks.
My work focuses on exploring and uncovering structure in complex human-robot systems to create more intelligent, interactive autonomy. This research agenda combines ideas from robotics, artificial intelligence, and control. By developing nonparametric approximations of human behaviors with minimal, realistic assumptions, we show how we can improve perception, planning, and control in for semi- and fully autonomous vehicles in multi-agent settings. These approaches are validated in high-fidelity simulation using tools from adaptive stress testing and through rigorous experiments in immersive human-in-the-loop testbeds and in fully outfitted autonomous test vehicles.
Katie is currently a Postdoctoral Research Scholar in the Stanford Intelligent Systems Laboratory working with Professor Mykel Kochenderfer. Prior to that, she received a B.S.E. with honors from Arizona State University in 2012 and a M.S. from UC Berkeley in 2015. In May of 2017, she earned her PhD in Electrical Engineering and Computer Sciences from the University of California, Berkeley, advised by Professor Ruzena Bajcsy. Her thesis was entitled “Tools for Trustworthy Autonomy: Robust Prediction, Intuitive Control, and Optimized Interaction,” which presented novel contributions to field of autonomy, merging ideas robotics, transportation, and control to address problems associated with human-in-the-loop. Her work considers the integration of autonomy into human dominated fields, in terms of safe interaction, with a strong emphasis on novel modeling methods, experimental design, robust learning, and control frameworks. She is particularly interested in finding safety guarantees for learning and adaptive systems and in non-parametric approximations of human behaviors that account for the combinatorial aspects of decision making and control. Currently, Katie is leading an industry sponsored research project to develop autonomous systems that can effectively perceive and interact with the environment and adapts to new, rare encounters.
Beyond research, Katie strives to impact the community beyond just technical contributions. She has been very active in mentoring and outreach throughout her academic career, particularly focusing on encouraging young women to pursue STEM careers. Within her graduate student community at Berkeley, Katie was co-president of both the Electrical Engineering Graduate Student Association and the Women in Computer Science and Electrical Engineering organization. In these positions, she sought to improve graduate student life, specifically targeting imposter syndrome and diversity issues. More recently, she has been engaging with the IEEE Intelligent Transportation Systems Society to organize regular Women in Intelligent Transportation Systems (WITS) events to foster a diverse and welcoming community within her field. She received the Demetri Angelakos Memorial Achievement Award for her contributions to the community.