The National Science Foundation (NSF) Division of Information and Intelligent Systems awarded a $420,530 Project Grant to the Regents of the University of California, doing business as the University of California, Berkeley, under the NSF's Computer and Information Science and Engineering (CFDA 47.070) program. The grant supports a 3-year collaborative research project aimed at "Aligning Robot Representations with Humans". The key objectives are to: 1) develop new types of human feedback and algorithms to help robots learn the correct representation of objectives and tasks that are aligned with human preferences, 2) explore methods for extracting the full task representation from novel human input, 3) enable robots to appropriately apply the learned representation in new contexts, and 4) extend the techniques to policy learning. The research seeks to address challenges in ensuring robots exhibit behaviors and objectives that are robust and well-aligned with human values.
Generated 5/14/24, 9:05 AM