I am a Ph.D. Candidate in Electrical Engineering and Computer Science at University of California, Merced. I am working on algorithms for human-robot interaction.
It is important that we enable the powerful AI systems to imbibe human values if we want to avoid any unintended consequence of the AI revolution. The AI systems should be aligned with human values: they should be designed to act as their operators want. However, it is not easy to specify human values in a way that AI can make use of. I am working on enabling AI systems to learn human values autonomously, with a focus on the situations where humans cannot directly demonstrate their preferences in actions or words. Such situations arise in several applications, for instance, my early research shows that human driving demonstrations do not reflect how they would prefer their autonomous car to drive. Demonstrations can also be unreliable when people do not fully understand the working of a robot or an AI system, e.g. orchestrating the full degrees of freedom in a robot arm. In my doctoral research, I am designing novel interactions that will allow AI systems or robots to seek alternative forms of guidance from humans and developing active learning algorithms that will enable them to learn human values faster.
Prior to this, I was a graduate student at Carnegie Mellon University, where I applied machine learning to problems in smart environments, e.g. human mobility prediction for human-robot rendezvous and occupancy estimation. I also have an M.S. from UC Berkeley, during which I developed sensor fusion methods for intelligent lighting systems.
Algorithmic Human Robot Interaction, Applied Machine Learning, Human Machine Interaction, Cloud Robotics, Crowd sourcing, Autonomous Vehicles