ASSET Seminar: Polyglot Robots: Versatile Goal-Based Task Specification for Robot Learning (Dinesh Jayaraman, University of Pennsylvania)

PRESENTATION ABSTRACT 

An important goal of the field sensorimotor robot learning is to do away with cumbersome expertise-intensive task specification, so that general-purpose robots of the future might learn large numbers of new skills. In this talk, I will discuss our recent work on algorithms that exploit goals as a versatile and accessible task specification interface. Goals might be specified through images, language, or physical objects, and may either be provided by a layperson or even discovered autonomously by a robot exploring its environment. I will show how unsupervised learning from large human action datasets can train goal-conditioned value functions for robots, how learned verification behaviors can in turn help to evaluate and acquire new skills, and how careful model-based reasoning can help a robot discover interesting goal-based tasks in an environment with no supervision.

BIO

Dinesh Jayaraman is an assistant professor at the University of Pennsylvania’s CIS department. His research group focuses on developing machine learning algorithms that attend to task-relevant components in each stage of the embodied perception-action loop: observation, representation, and action. Dinesh’s research has received an NSF CAREER award ’23, Best Paper Award at CORL ’22, AAAI New Faculty Highlights award ’21, an Amazon Research Award ’21, a Best Paper Runner-Up Award at ICRA ’18, a Best Application Paper Award at ACCV ’16, and been featured on the cover page of Science Robotics and in several press outlets.